If you are being abused, making a safety plan now may help you when you have to act quickly in the future. The following ideas are ways that other women have planned for their safety. Some of these ideas may work for you. You may come up with additional ideas for yourself. You know your own situation better than anyone else, so plan what will work best for you.

• Hide money or put it somewhere safe so you can leave quickly.
• Make copies of birth certificates, immunization records, Social Security numbers, and other important documents to keep in safe locations away from home such as at work, the homes of trusted family members or friends, or hidden in convenient locations.
• Hide a spare car key or bus or subway pass that you can grab quickly.
• Keep a list of hotline numbers, or memorize the 1-800-799-SAFE National Domestic Violence Hotline number.
• Develop a code with friends, family, or neighbors to let them know when you need help in an emergency. If you have children, teach them a signal (like a code word) that means they should call the police or go for help. You may want to have a special code for neighbors (like putting on a particular light or opening a certain window) that means you want them to call the police.
• Plan your exit. Know which doors, windows, stairwells, elevators, or fire escapes you can use if you have to leave quickly. Practice using them so that they feel familiar to you.
• Know how to reach the police and your local women’s shelter.
• Every day, think about where you can go immediately if you have to leave. Is a neighbor home today? A relative? A friend?
• Remove weapons from your home if you can.
• Something to think about: When you cannot get away and your partner becomes violent, which room is the safest for you to get to? Is there a room that has a phone and a lock on the door? Can you stay out of rooms with easy weapons, such as the kitchen?
• Try not to leave without your children. But if you have to leave your children with the abuser, call the police immediately after you escape.

Adapted from Chang JC. Domestic violence. In: Bieber EJ, et al, editors. Clinical Gynecology. Philadelphia, PA; Elsevier, Inc; 2006:79–89.

This information is provided by your physician and the Cleveland Clinic Journal of Medicine. It is not designed to replace a physician’s medical assessment and judgment.

This page may be reproduced noncommercially to share with patients. Any other reproduction is subject to Cleveland Clinic Journal of Medicine approval. Bulk color reprints available by calling 216-444-2661.

For patient information on hundreds of health topics, see the Center for Consumer Health Information web site, www.clevelandclinic.org/health

If you are being abused, making a safety plan now may help you when you have to act quickly in the future. The following ideas are ways that other women have planned for their safety. Some of these ideas may work for you. You may come up with additional ideas for yourself. You know your own situation better than anyone else, so plan what will work best for you.

• Hide money or put it somewhere safe so you can leave quickly.
• Make copies of birth certificates, immunization records, Social Security numbers, and other important documents to keep in safe locations away from home such as at work, the homes of trusted family members or friends, or hidden in convenient locations.
• Hide a spare car key or bus or subway pass that you can grab quickly.
• Keep a list of hotline numbers, or memorize the 1-800-799-SAFE National Domestic Violence Hotline number.
• Develop a code with friends, family, or neighbors to let them know when you need help in an emergency. If you have children, teach them a signal (like a code word) that means they should call the police or go for help. You may want to have a special code for neighbors (like putting on a particular light or opening a certain window) that means you want them to call the police.
• Plan your exit. Know which doors, windows, stairwells, elevators, or fire escapes you can use if you have to leave quickly. Practice using them so that they feel familiar to you.
• Know how to reach the police and your local women’s shelter.
• Every day, think about where you can go immediately if you have to leave. Is a neighbor home today? A relative? A friend?
• Remove weapons from your home if you can.
• Something to think about: When you cannot get away and your partner becomes violent, which room is the safest for you to get to? Is there a room that has a phone and a lock on the door? Can you stay out of rooms with easy weapons, such as the kitchen?
• Try not to leave without your children. But if you have to leave your children with the abuser, call the police immediately after you escape.

Adapted from Chang JC. Domestic violence. In: Bieber EJ, et al, editors. Clinical Gynecology. Philadelphia, PA; Elsevier, Inc; 2006:79–89.

This information is provided by your physician and the Cleveland Clinic Journal of Medicine. It is not designed to replace a physician’s medical assessment and judgment.

This page may be reproduced noncommercially to share with patients. Any other reproduction is subject to Cleveland Clinic Journal of Medicine approval. Bulk color reprints available by calling 216-444-2661.

For patient information on hundreds of health topics, see the Center for Consumer Health Information web site, www.clevelandclinic.org/health

If you are being abused, making a safety plan now may help you when you have to act quickly in the future. The following ideas are ways that other women have planned for their safety. Some of these ideas may work for you. You may come up with additional ideas for yourself. You know your own situation better than anyone else, so plan what will work best for you.

• Hide money or put it somewhere safe so you can leave quickly.
• Make copies of birth certificates, immunization records, Social Security numbers, and other important documents to keep in safe locations away from home such as at work, the homes of trusted family members or friends, or hidden in convenient locations.
• Hide a spare car key or bus or subway pass that you can grab quickly.
• Keep a list of hotline numbers, or memorize the 1-800-799-SAFE National Domestic Violence Hotline number.
• Develop a code with friends, family, or neighbors to let them know when you need help in an emergency. If you have children, teach them a signal (like a code word) that means they should call the police or go for help. You may want to have a special code for neighbors (like putting on a particular light or opening a certain window) that means you want them to call the police.
• Plan your exit. Know which doors, windows, stairwells, elevators, or fire escapes you can use if you have to leave quickly. Practice using them so that they feel familiar to you.
• Know how to reach the police and your local women’s shelter.
• Every day, think about where you can go immediately if you have to leave. Is a neighbor home today? A relative? A friend?
• Remove weapons from your home if you can.
• Something to think about: When you cannot get away and your partner becomes violent, which room is the safest for you to get to? Is there a room that has a phone and a lock on the door? Can you stay out of rooms with easy weapons, such as the kitchen?
• Try not to leave without your children. But if you have to leave your children with the abuser, call the police immediately after you escape.

Adapted from Chang JC. Domestic violence. In: Bieber EJ, et al, editors. Clinical Gynecology. Philadelphia, PA; Elsevier, Inc; 2006:79–89.

This information is provided by your physician and the Cleveland Clinic Journal of Medicine. It is not designed to replace a physician’s medical assessment and judgment.

This page may be reproduced noncommercially to share with patients. Any other reproduction is subject to Cleveland Clinic Journal of Medicine approval. Bulk color reprints available by calling 216-444-2661.

For patient information on hundreds of health topics, see the Center for Consumer Health Information web site, www.clevelandclinic.org/health

Frail older adults deserve guidelines that take frailty into account while assessing the potential benefit and risks of treatment.

Specifically, our group—the Dalhousie Academic Detailing Service (ADS) and the Palliative and Therapeutic Harmonization (PATH) program—recommends that physicians strive to achieve more liberal treatment targets for elderly frail patients who have high blood pressure,¹ as evidence does not support an aggressive approach in the frail elderly and the potential exists for harm.

This article reviews the evidence and reasoning that were used to develop and promote a guideline for drug treatment of hypertension in frail older adults. Our recommendations differ from other guidelines in that they focus as much on stopping or decreasing therapy as on starting or increasing it.

FRAILTY INCREASES THE RISK OF ADVERSE EFFECTS

The word frail, applied to older adults, describes those who have complex medical illnesses severe enough to compromise their ability to live independently.² Many have multiple coexisting medical problems for which they take numerous drugs, in addition to dementia, impaired mobility, compromised functional ability, or a history of falling.

Frailty denotes vulnerability; it increases the risk of adverse effects from medical and surgical procedures,³ complicates drug therapy,⁴ prolongs hospital length of stay,⁵ leads to functional and cognitive decline,⁶ increases the risk of institutionalization,⁷ and reduces life expectancy⁸—all of which affect the benefit and harm of medical treatments.

Guidelines for treating hypertension^9–11 now acknowledge that little evidence exists to support starting treatment for systolic blood pressure between 140 and 160 mm Hg or aiming for a target of less than 140 mm Hg for “very old” adults, commonly defined as over the age of 80. New guidelines loosen the treatment targets for the very old, but they do not specify targets for the frail and do not describe how to recognize or measure frailty.

RECOGNIZING AND MEASURING FRAILTY

A number of tools are available to recognize and measure frailty.¹²

The Fried frailty assessment¹³ has five items:

• Unintentional weight loss
• Self-reported exhaustion
• Weakness in grip
• Slow walking speed
• Low physical activity and energy expenditure.

People are deemed frail if they have three or more of these five. However, experts disagree about whether this system is too sensitive¹⁴ or not sensitive enough.^15,16

The FRAIL questionnaire¹⁷ also has five items:

• Fatigue
• Resistance (inability to climb stairs)
• Ambulation (inability to walk 1 city block)
• Illness (more than 5 major illnesses)
• Weight loss.

People are deemed frail if they have at least three of these five items, and “prefrail” if they have two.

These and other tools are limited by being dichotomous: they classify people as being either frail or not frail^18–20 but do not define the spectrum of frailty.

Other frailty assessments such as the Frailty Index²¹ identify frailty based on the number of accumulated health deficits but take a long time to complete, making them difficult to use in busy clinical settings.^22–24

The Clinical Frailty Scale⁷ is a validated scale that categorizes frailty based on physical and functional indicators of health, such as cognition, function, and mobility, with scores that range from 1 (very fit) to 9 (terminally ill).^7,12

The Frailty Assessment for Care-planning Tool (FACT) uses scaling compatible with the Clinical Frailty Scale but has been developed for use as a practical and interpretable frailty screening tool for nonexperts (Table 1). The FACT assesses cognition, mobility, function, and the social situation, using a combination of caregiver report and objective measures. To assess cognition, a health care professional uses items from the Mini-Cog²⁵ (ie, the ability to draw an analog clock face and then recall three unrelated items following the clock-drawing test) and the memory axis of the Brief Cognitive Rating Scale²⁶ (ie, the ability to recall current events, the current US president, and the names of children or spouse). Mobility, function, and social circumstance scores are assigned according to the caregiver report of the patient’s baseline status.

The FACT can be completed in busy clinical settings. Once a caregiver is identified, it takes about 5 minutes to complete.

Our guideline^27–31 is intended for those with a score of 7 or more on the Clinical Frailty Scale or FACT,^7,12 a score we chose because it describes people who are severely frail with shortened life expectancy.⁸ At this level, people need help with all instrumental activities of daily living (eg, handling finances, medication management, household chores, and shopping) as well as with basic activities of daily living such as bathing or dressing.

REVIEWING THE LIMITED EVIDENCE

We found no studies that addressed the risks and benefits of treating hypertension in frail older adults; therefore, we concentrated on studies that enrolled individuals who were chronologically old but not frail. We reviewed prominent guidelines,^9–11,32,33 the evidence base for these guidelines,^34–44 and Cochrane reviews.^45,46 A detailed description of the evidence used to build our recommendation can be found online.³¹

When we deliberated on treatment targets, we reviewed evidence from two types of randomized controlled trials⁴⁷:

Drug treatment trials randomize patients to different treatments, such as placebo versus a drug or one drug compared with another drug. Patients in different treatment groups may achieve different blood pressures and clinical outcomes, and this information is then used to define optimal targets. However, it may be difficult to determine if the benefit came from lowering blood pressure or from some other effect of the drug, which can be independent of blood pressure lowering.

Treat-to-target trials randomize patients to different blood pressure goals, but the groups are treated with the same or similar drugs. Therefore, any identified benefit can be attributed to the differences in blood pressure rather than the medications used. Compared with a drug treatment trial, this type of trial provides stronger evidence about optimal targets.

We also considered the characteristics of frailty, the dilemma of polypharmacy, and the relevance of the available scientific evidence to those who are frail.

Drug treatment trials

A Cochrane review⁴⁵ of 15 studies with approximately 24,000 elderly participants found that treating hypertension decreased the rates of cardiovascular morbidity and mortality as well as fatal and nonfatal stroke in the “elderly” (defined as age ≥ 60) and “very elderly” (age ≥ 80). However, in the very elderly, all-cause mortality rates were not statistically significantly different with treatment compared with placebo. The mean duration of treatment was 4.5 years in the elderly and 2.2 years in the very elderly (Table 2). Of importance, all the trials enrolled only those individuals whose systolic blood pressure was at least 160 mm Hg at baseline.

None of the studies were treat-to-target trials—patients were assigned either active medication or placebo. Thus, these trials provide evidence of benefit for treating hypertension in the elderly and very elderly but do not identify the optimal target. All of the drug treatment trials showed benefit, but none achieved a systolic pressure lower than 140 mm Hg with active treatment (Table 3). Therefore, these studies do not support a systolic target of less than 140 mm Hg in the elderly.

Treat-to-target trials: JATOS and VALISH

The Japanese Trial to Assess Optimal Systolic Blood Pressure in Elderly Hypertensive Patients (JATOS)⁴² and the Valsartan in Elderly Isolated Systolic Hypertension (VALISH) study⁴³ each enrolled more than 3,000 people age 65 or older (mean age approximately 75). Patients were randomized to either a strict systolic target of less than 140 mm Hg or a higher (more permissive) target of 140 to 160 mm Hg in JATOS and 140 to 149 mm Hg in VALISH.

In both trials, the group with strict targets achieved a systolic pressure of approximately 136 mm Hg, while the group with higher blood pressure targets achieved a systolic pressure of 146 mm Hg in JATOS and 142 mm Hg in VALISH. Despite these differences, there was no statistically significant difference in the primary outcome.

Thus, treat-to-target studies also fail to support a systolic target of less than 140 mm Hg in the elderly, although it is important to recognize the limitations of the studies. Approximately 15% of the participants had cardiovascular disease, so the applicability of the findings to patients with target-organ damage is uncertain. In addition, there were fewer efficacy outcome events than expected, which suggests that the studies were underpowered.

When to start drug treatment

In each of the drug treatment and treat-to-target trials, the inclusion criterion for study entry was a systolic blood pressure above 160 mm Hg, with a mean blood pressure at entry into the drug treatment trials of 182/95 mm Hg.⁴⁶ Thus, data support starting treatment if the systolic blood pressure is above 160 mm Hg, but not lower.

Notably, in all but one study,⁴⁶ at least two-thirds of the participants took no more than two antihypertensive medications. Since adverse events become more common as the number of medications increases, the benefit of adding a third drug to lower blood pressure is uncertain.

Evidence in the ‘very elderly’: HYVET

With the exception of the Hypertension in the Very Elderly Trial (HYVET),⁴⁴ the mean age of elderly patients in the reported studies was between 67 and 76.

HYVET patients were age 80 and older (mean age 84) and were randomized to receive either indapamide (with or without perindopril) or placebo. The trial was stopped early at 2 years because the mortality rate was lower in the treatment group (10.1%) than in the placebo group (12.3%) (number needed to treat 46, 95% confidence interval 24–637, P = .02). There was no significant difference in the primary outcome of fatal and nonfatal stroke.

Notably, trials that are stopped early may overestimate treatment benefit.⁴⁸

Evidence in frail older adults

While the above studies provide some information about managing hypertension in the elderly, the participants were generally healthy. HYVET⁴⁴ specifically excluded those with a standing systolic blood pressure of less than 140 mm Hg and enrolled few patients with orthostasis (7.9% in the placebo group and 8.8% in the treatment group), a condition commonly associated with frailty. As such, these studies may be less relevant to the frail elderly, who are at higher risk of adverse drug events and have competing risks for morbidity and mortality.

Observational studies, in fact, raise questions about whether tight blood pressure control improves clinical outcomes for the very elderly. In the Leiden 85-plus study, lower systolic blood pressure was associated with lower cognitive scores, worse functional ability,^49,50 and a higher mortality rate⁵¹ compared with higher systolic pressure, although it is uncertain whether these outcomes were indicative of underlying disease that could result in lower blood pressure or an effect of blood pressure-lowering.

The National Health and Nutrition Examination Survey⁵² found an association between blood pressure and mortality rate that varied by walking speed. For slower walkers (based on the 6-minute walk test), higher systolic pressures were not associated with a higher risk of death, suggesting that when older adults are frail (as indicated by their slow walking speed) they are less likely to benefit from aggressive treatment of hypertension.

People at high risk because of stroke

Because the evidence is limited, it is even more difficult to judge whether lowering blood pressure below 140 mm Hg is beneficial for frail patients who have a history of stroke, compared with the possibility that medications will cause adverse effects such as weakness, orthostasis, and falls. When reviewing the evidence to answer this question, we especially looked at outcomes that affect quality of life, such as nonfatal stroke leading to disability. In contrast, because the frail elderly have competing causes of mortality, we could not assume that a mortality benefit shown in nonfrail populations could be applied to frail populations.

The PROGRESS trial (Perindopril Protection Against Recurrent Stroke Study)⁵³ was in patients with a history of stroke or transient ischemic attack and a mean age of 64, who were treated with either perindopril (with or without indapamide) or placebo.

At almost 4 years, the rate of disabling stroke was 2.7% in the treatment group and 4.3% in the placebo group, a relative risk reduction of 38% and an absolute risk reduction of 1.64% (number needed to treat 61, 95% confidence interval 39–139). The relative risk reduction for all strokes (fatal and nonfatal) was similar across a range of baseline systolic pressures, but the absolute risk reduction was greater in the prespecified subgroup that had hypertension at baseline (mean blood pressure 159/94 mm Hg) than in the normotensive subgroup (mean blood pressure 136/79 mm Hg), suggesting that treatment is most beneficial for those with higher systolic blood pressures. Also, the benefit was only demonstrated in the subgroup that received two antihypertensive medications; those who received perindopril alone showed no benefit.

This study involved relatively young patients in relatively good health except for their strokes. The extent to which the results can be extrapolated to older, frail adults is uncertain because of the time needed to achieve benefit and because of the added vulnerability of frailty, which could make treatment with two antihypertensive medications riskier.

PRoFESS (Prevention Regimen for Effectively Avoiding Second Strokes),⁵⁴ another study in patients with previous stroke (mean age 66) showed no benefit over 2.5 years in the primary outcome of stroke using telmesartan 80 mg daily compared with placebo. This result is concordant with that of PROGRESS,⁵³ in which patients who took only one medication did not show a significant decrease in the rate of stroke.

A possible reason for the lack of benefit from monotherapy was that the differences in blood pressure between the placebo group and the treatment group on monotherapy were small in both studies (3.8/2.0 mm Hg in PRoFESS, 5/3 mm Hg in PROGRESS). In contrast, patients on dual therapy in PROGRESS decreased their blood pressure by 12/5 mm Hg compared with placebo.

CURRENT HYPERTENSION GUIDELINES

Current guidelines make reference to the elderly, but we found none that made specific recommendations for the frail elderly.

JNC 8

In December 2013, members of the Eighth Joint National Committee on the Prevention, Detection, Evaluation, and Treatment of High Blood Pressure (JNC 8) released new recommendations.³² One significant revision was to support higher blood pressure targets for older adults (age 60 and older). Whereas JNC 7 stated that lowering blood pressure below 140/90 mm Hg reduced cardiovascular complications,³³ JNC 8 now acknowledges that there is no strong evidence to support blood pressure targets below 150/90 mm Hg for hypertensive persons without kidney disease or diabetes age 60 and older. Thus, in the general population age 60 and older, JNC 8 recommends starting antihypertensive treatment when blood pressure is 150/90 mm Hg or higher, and treating to a goal blood pressure of less than 150/90 mm Hg. JNC 8 makes no recommendation about how to adjust blood pressure targets for frailty or how to measure blood pressure.

American College of Cardiology and American Heart Association

In 2011, the American College of Cardiology and American Heart Association published a consensus document on the management of hypertension in the elderly.⁹

They acknowledged that the generally recommended blood pressure goal of lower than 140/90 mm Hg in uncomplicated elderly patients is based on expert opinion rather than on data from randomized controlled trials, but nevertheless recommended a target systolic pressure lower than 140 mm Hg for older adults, except for octogenarians.

For those over age 80, systolic levels of 140 to 145 mm Hg can be acceptable if tolerated and if the patient does not experience orthostasis when standing. Systolic pressure lower than 130 mm Hg and diastolic pressures lower than 65 mm Hg should be avoided in this age group.

The document acknowledges that systolic pressure may have to remain above 150 mm Hg if there is no response to four “well-selected drugs” or if there are unacceptable side effects. In these cases, the lowest “safely achieved” systolic blood pressure should be the goal.

Canadian Hypertension Education Program

The 2014 Canadian Hypertension Education Program (CHEP) report makes several recommendations for the “very elderly,” a group they define as over the age of 80. The CHEP website and resources include the following recommendations¹⁰:

• For the very elderly without diabetes or target-organ damage, drug therapy should be initiated when systolic blood pressure is higher than 160 mm Hg to reach a systolic blood pressure target lower than 150 mm Hg. This is a grade C level recommendation, indicating that it is based on low-quality trials, unvalidated surrogate outcomes, or results from nonrandomized observational studies.
• For the very elderly with macrovascular target-organ damage, antihypertensive therapy should be considered if systolic blood pressure readings average 140 mm Hg or higher (grade D for 140 to 160 mm Hg; grade A for higher than 160 mm Hg), although caution should be exercised in elderly patients who are frail. (Grade D recommendations are the weakest, as they are based on low-powered, imprecise studies or expert opinion, whereas grade A recommendations are based on the strongest evidence from high-quality randomized clinical trials.)
• Decisions regarding initiating and intensifying pharmacotherapy in the very elderly should be based on an individualized risk-benefit analysis.

The European Society of Hypertension and European Society of Cardiology

The 2013 guidelines from the European Society of Hypertension and the European Society of Cardiology¹¹ recommend that for elderly patients under age 80, antihypertensive treatment may be considered at systolic values higher than 140 mm Hg and aimed at values lower than 140 mm Hg if the patient is fit and treatment is well tolerated.

For those over age 80 with an initial systolic pressure of 160 mm Hg or higher, the guidelines recommend lowering systolic pressure to between 150 and 140 mm Hg, provided the patient is in good physical and mental condition. In frail elderly patients, they recommend leaving decisions on antihypertensive therapy to the treating physician, based on monitoring of the clinical effects of treatment.¹¹

The ADS/PATH guidelines

When finalizing our recommendations,¹ we considered the characteristics of frailty and the following key points from the evidence:

• Although evidence from drug treatment trials indicates that there is benefit in treating healthy older adults who have hypertension, the benefit of treating frail older adults is unknown.
• Major trials enrolled elderly patients only if they had systolic blood pressures of at least 160 mm Hg. Therefore, evidence supports initiating pharmacotherapy at a systolic pressure of 160 mm Hg or higher.
• No evidence from randomized controlled trials supports a systolic target lower than 140 mm Hg in the elderly, and there is some evidence that such a target does not benefit.
• The benefit of adding a third medication to lower blood pressure has not been studied.
• Frailty makes the potential benefits of strict blood pressure targets even less certain and increases the possibility of harm from adverse drug events.
• The only study of very old adults, HYVET,⁴⁴ enrolled relatively healthy older adults and few with orthostasis, while excluding those with a standing systolic blood pressure lower than 140 mm Hg.

OUR RECOMMENDATIONS

Based on the above, we advise against unnecessarily strict targets and recommend stopping antihypertensive medications that are used for the sole purpose of keeping the systolic blood pressure below 140 mm Hg. Our guidelines are unique in that they focus equally on when to stop and when to start medications. We concluded that without evidence of definitive benefit, “less is more” with frailty.⁵⁵ We believe that if physicians and health professionals understand the limitations of the evidence, they can be more confident in stopping medications that lower blood pressure to an unnecessarily low level.

We recommend the following (Table 4):

Before treating

• Carefully review the risks and the potential but unproven benefits of treatment.
• To avoid overtreatment, treatment decisions should be based on blood pressure measurements in the seated (not supine) position, while also considering the presence of orthostasis.
• To evaluate orthostasis, measure blood pressure in the supine position, then immediately on standing, and again after 2 minutes. Ask the patient if he or she feels light-headed or dizzy when standing.

Stop treatment

• If the seated systolic blood pressure is less than 140 mm Hg, medications can be tapered and discontinued to achieve the targets described below.
• Before discontinuation, consider whether the medications are treating additional conditions such as rate control for atrial fibrillation or symptomatic management of heart failure.
• It is uncertain whether to discontinue treatment when there is a history of stroke. Consider that treatment with two medications resulted in an absolute risk reduction for disabling stroke of 1.64% over approximately 4 years for adults with previous stroke and a mean age of 64,⁵⁷ an effect that may be more prominent at higher systolic pressures.

Start treatment

• Consider starting treatment when systolic pressure is 160 mm Hg or higher.
• Aim for a seated systolic pressure between 140 and 160 mm Hg if there are no adverse effects from treatment that affect quality of life.
• If there is symptomatic orthostasis or if standing systolic pressure is lower than 140 mm Hg, the target seated systolic pressure can be adjusted upwards.
• In the severely frail nearing the end of life, a target systolic pressure of 160 to 190 mm Hg is reasonable.
• The blood pressure target is the same in people with diabetes.
• In general, use no more than two medications.

Dissemination and implementation

The ADS/PATH guideline is intended for use by physicians and other health professionals (eg, pharmacists and nurses) who care for frail older adults or who work in long-term care facilities. Since creating our guideline, we have disseminated it to physicians, pharmacists, and other health professionals through academic detailing, large conferences, and interactive webinars.

While we do not have objective evidence of practice change, our evaluation data found that 34% of 403 family physicians who received academic detailing indicated that the guideline would change their practice, while 36% stated that the guideline confirmed their practice, an indication that family physicians are sensitive to the needs of the frail elderly.

Because health professionals may be wary of stopping medications and not meeting recommended targets, there may be barriers to adopting this guideline. However, our experience with the PATH program indicates that these barriers can be overcome using effective communication strategies between health professionals and consumers.

AN APPROACH APPROPRIATE TO FRAILTY

There is no direct evidence for systolic blood pressure targets in the frail elderly, so we applied evidence from the nonfrail elderly. Our recommendations differ somewhat from those of other groups, which recommend targets below 140 to 150 mm Hg for older adults, although some do advise caution in the elderly for whom a substantial fall in blood pressure might be poorly tolerated. Despite these messages, we believe that clearer guidance is needed to direct health practitioners toward models that acknowledge that frail patients are in a precarious balance of health and may be harmed by treatments that strive to lower blood pressure to unproven targets. For this reason, our guideline clearly indicates when to decrease or stop drug treatment.

After physicians and health professionals examine the evidence and more fully understand the benefits and harms of treating frail older adults, we are confident that they will be more comfortable stopping medications that lower blood pressure to an unnecessarily low level and instead use an approach that is more appropriate to frailty. We hope clinicians can use this guideline with the same enthusiasm applied to other guidelines, and we welcome discussion.

Acknowledgments: We would like to thank and acknowledge Tanya MacLeod and Kathryn Yuill for their review of and advice about the manuscript.

Frail older adults deserve guidelines that take frailty into account while assessing the potential benefit and risks of treatment.

Specifically, our group—the Dalhousie Academic Detailing Service (ADS) and the Palliative and Therapeutic Harmonization (PATH) program—recommends that physicians strive to achieve more liberal treatment targets for elderly frail patients who have high blood pressure,¹ as evidence does not support an aggressive approach in the frail elderly and the potential exists for harm.

This article reviews the evidence and reasoning that were used to develop and promote a guideline for drug treatment of hypertension in frail older adults. Our recommendations differ from other guidelines in that they focus as much on stopping or decreasing therapy as on starting or increasing it.

FRAILTY INCREASES THE RISK OF ADVERSE EFFECTS

The word frail, applied to older adults, describes those who have complex medical illnesses severe enough to compromise their ability to live independently.² Many have multiple coexisting medical problems for which they take numerous drugs, in addition to dementia, impaired mobility, compromised functional ability, or a history of falling.

Frailty denotes vulnerability; it increases the risk of adverse effects from medical and surgical procedures,³ complicates drug therapy,⁴ prolongs hospital length of stay,⁵ leads to functional and cognitive decline,⁶ increases the risk of institutionalization,⁷ and reduces life expectancy⁸—all of which affect the benefit and harm of medical treatments.

Guidelines for treating hypertension^9–11 now acknowledge that little evidence exists to support starting treatment for systolic blood pressure between 140 and 160 mm Hg or aiming for a target of less than 140 mm Hg for “very old” adults, commonly defined as over the age of 80. New guidelines loosen the treatment targets for the very old, but they do not specify targets for the frail and do not describe how to recognize or measure frailty.

RECOGNIZING AND MEASURING FRAILTY

A number of tools are available to recognize and measure frailty.¹²

The Fried frailty assessment¹³ has five items:

• Unintentional weight loss
• Self-reported exhaustion
• Weakness in grip
• Slow walking speed
• Low physical activity and energy expenditure.

People are deemed frail if they have three or more of these five. However, experts disagree about whether this system is too sensitive¹⁴ or not sensitive enough.^15,16

The FRAIL questionnaire¹⁷ also has five items:

• Fatigue
• Resistance (inability to climb stairs)
• Ambulation (inability to walk 1 city block)
• Illness (more than 5 major illnesses)
• Weight loss.

People are deemed frail if they have at least three of these five items, and “prefrail” if they have two.

These and other tools are limited by being dichotomous: they classify people as being either frail or not frail^18–20 but do not define the spectrum of frailty.

Other frailty assessments such as the Frailty Index²¹ identify frailty based on the number of accumulated health deficits but take a long time to complete, making them difficult to use in busy clinical settings.^22–24

The Clinical Frailty Scale⁷ is a validated scale that categorizes frailty based on physical and functional indicators of health, such as cognition, function, and mobility, with scores that range from 1 (very fit) to 9 (terminally ill).^7,12

The Frailty Assessment for Care-planning Tool (FACT) uses scaling compatible with the Clinical Frailty Scale but has been developed for use as a practical and interpretable frailty screening tool for nonexperts (Table 1). The FACT assesses cognition, mobility, function, and the social situation, using a combination of caregiver report and objective measures. To assess cognition, a health care professional uses items from the Mini-Cog²⁵ (ie, the ability to draw an analog clock face and then recall three unrelated items following the clock-drawing test) and the memory axis of the Brief Cognitive Rating Scale²⁶ (ie, the ability to recall current events, the current US president, and the names of children or spouse). Mobility, function, and social circumstance scores are assigned according to the caregiver report of the patient’s baseline status.

The FACT can be completed in busy clinical settings. Once a caregiver is identified, it takes about 5 minutes to complete.

Our guideline^27–31 is intended for those with a score of 7 or more on the Clinical Frailty Scale or FACT,^7,12 a score we chose because it describes people who are severely frail with shortened life expectancy.⁸ At this level, people need help with all instrumental activities of daily living (eg, handling finances, medication management, household chores, and shopping) as well as with basic activities of daily living such as bathing or dressing.

REVIEWING THE LIMITED EVIDENCE

We found no studies that addressed the risks and benefits of treating hypertension in frail older adults; therefore, we concentrated on studies that enrolled individuals who were chronologically old but not frail. We reviewed prominent guidelines,^9–11,32,33 the evidence base for these guidelines,^34–44 and Cochrane reviews.^45,46 A detailed description of the evidence used to build our recommendation can be found online.³¹

When we deliberated on treatment targets, we reviewed evidence from two types of randomized controlled trials⁴⁷:

Drug treatment trials randomize patients to different treatments, such as placebo versus a drug or one drug compared with another drug. Patients in different treatment groups may achieve different blood pressures and clinical outcomes, and this information is then used to define optimal targets. However, it may be difficult to determine if the benefit came from lowering blood pressure or from some other effect of the drug, which can be independent of blood pressure lowering.

Treat-to-target trials randomize patients to different blood pressure goals, but the groups are treated with the same or similar drugs. Therefore, any identified benefit can be attributed to the differences in blood pressure rather than the medications used. Compared with a drug treatment trial, this type of trial provides stronger evidence about optimal targets.

We also considered the characteristics of frailty, the dilemma of polypharmacy, and the relevance of the available scientific evidence to those who are frail.

Drug treatment trials

A Cochrane review⁴⁵ of 15 studies with approximately 24,000 elderly participants found that treating hypertension decreased the rates of cardiovascular morbidity and mortality as well as fatal and nonfatal stroke in the “elderly” (defined as age ≥ 60) and “very elderly” (age ≥ 80). However, in the very elderly, all-cause mortality rates were not statistically significantly different with treatment compared with placebo. The mean duration of treatment was 4.5 years in the elderly and 2.2 years in the very elderly (Table 2). Of importance, all the trials enrolled only those individuals whose systolic blood pressure was at least 160 mm Hg at baseline.

None of the studies were treat-to-target trials—patients were assigned either active medication or placebo. Thus, these trials provide evidence of benefit for treating hypertension in the elderly and very elderly but do not identify the optimal target. All of the drug treatment trials showed benefit, but none achieved a systolic pressure lower than 140 mm Hg with active treatment (Table 3). Therefore, these studies do not support a systolic target of less than 140 mm Hg in the elderly.

Treat-to-target trials: JATOS and VALISH

The Japanese Trial to Assess Optimal Systolic Blood Pressure in Elderly Hypertensive Patients (JATOS)⁴² and the Valsartan in Elderly Isolated Systolic Hypertension (VALISH) study⁴³ each enrolled more than 3,000 people age 65 or older (mean age approximately 75). Patients were randomized to either a strict systolic target of less than 140 mm Hg or a higher (more permissive) target of 140 to 160 mm Hg in JATOS and 140 to 149 mm Hg in VALISH.

In both trials, the group with strict targets achieved a systolic pressure of approximately 136 mm Hg, while the group with higher blood pressure targets achieved a systolic pressure of 146 mm Hg in JATOS and 142 mm Hg in VALISH. Despite these differences, there was no statistically significant difference in the primary outcome.

Thus, treat-to-target studies also fail to support a systolic target of less than 140 mm Hg in the elderly, although it is important to recognize the limitations of the studies. Approximately 15% of the participants had cardiovascular disease, so the applicability of the findings to patients with target-organ damage is uncertain. In addition, there were fewer efficacy outcome events than expected, which suggests that the studies were underpowered.

When to start drug treatment

In each of the drug treatment and treat-to-target trials, the inclusion criterion for study entry was a systolic blood pressure above 160 mm Hg, with a mean blood pressure at entry into the drug treatment trials of 182/95 mm Hg.⁴⁶ Thus, data support starting treatment if the systolic blood pressure is above 160 mm Hg, but not lower.

Notably, in all but one study,⁴⁶ at least two-thirds of the participants took no more than two antihypertensive medications. Since adverse events become more common as the number of medications increases, the benefit of adding a third drug to lower blood pressure is uncertain.

Evidence in the ‘very elderly’: HYVET

With the exception of the Hypertension in the Very Elderly Trial (HYVET),⁴⁴ the mean age of elderly patients in the reported studies was between 67 and 76.

HYVET patients were age 80 and older (mean age 84) and were randomized to receive either indapamide (with or without perindopril) or placebo. The trial was stopped early at 2 years because the mortality rate was lower in the treatment group (10.1%) than in the placebo group (12.3%) (number needed to treat 46, 95% confidence interval 24–637, P = .02). There was no significant difference in the primary outcome of fatal and nonfatal stroke.

Notably, trials that are stopped early may overestimate treatment benefit.⁴⁸

Evidence in frail older adults

While the above studies provide some information about managing hypertension in the elderly, the participants were generally healthy. HYVET⁴⁴ specifically excluded those with a standing systolic blood pressure of less than 140 mm Hg and enrolled few patients with orthostasis (7.9% in the placebo group and 8.8% in the treatment group), a condition commonly associated with frailty. As such, these studies may be less relevant to the frail elderly, who are at higher risk of adverse drug events and have competing risks for morbidity and mortality.

Observational studies, in fact, raise questions about whether tight blood pressure control improves clinical outcomes for the very elderly. In the Leiden 85-plus study, lower systolic blood pressure was associated with lower cognitive scores, worse functional ability,^49,50 and a higher mortality rate⁵¹ compared with higher systolic pressure, although it is uncertain whether these outcomes were indicative of underlying disease that could result in lower blood pressure or an effect of blood pressure-lowering.

The National Health and Nutrition Examination Survey⁵² found an association between blood pressure and mortality rate that varied by walking speed. For slower walkers (based on the 6-minute walk test), higher systolic pressures were not associated with a higher risk of death, suggesting that when older adults are frail (as indicated by their slow walking speed) they are less likely to benefit from aggressive treatment of hypertension.

People at high risk because of stroke

Because the evidence is limited, it is even more difficult to judge whether lowering blood pressure below 140 mm Hg is beneficial for frail patients who have a history of stroke, compared with the possibility that medications will cause adverse effects such as weakness, orthostasis, and falls. When reviewing the evidence to answer this question, we especially looked at outcomes that affect quality of life, such as nonfatal stroke leading to disability. In contrast, because the frail elderly have competing causes of mortality, we could not assume that a mortality benefit shown in nonfrail populations could be applied to frail populations.

The PROGRESS trial (Perindopril Protection Against Recurrent Stroke Study)⁵³ was in patients with a history of stroke or transient ischemic attack and a mean age of 64, who were treated with either perindopril (with or without indapamide) or placebo.

At almost 4 years, the rate of disabling stroke was 2.7% in the treatment group and 4.3% in the placebo group, a relative risk reduction of 38% and an absolute risk reduction of 1.64% (number needed to treat 61, 95% confidence interval 39–139). The relative risk reduction for all strokes (fatal and nonfatal) was similar across a range of baseline systolic pressures, but the absolute risk reduction was greater in the prespecified subgroup that had hypertension at baseline (mean blood pressure 159/94 mm Hg) than in the normotensive subgroup (mean blood pressure 136/79 mm Hg), suggesting that treatment is most beneficial for those with higher systolic blood pressures. Also, the benefit was only demonstrated in the subgroup that received two antihypertensive medications; those who received perindopril alone showed no benefit.

This study involved relatively young patients in relatively good health except for their strokes. The extent to which the results can be extrapolated to older, frail adults is uncertain because of the time needed to achieve benefit and because of the added vulnerability of frailty, which could make treatment with two antihypertensive medications riskier.

PRoFESS (Prevention Regimen for Effectively Avoiding Second Strokes),⁵⁴ another study in patients with previous stroke (mean age 66) showed no benefit over 2.5 years in the primary outcome of stroke using telmesartan 80 mg daily compared with placebo. This result is concordant with that of PROGRESS,⁵³ in which patients who took only one medication did not show a significant decrease in the rate of stroke.

A possible reason for the lack of benefit from monotherapy was that the differences in blood pressure between the placebo group and the treatment group on monotherapy were small in both studies (3.8/2.0 mm Hg in PRoFESS, 5/3 mm Hg in PROGRESS). In contrast, patients on dual therapy in PROGRESS decreased their blood pressure by 12/5 mm Hg compared with placebo.

CURRENT HYPERTENSION GUIDELINES

Current guidelines make reference to the elderly, but we found none that made specific recommendations for the frail elderly.

JNC 8

In December 2013, members of the Eighth Joint National Committee on the Prevention, Detection, Evaluation, and Treatment of High Blood Pressure (JNC 8) released new recommendations.³² One significant revision was to support higher blood pressure targets for older adults (age 60 and older). Whereas JNC 7 stated that lowering blood pressure below 140/90 mm Hg reduced cardiovascular complications,³³ JNC 8 now acknowledges that there is no strong evidence to support blood pressure targets below 150/90 mm Hg for hypertensive persons without kidney disease or diabetes age 60 and older. Thus, in the general population age 60 and older, JNC 8 recommends starting antihypertensive treatment when blood pressure is 150/90 mm Hg or higher, and treating to a goal blood pressure of less than 150/90 mm Hg. JNC 8 makes no recommendation about how to adjust blood pressure targets for frailty or how to measure blood pressure.

American College of Cardiology and American Heart Association

In 2011, the American College of Cardiology and American Heart Association published a consensus document on the management of hypertension in the elderly.⁹

They acknowledged that the generally recommended blood pressure goal of lower than 140/90 mm Hg in uncomplicated elderly patients is based on expert opinion rather than on data from randomized controlled trials, but nevertheless recommended a target systolic pressure lower than 140 mm Hg for older adults, except for octogenarians.

For those over age 80, systolic levels of 140 to 145 mm Hg can be acceptable if tolerated and if the patient does not experience orthostasis when standing. Systolic pressure lower than 130 mm Hg and diastolic pressures lower than 65 mm Hg should be avoided in this age group.

The document acknowledges that systolic pressure may have to remain above 150 mm Hg if there is no response to four “well-selected drugs” or if there are unacceptable side effects. In these cases, the lowest “safely achieved” systolic blood pressure should be the goal.

Canadian Hypertension Education Program

The 2014 Canadian Hypertension Education Program (CHEP) report makes several recommendations for the “very elderly,” a group they define as over the age of 80. The CHEP website and resources include the following recommendations¹⁰:

• For the very elderly without diabetes or target-organ damage, drug therapy should be initiated when systolic blood pressure is higher than 160 mm Hg to reach a systolic blood pressure target lower than 150 mm Hg. This is a grade C level recommendation, indicating that it is based on low-quality trials, unvalidated surrogate outcomes, or results from nonrandomized observational studies.
• For the very elderly with macrovascular target-organ damage, antihypertensive therapy should be considered if systolic blood pressure readings average 140 mm Hg or higher (grade D for 140 to 160 mm Hg; grade A for higher than 160 mm Hg), although caution should be exercised in elderly patients who are frail. (Grade D recommendations are the weakest, as they are based on low-powered, imprecise studies or expert opinion, whereas grade A recommendations are based on the strongest evidence from high-quality randomized clinical trials.)
• Decisions regarding initiating and intensifying pharmacotherapy in the very elderly should be based on an individualized risk-benefit analysis.

The European Society of Hypertension and European Society of Cardiology

The 2013 guidelines from the European Society of Hypertension and the European Society of Cardiology¹¹ recommend that for elderly patients under age 80, antihypertensive treatment may be considered at systolic values higher than 140 mm Hg and aimed at values lower than 140 mm Hg if the patient is fit and treatment is well tolerated.

For those over age 80 with an initial systolic pressure of 160 mm Hg or higher, the guidelines recommend lowering systolic pressure to between 150 and 140 mm Hg, provided the patient is in good physical and mental condition. In frail elderly patients, they recommend leaving decisions on antihypertensive therapy to the treating physician, based on monitoring of the clinical effects of treatment.¹¹

The ADS/PATH guidelines

When finalizing our recommendations,¹ we considered the characteristics of frailty and the following key points from the evidence:

• Although evidence from drug treatment trials indicates that there is benefit in treating healthy older adults who have hypertension, the benefit of treating frail older adults is unknown.
• Major trials enrolled elderly patients only if they had systolic blood pressures of at least 160 mm Hg. Therefore, evidence supports initiating pharmacotherapy at a systolic pressure of 160 mm Hg or higher.
• No evidence from randomized controlled trials supports a systolic target lower than 140 mm Hg in the elderly, and there is some evidence that such a target does not benefit.
• The benefit of adding a third medication to lower blood pressure has not been studied.
• Frailty makes the potential benefits of strict blood pressure targets even less certain and increases the possibility of harm from adverse drug events.
• The only study of very old adults, HYVET,⁴⁴ enrolled relatively healthy older adults and few with orthostasis, while excluding those with a standing systolic blood pressure lower than 140 mm Hg.

OUR RECOMMENDATIONS

Based on the above, we advise against unnecessarily strict targets and recommend stopping antihypertensive medications that are used for the sole purpose of keeping the systolic blood pressure below 140 mm Hg. Our guidelines are unique in that they focus equally on when to stop and when to start medications. We concluded that without evidence of definitive benefit, “less is more” with frailty.⁵⁵ We believe that if physicians and health professionals understand the limitations of the evidence, they can be more confident in stopping medications that lower blood pressure to an unnecessarily low level.

We recommend the following (Table 4):

Before treating

• Carefully review the risks and the potential but unproven benefits of treatment.
• To avoid overtreatment, treatment decisions should be based on blood pressure measurements in the seated (not supine) position, while also considering the presence of orthostasis.
• To evaluate orthostasis, measure blood pressure in the supine position, then immediately on standing, and again after 2 minutes. Ask the patient if he or she feels light-headed or dizzy when standing.

Stop treatment

• If the seated systolic blood pressure is less than 140 mm Hg, medications can be tapered and discontinued to achieve the targets described below.
• Before discontinuation, consider whether the medications are treating additional conditions such as rate control for atrial fibrillation or symptomatic management of heart failure.
• It is uncertain whether to discontinue treatment when there is a history of stroke. Consider that treatment with two medications resulted in an absolute risk reduction for disabling stroke of 1.64% over approximately 4 years for adults with previous stroke and a mean age of 64,⁵⁷ an effect that may be more prominent at higher systolic pressures.

Start treatment

• Consider starting treatment when systolic pressure is 160 mm Hg or higher.
• Aim for a seated systolic pressure between 140 and 160 mm Hg if there are no adverse effects from treatment that affect quality of life.
• If there is symptomatic orthostasis or if standing systolic pressure is lower than 140 mm Hg, the target seated systolic pressure can be adjusted upwards.
• In the severely frail nearing the end of life, a target systolic pressure of 160 to 190 mm Hg is reasonable.
• The blood pressure target is the same in people with diabetes.
• In general, use no more than two medications.

Dissemination and implementation

The ADS/PATH guideline is intended for use by physicians and other health professionals (eg, pharmacists and nurses) who care for frail older adults or who work in long-term care facilities. Since creating our guideline, we have disseminated it to physicians, pharmacists, and other health professionals through academic detailing, large conferences, and interactive webinars.

While we do not have objective evidence of practice change, our evaluation data found that 34% of 403 family physicians who received academic detailing indicated that the guideline would change their practice, while 36% stated that the guideline confirmed their practice, an indication that family physicians are sensitive to the needs of the frail elderly.

Because health professionals may be wary of stopping medications and not meeting recommended targets, there may be barriers to adopting this guideline. However, our experience with the PATH program indicates that these barriers can be overcome using effective communication strategies between health professionals and consumers.

AN APPROACH APPROPRIATE TO FRAILTY

There is no direct evidence for systolic blood pressure targets in the frail elderly, so we applied evidence from the nonfrail elderly. Our recommendations differ somewhat from those of other groups, which recommend targets below 140 to 150 mm Hg for older adults, although some do advise caution in the elderly for whom a substantial fall in blood pressure might be poorly tolerated. Despite these messages, we believe that clearer guidance is needed to direct health practitioners toward models that acknowledge that frail patients are in a precarious balance of health and may be harmed by treatments that strive to lower blood pressure to unproven targets. For this reason, our guideline clearly indicates when to decrease or stop drug treatment.

After physicians and health professionals examine the evidence and more fully understand the benefits and harms of treating frail older adults, we are confident that they will be more comfortable stopping medications that lower blood pressure to an unnecessarily low level and instead use an approach that is more appropriate to frailty. We hope clinicians can use this guideline with the same enthusiasm applied to other guidelines, and we welcome discussion.

Acknowledgments: We would like to thank and acknowledge Tanya MacLeod and Kathryn Yuill for their review of and advice about the manuscript.

Frail older adults deserve guidelines that take frailty into account while assessing the potential benefit and risks of treatment.

Specifically, our group—the Dalhousie Academic Detailing Service (ADS) and the Palliative and Therapeutic Harmonization (PATH) program—recommends that physicians strive to achieve more liberal treatment targets for elderly frail patients who have high blood pressure,¹ as evidence does not support an aggressive approach in the frail elderly and the potential exists for harm.

This article reviews the evidence and reasoning that were used to develop and promote a guideline for drug treatment of hypertension in frail older adults. Our recommendations differ from other guidelines in that they focus as much on stopping or decreasing therapy as on starting or increasing it.

FRAILTY INCREASES THE RISK OF ADVERSE EFFECTS

The word frail, applied to older adults, describes those who have complex medical illnesses severe enough to compromise their ability to live independently.² Many have multiple coexisting medical problems for which they take numerous drugs, in addition to dementia, impaired mobility, compromised functional ability, or a history of falling.

Frailty denotes vulnerability; it increases the risk of adverse effects from medical and surgical procedures,³ complicates drug therapy,⁴ prolongs hospital length of stay,⁵ leads to functional and cognitive decline,⁶ increases the risk of institutionalization,⁷ and reduces life expectancy⁸—all of which affect the benefit and harm of medical treatments.

Guidelines for treating hypertension^9–11 now acknowledge that little evidence exists to support starting treatment for systolic blood pressure between 140 and 160 mm Hg or aiming for a target of less than 140 mm Hg for “very old” adults, commonly defined as over the age of 80. New guidelines loosen the treatment targets for the very old, but they do not specify targets for the frail and do not describe how to recognize or measure frailty.

RECOGNIZING AND MEASURING FRAILTY

A number of tools are available to recognize and measure frailty.¹²

The Fried frailty assessment¹³ has five items:

• Unintentional weight loss
• Self-reported exhaustion
• Weakness in grip
• Slow walking speed
• Low physical activity and energy expenditure.

People are deemed frail if they have three or more of these five. However, experts disagree about whether this system is too sensitive¹⁴ or not sensitive enough.^15,16

The FRAIL questionnaire¹⁷ also has five items:

• Fatigue
• Resistance (inability to climb stairs)
• Ambulation (inability to walk 1 city block)
• Illness (more than 5 major illnesses)
• Weight loss.

People are deemed frail if they have at least three of these five items, and “prefrail” if they have two.

These and other tools are limited by being dichotomous: they classify people as being either frail or not frail^18–20 but do not define the spectrum of frailty.

Other frailty assessments such as the Frailty Index²¹ identify frailty based on the number of accumulated health deficits but take a long time to complete, making them difficult to use in busy clinical settings.^22–24

The Clinical Frailty Scale⁷ is a validated scale that categorizes frailty based on physical and functional indicators of health, such as cognition, function, and mobility, with scores that range from 1 (very fit) to 9 (terminally ill).^7,12

The Frailty Assessment for Care-planning Tool (FACT) uses scaling compatible with the Clinical Frailty Scale but has been developed for use as a practical and interpretable frailty screening tool for nonexperts (Table 1). The FACT assesses cognition, mobility, function, and the social situation, using a combination of caregiver report and objective measures. To assess cognition, a health care professional uses items from the Mini-Cog²⁵ (ie, the ability to draw an analog clock face and then recall three unrelated items following the clock-drawing test) and the memory axis of the Brief Cognitive Rating Scale²⁶ (ie, the ability to recall current events, the current US president, and the names of children or spouse). Mobility, function, and social circumstance scores are assigned according to the caregiver report of the patient’s baseline status.

The FACT can be completed in busy clinical settings. Once a caregiver is identified, it takes about 5 minutes to complete.

Our guideline^27–31 is intended for those with a score of 7 or more on the Clinical Frailty Scale or FACT,^7,12 a score we chose because it describes people who are severely frail with shortened life expectancy.⁸ At this level, people need help with all instrumental activities of daily living (eg, handling finances, medication management, household chores, and shopping) as well as with basic activities of daily living such as bathing or dressing.

REVIEWING THE LIMITED EVIDENCE

We found no studies that addressed the risks and benefits of treating hypertension in frail older adults; therefore, we concentrated on studies that enrolled individuals who were chronologically old but not frail. We reviewed prominent guidelines,^9–11,32,33 the evidence base for these guidelines,^34–44 and Cochrane reviews.^45,46 A detailed description of the evidence used to build our recommendation can be found online.³¹

When we deliberated on treatment targets, we reviewed evidence from two types of randomized controlled trials⁴⁷:

Drug treatment trials randomize patients to different treatments, such as placebo versus a drug or one drug compared with another drug. Patients in different treatment groups may achieve different blood pressures and clinical outcomes, and this information is then used to define optimal targets. However, it may be difficult to determine if the benefit came from lowering blood pressure or from some other effect of the drug, which can be independent of blood pressure lowering.

Treat-to-target trials randomize patients to different blood pressure goals, but the groups are treated with the same or similar drugs. Therefore, any identified benefit can be attributed to the differences in blood pressure rather than the medications used. Compared with a drug treatment trial, this type of trial provides stronger evidence about optimal targets.

We also considered the characteristics of frailty, the dilemma of polypharmacy, and the relevance of the available scientific evidence to those who are frail.

Drug treatment trials

A Cochrane review⁴⁵ of 15 studies with approximately 24,000 elderly participants found that treating hypertension decreased the rates of cardiovascular morbidity and mortality as well as fatal and nonfatal stroke in the “elderly” (defined as age ≥ 60) and “very elderly” (age ≥ 80). However, in the very elderly, all-cause mortality rates were not statistically significantly different with treatment compared with placebo. The mean duration of treatment was 4.5 years in the elderly and 2.2 years in the very elderly (Table 2). Of importance, all the trials enrolled only those individuals whose systolic blood pressure was at least 160 mm Hg at baseline.

None of the studies were treat-to-target trials—patients were assigned either active medication or placebo. Thus, these trials provide evidence of benefit for treating hypertension in the elderly and very elderly but do not identify the optimal target. All of the drug treatment trials showed benefit, but none achieved a systolic pressure lower than 140 mm Hg with active treatment (Table 3). Therefore, these studies do not support a systolic target of less than 140 mm Hg in the elderly.

Treat-to-target trials: JATOS and VALISH

The Japanese Trial to Assess Optimal Systolic Blood Pressure in Elderly Hypertensive Patients (JATOS)⁴² and the Valsartan in Elderly Isolated Systolic Hypertension (VALISH) study⁴³ each enrolled more than 3,000 people age 65 or older (mean age approximately 75). Patients were randomized to either a strict systolic target of less than 140 mm Hg or a higher (more permissive) target of 140 to 160 mm Hg in JATOS and 140 to 149 mm Hg in VALISH.

In both trials, the group with strict targets achieved a systolic pressure of approximately 136 mm Hg, while the group with higher blood pressure targets achieved a systolic pressure of 146 mm Hg in JATOS and 142 mm Hg in VALISH. Despite these differences, there was no statistically significant difference in the primary outcome.

Thus, treat-to-target studies also fail to support a systolic target of less than 140 mm Hg in the elderly, although it is important to recognize the limitations of the studies. Approximately 15% of the participants had cardiovascular disease, so the applicability of the findings to patients with target-organ damage is uncertain. In addition, there were fewer efficacy outcome events than expected, which suggests that the studies were underpowered.

When to start drug treatment

In each of the drug treatment and treat-to-target trials, the inclusion criterion for study entry was a systolic blood pressure above 160 mm Hg, with a mean blood pressure at entry into the drug treatment trials of 182/95 mm Hg.⁴⁶ Thus, data support starting treatment if the systolic blood pressure is above 160 mm Hg, but not lower.

Notably, in all but one study,⁴⁶ at least two-thirds of the participants took no more than two antihypertensive medications. Since adverse events become more common as the number of medications increases, the benefit of adding a third drug to lower blood pressure is uncertain.

Evidence in the ‘very elderly’: HYVET

With the exception of the Hypertension in the Very Elderly Trial (HYVET),⁴⁴ the mean age of elderly patients in the reported studies was between 67 and 76.

HYVET patients were age 80 and older (mean age 84) and were randomized to receive either indapamide (with or without perindopril) or placebo. The trial was stopped early at 2 years because the mortality rate was lower in the treatment group (10.1%) than in the placebo group (12.3%) (number needed to treat 46, 95% confidence interval 24–637, P = .02). There was no significant difference in the primary outcome of fatal and nonfatal stroke.

Notably, trials that are stopped early may overestimate treatment benefit.⁴⁸

Evidence in frail older adults

While the above studies provide some information about managing hypertension in the elderly, the participants were generally healthy. HYVET⁴⁴ specifically excluded those with a standing systolic blood pressure of less than 140 mm Hg and enrolled few patients with orthostasis (7.9% in the placebo group and 8.8% in the treatment group), a condition commonly associated with frailty. As such, these studies may be less relevant to the frail elderly, who are at higher risk of adverse drug events and have competing risks for morbidity and mortality.

Observational studies, in fact, raise questions about whether tight blood pressure control improves clinical outcomes for the very elderly. In the Leiden 85-plus study, lower systolic blood pressure was associated with lower cognitive scores, worse functional ability,^49,50 and a higher mortality rate⁵¹ compared with higher systolic pressure, although it is uncertain whether these outcomes were indicative of underlying disease that could result in lower blood pressure or an effect of blood pressure-lowering.

The National Health and Nutrition Examination Survey⁵² found an association between blood pressure and mortality rate that varied by walking speed. For slower walkers (based on the 6-minute walk test), higher systolic pressures were not associated with a higher risk of death, suggesting that when older adults are frail (as indicated by their slow walking speed) they are less likely to benefit from aggressive treatment of hypertension.

People at high risk because of stroke

Because the evidence is limited, it is even more difficult to judge whether lowering blood pressure below 140 mm Hg is beneficial for frail patients who have a history of stroke, compared with the possibility that medications will cause adverse effects such as weakness, orthostasis, and falls. When reviewing the evidence to answer this question, we especially looked at outcomes that affect quality of life, such as nonfatal stroke leading to disability. In contrast, because the frail elderly have competing causes of mortality, we could not assume that a mortality benefit shown in nonfrail populations could be applied to frail populations.

The PROGRESS trial (Perindopril Protection Against Recurrent Stroke Study)⁵³ was in patients with a history of stroke or transient ischemic attack and a mean age of 64, who were treated with either perindopril (with or without indapamide) or placebo.

At almost 4 years, the rate of disabling stroke was 2.7% in the treatment group and 4.3% in the placebo group, a relative risk reduction of 38% and an absolute risk reduction of 1.64% (number needed to treat 61, 95% confidence interval 39–139). The relative risk reduction for all strokes (fatal and nonfatal) was similar across a range of baseline systolic pressures, but the absolute risk reduction was greater in the prespecified subgroup that had hypertension at baseline (mean blood pressure 159/94 mm Hg) than in the normotensive subgroup (mean blood pressure 136/79 mm Hg), suggesting that treatment is most beneficial for those with higher systolic blood pressures. Also, the benefit was only demonstrated in the subgroup that received two antihypertensive medications; those who received perindopril alone showed no benefit.

This study involved relatively young patients in relatively good health except for their strokes. The extent to which the results can be extrapolated to older, frail adults is uncertain because of the time needed to achieve benefit and because of the added vulnerability of frailty, which could make treatment with two antihypertensive medications riskier.

PRoFESS (Prevention Regimen for Effectively Avoiding Second Strokes),⁵⁴ another study in patients with previous stroke (mean age 66) showed no benefit over 2.5 years in the primary outcome of stroke using telmesartan 80 mg daily compared with placebo. This result is concordant with that of PROGRESS,⁵³ in which patients who took only one medication did not show a significant decrease in the rate of stroke.

A possible reason for the lack of benefit from monotherapy was that the differences in blood pressure between the placebo group and the treatment group on monotherapy were small in both studies (3.8/2.0 mm Hg in PRoFESS, 5/3 mm Hg in PROGRESS). In contrast, patients on dual therapy in PROGRESS decreased their blood pressure by 12/5 mm Hg compared with placebo.

CURRENT HYPERTENSION GUIDELINES

Current guidelines make reference to the elderly, but we found none that made specific recommendations for the frail elderly.

JNC 8

In December 2013, members of the Eighth Joint National Committee on the Prevention, Detection, Evaluation, and Treatment of High Blood Pressure (JNC 8) released new recommendations.³² One significant revision was to support higher blood pressure targets for older adults (age 60 and older). Whereas JNC 7 stated that lowering blood pressure below 140/90 mm Hg reduced cardiovascular complications,³³ JNC 8 now acknowledges that there is no strong evidence to support blood pressure targets below 150/90 mm Hg for hypertensive persons without kidney disease or diabetes age 60 and older. Thus, in the general population age 60 and older, JNC 8 recommends starting antihypertensive treatment when blood pressure is 150/90 mm Hg or higher, and treating to a goal blood pressure of less than 150/90 mm Hg. JNC 8 makes no recommendation about how to adjust blood pressure targets for frailty or how to measure blood pressure.

American College of Cardiology and American Heart Association

In 2011, the American College of Cardiology and American Heart Association published a consensus document on the management of hypertension in the elderly.⁹

They acknowledged that the generally recommended blood pressure goal of lower than 140/90 mm Hg in uncomplicated elderly patients is based on expert opinion rather than on data from randomized controlled trials, but nevertheless recommended a target systolic pressure lower than 140 mm Hg for older adults, except for octogenarians.

For those over age 80, systolic levels of 140 to 145 mm Hg can be acceptable if tolerated and if the patient does not experience orthostasis when standing. Systolic pressure lower than 130 mm Hg and diastolic pressures lower than 65 mm Hg should be avoided in this age group.

The document acknowledges that systolic pressure may have to remain above 150 mm Hg if there is no response to four “well-selected drugs” or if there are unacceptable side effects. In these cases, the lowest “safely achieved” systolic blood pressure should be the goal.

Canadian Hypertension Education Program

The 2014 Canadian Hypertension Education Program (CHEP) report makes several recommendations for the “very elderly,” a group they define as over the age of 80. The CHEP website and resources include the following recommendations¹⁰:

• For the very elderly without diabetes or target-organ damage, drug therapy should be initiated when systolic blood pressure is higher than 160 mm Hg to reach a systolic blood pressure target lower than 150 mm Hg. This is a grade C level recommendation, indicating that it is based on low-quality trials, unvalidated surrogate outcomes, or results from nonrandomized observational studies.
• For the very elderly with macrovascular target-organ damage, antihypertensive therapy should be considered if systolic blood pressure readings average 140 mm Hg or higher (grade D for 140 to 160 mm Hg; grade A for higher than 160 mm Hg), although caution should be exercised in elderly patients who are frail. (Grade D recommendations are the weakest, as they are based on low-powered, imprecise studies or expert opinion, whereas grade A recommendations are based on the strongest evidence from high-quality randomized clinical trials.)
• Decisions regarding initiating and intensifying pharmacotherapy in the very elderly should be based on an individualized risk-benefit analysis.

The European Society of Hypertension and European Society of Cardiology

The 2013 guidelines from the European Society of Hypertension and the European Society of Cardiology¹¹ recommend that for elderly patients under age 80, antihypertensive treatment may be considered at systolic values higher than 140 mm Hg and aimed at values lower than 140 mm Hg if the patient is fit and treatment is well tolerated.

For those over age 80 with an initial systolic pressure of 160 mm Hg or higher, the guidelines recommend lowering systolic pressure to between 150 and 140 mm Hg, provided the patient is in good physical and mental condition. In frail elderly patients, they recommend leaving decisions on antihypertensive therapy to the treating physician, based on monitoring of the clinical effects of treatment.¹¹

The ADS/PATH guidelines

When finalizing our recommendations,¹ we considered the characteristics of frailty and the following key points from the evidence:

• Although evidence from drug treatment trials indicates that there is benefit in treating healthy older adults who have hypertension, the benefit of treating frail older adults is unknown.
• Major trials enrolled elderly patients only if they had systolic blood pressures of at least 160 mm Hg. Therefore, evidence supports initiating pharmacotherapy at a systolic pressure of 160 mm Hg or higher.
• No evidence from randomized controlled trials supports a systolic target lower than 140 mm Hg in the elderly, and there is some evidence that such a target does not benefit.
• The benefit of adding a third medication to lower blood pressure has not been studied.
• Frailty makes the potential benefits of strict blood pressure targets even less certain and increases the possibility of harm from adverse drug events.
• The only study of very old adults, HYVET,⁴⁴ enrolled relatively healthy older adults and few with orthostasis, while excluding those with a standing systolic blood pressure lower than 140 mm Hg.

OUR RECOMMENDATIONS

Based on the above, we advise against unnecessarily strict targets and recommend stopping antihypertensive medications that are used for the sole purpose of keeping the systolic blood pressure below 140 mm Hg. Our guidelines are unique in that they focus equally on when to stop and when to start medications. We concluded that without evidence of definitive benefit, “less is more” with frailty.⁵⁵ We believe that if physicians and health professionals understand the limitations of the evidence, they can be more confident in stopping medications that lower blood pressure to an unnecessarily low level.

We recommend the following (Table 4):

Before treating

• Carefully review the risks and the potential but unproven benefits of treatment.
• To avoid overtreatment, treatment decisions should be based on blood pressure measurements in the seated (not supine) position, while also considering the presence of orthostasis.
• To evaluate orthostasis, measure blood pressure in the supine position, then immediately on standing, and again after 2 minutes. Ask the patient if he or she feels light-headed or dizzy when standing.

Stop treatment

• If the seated systolic blood pressure is less than 140 mm Hg, medications can be tapered and discontinued to achieve the targets described below.
• Before discontinuation, consider whether the medications are treating additional conditions such as rate control for atrial fibrillation or symptomatic management of heart failure.
• It is uncertain whether to discontinue treatment when there is a history of stroke. Consider that treatment with two medications resulted in an absolute risk reduction for disabling stroke of 1.64% over approximately 4 years for adults with previous stroke and a mean age of 64,⁵⁷ an effect that may be more prominent at higher systolic pressures.

Start treatment

• Consider starting treatment when systolic pressure is 160 mm Hg or higher.
• Aim for a seated systolic pressure between 140 and 160 mm Hg if there are no adverse effects from treatment that affect quality of life.
• If there is symptomatic orthostasis or if standing systolic pressure is lower than 140 mm Hg, the target seated systolic pressure can be adjusted upwards.
• In the severely frail nearing the end of life, a target systolic pressure of 160 to 190 mm Hg is reasonable.
• The blood pressure target is the same in people with diabetes.
• In general, use no more than two medications.

Dissemination and implementation

The ADS/PATH guideline is intended for use by physicians and other health professionals (eg, pharmacists and nurses) who care for frail older adults or who work in long-term care facilities. Since creating our guideline, we have disseminated it to physicians, pharmacists, and other health professionals through academic detailing, large conferences, and interactive webinars.

While we do not have objective evidence of practice change, our evaluation data found that 34% of 403 family physicians who received academic detailing indicated that the guideline would change their practice, while 36% stated that the guideline confirmed their practice, an indication that family physicians are sensitive to the needs of the frail elderly.

Because health professionals may be wary of stopping medications and not meeting recommended targets, there may be barriers to adopting this guideline. However, our experience with the PATH program indicates that these barriers can be overcome using effective communication strategies between health professionals and consumers.

AN APPROACH APPROPRIATE TO FRAILTY

There is no direct evidence for systolic blood pressure targets in the frail elderly, so we applied evidence from the nonfrail elderly. Our recommendations differ somewhat from those of other groups, which recommend targets below 140 to 150 mm Hg for older adults, although some do advise caution in the elderly for whom a substantial fall in blood pressure might be poorly tolerated. Despite these messages, we believe that clearer guidance is needed to direct health practitioners toward models that acknowledge that frail patients are in a precarious balance of health and may be harmed by treatments that strive to lower blood pressure to unproven targets. For this reason, our guideline clearly indicates when to decrease or stop drug treatment.

After physicians and health professionals examine the evidence and more fully understand the benefits and harms of treating frail older adults, we are confident that they will be more comfortable stopping medications that lower blood pressure to an unnecessarily low level and instead use an approach that is more appropriate to frailty. We hope clinicians can use this guideline with the same enthusiasm applied to other guidelines, and we welcome discussion.

Acknowledgments: We would like to thank and acknowledge Tanya MacLeod and Kathryn Yuill for their review of and advice about the manuscript.

Your patient has had an ischemic stroke, and so far you have found no obvious cause such as atrial fibrillation or carotid disease. Should you look for a patent foramen ovale (PFO)? And if you find it, what should you do?

See related commentary

This scenario continues to challenge primary care physicians and subspecialists and requires an understanding of the relationship between PFO and cryptogenic stroke, as well as familiarity with current data on the safety and effectiveness of the management options. PFO is known to be associated with cryptogenic stroke, but many questions remain, including:

• How can we tell if PFO is a culprit (“pathologic”) or an innocent bystander (“incidental”) in a patient who has had a cryptogenic stroke?
• Should stroke patients receive different medical therapy if they have a PFO? In particular, should they receive warfarin in addition to aspirin? And what about the novel oral anticoagulants?
• Which patients should undergo percutaneous closure of the PFO?
• Should we even be looking for PFO in stroke patients at this point, if we cannot say with certainty what we should do if we find it?

WHY IS THIS IMPORTANT?

Cerebrovascular disease is common and costly. The estimated yearly incidence of stroke in the United States is 795,000 events, at a cost of nearly $30 billion.¹ The incidence of stroke in Europe is more than 1 million annually.²

During the diagnostic evaluation of stroke or transient ischemic attack (TIA), PFO is occasionally discovered incidentally by echocardiography. The management decisions that follow often fall to the primary care physician, who must decipher the conflicting data currently available and explain the options to the patient.

Although reviews have been published on this subject,³ several newer key trials and data on risk stratification warrant consideration.

DEFINITIONS

PFO is the failure of the septum primum to fuse with the septum secundum, so that a communication remains between the atria (Figure 1). The diagnosis is commonly made by echocardiography, when agitated saline is injected into the venous system and bubbles can be seen in the left atrium within three to five cardiac cycles (see video).

Atrial septal aneurysm is loosely defined as a septal excursion or bulging of at least 10 to 15 mm into the left and right atria during the cardiac cycle (Figure 2). The combination of PFO and atrial septal aneurysm may be more of a risk factor for stroke than PFO alone (see discussion below).

Cryptogenic stroke. The diagnostic workup of stroke fails to elucidate a clear cause in up to 40% of cases, which are thus called cryptogenic.⁴ The workup varies, but typically includes a search for a cardioembolic source and for atherosclerotic disease. Embolic sources are evaluated for by electrocardiography, transthoracic echocardiography, and possibly imaging of the aortic arch. Evaluation for atherosclerotic disease of the intracranial and extracranial arteries includes magnetic resonance angiography or, if that is unavailable, computed tomographic angiography or carotid Doppler ultrasonography. If no source is found, long-term cardiac monitoring may be used to detect paroxysmal atrial fibrillation, which may be more common than previously thought.

PFO AND CRYPTOGENIC STROKE ARE COMMON

As noted, there are approximately 800,000 strokes every year in the United States. If 25% to 40% of them are cryptogenic (the true prevalence warrants more evaluation),^4,5 then 200,000 to 320,000 strokes are cryptogenic.

Autopsy studies indicate that 25% of the general population have a PFO, and if the prevalence is the same in people with cryptogenic stroke, that would equal 80,000 people with both cryptogenic stroke and PFO every year. However, the prevalence of PFO in patients with cryptogenic stroke appears to be significantly higher than in the general population.⁶ Although these numbers are crude estimates, they provide some insight into the prevalence of this clinical presentation.

HOW ARE CRYPTOGENIC STROKE AND PFO RELATED?

The exact relationship between PFO and cryptogenic stroke is unknown, although cases have been reported of thrombus in transit through a PFO, supporting paradoxical embolism as the plausible cause in stroke patients with PFO.^7–9

There is clear evidence that the two conditions are associated by more than chance. Homma and Sacco⁶ reported that, in several studies, 93 (46%) of 202 patients under age 55 with cryptogenic stroke had PFOs, compared with 29 (11%) of 271 controls (P < .05 in all studies).⁶

In their evaluation of 23 case-control studies, Alsheikh-Ali et al¹⁰ found that the summary odds ratio (OR) for PFO in cryptogenic stroke vs PFO in control patients was 2.9 (95% confidence interval [CI] 2.1–4), largely driven by an OR of 5.1 (3.3–7.8) in those under age 55. Through Bayesian probability theory, this correlated with only a 33% probability that PFO in a patient with cryptogenic stroke was an innocent bystander rather than the culprit.¹⁰

IS PFO A RISK FACTOR FOR STROKE?

One of the more puzzling aspects of the relationship of PFO to cryptogenic stroke is that despite a clear association, there is little evidence that the relationship is causal.

Di Tullio et al¹¹ followed 1,100 people who had no history of stroke and found that the risk of a first stroke in those with a PFO was not significantly higher than in those without a PFO, regardless of age, sex, or ethnic or racial group. At 80 months, the hazard ratio of stroke in people who had a PFO was 1.64 (95% CI 0.87–3.09).¹¹ The findings were similar at 11 years, with a hazard ratio of 1.10 (95% CI 0.64–1.91).¹²

A prospective study of 585 patients found a similar risk of stroke in those with and without a PFO, with a hazard ratio of 1.46 (95% CI 0.74–2.88; P = .28).¹³

These prospective trials suggest that although previous studies have found a higher prevalence of PFOs in patients with cryptogenic stroke than in patients without stroke, there appears to be very little if any increased risk from baseline for a first stroke or TIA.

The lack of statistical significance in these trials should be interpreted with some caution, as a small increased risk is difficult to show if the event rate is low (approximately 10% of patients had events over 11 years in the study by Di Tullio et al¹²).

HOW DO WE KNOW IF A PFO IS A CULPRIT OR BYSTANDER?

Unfortunately, this is largely unanswered, though experts have suggested that echocardiographic features of the PFO, radiographic characteristics of the stroke, and clinical features of the patient may provide useful information.

‘High-risk’ features on echocardiography

Certain features of PFO may portend a high risk of cerebrovascular events. Both right-to-left shunting at rest and septal hypermobility were found in one study¹⁴ to be more common in patients with a PFO who had a stroke or TIA than in patients with a PFO but no cerebrovascular events. Also, patients who had these features and had a stroke had a higher risk of recurrence than stroke patients without these features (12.5% vs 4.3%, P = .05).¹⁴

Septal hypermobility and shunting at rest are easily diagnosed by echocardiography, and detecting these “high-risk” features would be useful if they could identify patients who would benefit from special therapy, such as percutaneous closure of the PFO.

Unfortunately, when investigators looked at these features in subgroup analysis of the major randomized controlled trials of percutaneous closure vs medical therapy, the results were mixed.

CLOSURE 1 (the Evaluation of the STARFlex Septal Closure System in Patients With a Stroke and/or Transient Ischemic Attack Due to Presumed Paradoxical Embolism Through a Patent Foramen Ovale)¹⁵ found percutaneous closure to be no better than medical therapy, regardless of shunt size or the presence of atrial septal aneurysm.

Similarly, the PC trial (Clinical Trial Comparing Percutaneous Closure of Patent Foramen Ovale Using the Amplatzer PFO Occluder With Medical Treatment in Patients With Cryptogenic Embolism)¹⁶ found no statistically significant benefit of closure in those with atrial septal aneurysm.

In contrast, the RESPECT trial (Randomized Evaluation of Recurrent Stroke Comparing PFO Closure to Established Current Standard of Care Treatment)¹⁷ showed percutaneous closure to be beneficial in patients with atrial septal aneurysm or large shunt.

Radiographic characteristics of the stroke

Another area of interest in trying to identify culprit PFOs is the radiographic characteristics of the stroke.

In a study comparing patients with stroke related to atrial fibrillation vs patients with cryptogenic stroke and a known PFO, those in the latter group were more likely to have a single cortical infarction (34.2% vs 3.1%; P < .001) or multiple small scattered lesions (23.1% vs 5.9%; P < .01).¹⁸ Similarly, in a large database of patients with cryptogenic stroke and known PFO status, a superficially located stroke was associated with the presence of PFO (OR 1.54; P < .0001).¹⁹

Although these findings do not tell us with certainty that a patient’s PFO was the cause of his or her stroke, they provide guidance when dealing with the uncertainty of how to manage a patient with PFO. They may be useful in clinical practice, for example, when discussing treatment options with a young patient with cryptogenic stroke who has no risk factors and a superficial single infarct and who is found to have a PFO with a right-to-left shunting at rest.

Patient characteristics

Kent et al²⁰ developed a 10-point index (the RoPE score) in an attempt to assign a probability to whether a stroke was PFO-related. Points were assigned for patients who were younger, who had a cortical stroke on neuroimaging, and who did not have diabetes, hypertension, smoking, or prior stroke or TIA. Patients with cryptogenic stroke with a higher RoPE score were more likely to have a PFO and thus had a higher likelihood that the index event was related to PFO. Of note, the patients with the highest likelihood of PFO-related stroke were the least likely to have a recurrence (RoPE score of 9 to 10; PFO-attributable fraction 88%; estimated 2-year recurrence rate 2%; 95% CI 0%–4%), whereas those with a low RoPE score have more traditional risk factors for stroke and thus are more likely to have a recurrence (RoPE 0 to 3; estimated 2-year recurrence rate 20%; 95% CI 12%–28%).²⁰

Again, this sheds light on a difficulty faced by randomized controlled trials: the patients who may benefit from closure of a PFO may very well be those with the lowest recurrence rates without intervention.

The RoPE index was examined in an attempt to validate previously described morphologic criteria of “high-risk” PFO,²¹ though none of the previously described “high-risk” echocardiographic features (large physiologic size, hypermobile septum, shunt at rest) were more common in the group with presumed PFO-attributable stroke (RoPE score > 6). This underscores the difficulty of distinguishing pathologic PFO from incidental PFO.

KEY TREATMENT CONSIDERATIONS FOR SECONDARY PREVENTION

Given the complicated relationship between PFO and cryptogenic stroke, there has been much debate over management strategies. The three options are surgical closure, percutaneous closure with a device, and medical therapy. The goal of all three is to prevent the recurrence of stroke or TIA.

Surgical closure has largely been supplanted by percutaneous closure, but is still done in specific situations such as when a PFO is found incidentally on transesophageal echocardiography during surgery for another cardiac condition. The data on such cases²² tend to support the argument that asymptomatic PFOs in the general population have a relatively benign natural history.

Thus, the two key questions about management that warrant discussion are: is anticoagulation superior to antiplatelet therapy? And is percutaneous closure superior to medical management?

Anticoagulant vs antiplatelet therapy

Whether to treat with aspirin or with a vitamin K antagonist has been a subject of debate, although there is no strong evidence to suggest that anticoagulation is superior to antiplatelet therapy.

The concern that aspirin alone is insufficient in some patients stems from a study by Mas et al,²³ who followed 581 patients with cryptogenic stroke who had a PFO alone, a PFO with an atrial septal aneurysm, or neither. The rate of stroke recurrence at 4 years on aspirin therapy was 2.3% in those with a PFO alone, 15.2% in those with a PFO with an atrial septal aneurysm, and 4.2% in those with neither.

Many have concluded that aspirin therapy does not sufficiently protect those with both PFO and atrial septal aneurysm, given the high recurrence rate in this group. This might lead to the suggestion that anticoagulation could be of benefit in these patients.

However, the Patent Foramen Ovale in Cryptogenic Stroke Study (PiCSS)²⁴ and the Spanish Multicenter Study Into Right-to-Left Shunt in Cryptogenic Stroke (CODICIA)²⁵ found similar recurrence rates in patients with PFO and atrial septal aneurysm compared with those with only PFO. In these two studies, recurrence rates were similar regardless of whether patients were taking aspirin or warfarin.

In a study that followed 140 consecutive patients with both stroke and PFO, those treated in a nonrandomized fashion with antiplatelet agents had no difference in the recurrence rate compared with those treated with anticoagulation.²⁶

Although uncertainty remains because no head-to-head randomized controlled trial has been done, some patients with PFO have other indications for anticoagulation, most commonly atrial fibrillation and venous thromboembolic disease.

There are currently no data on the use of novel oral anticoagulants in this setting.

Is percutaneous closure better than medical therapy?

When cryptogenic stroke is treated with antiplatelet therapy or anticoagulation therapy, the recurrence rate is the same whether or not the patient has a PFO.^23–25 The belief that medical therapy offers adequate secondary protection is supported by a meta-analysis of 15 studies that found no increased risk of recurrent ischemic events in those with a PFO on medical therapy (antiplatelet or anticoagulant) vs those without a PFO (relative risk 1.1, 95% CI 0.8–1.5).²⁷

Despite the conflicting evidence, percutaneous closure of PFO is still performed, mostly on a case-by-case basis. This has been supported by an apparent benefit in observational studies.

A systematic review of 52 single-arm studies and 7 comparative nonrandomized studies of patients with PFO and cryptogenic stroke found the rate of recurrent stroke to be 0.36 per 100 person-years with percutaneous closure vs 2.53 per 100 person-years with medical therapy.²⁸ However, three long-awaited randomized controlled trials (CLOSURE 1, the PC trial, and RESPECT) failed to show a significant reduction in primary end points with percutaneous closure vs standard medical therapy.^15–17

These trials had several limitations: event rates were low, medical therapy varied by provider, and enrollment was slowed by out-of-study percutaneous closure in patients perceived to be at high risk (though, as discussed above, high risk is difficult to determine).

Intention-to-treat analysis in RESPECT showed no benefit from percutaneous closure, but a favorable outcome was noted with closure in as-treated analysis (HR 0.27; 95% CI 0.1–0.75; P = .007) and per-protocol analysis (HR 0.37; 95% CI 0.14–0.96; P = .03) of the 980 randomized patients.¹⁷ This suggests some benefit, as does the CLOSURE 1 trial, in which 3 of the 12 recurrent strokes in the percutaneous closure group occurred before the device was implanted.¹⁵

The low event rates in these studies prompted several meta-analyses.^29–35 However, only two suggested a benefit of percutaneous closure over medical therapy. In one recent meta-analysis,²⁹ observational study data suggested benefit from percutaneous closure, whereas three randomized controlled trials failed to show a statistically significant benefit.

The conclusions of the meta-analyses must be interpreted with caution because of inherent differences in the randomized controlled trials, including the closure device used, inclusion criteria, study end points, and variations in medical therapy.

Devices differ

A meta-analysis by Khan et al³⁵ showed a benefit of percutaneous closure when evaluating only studies using the Amplatzer PFO occluder (AGA Medical), as in RESPECT and the PC trial.³⁵ As data accumulate, it is important to remember that there are differences between devices. Ongoing trials continue to investigate the Amplatzer device (NCT01550588) and the GORE HELEX Septal Occluder/GORE Septal Occluder (Gore Medical) (NCT00738894).

In another meta-analysis, Pineda et al³¹ found a benefit with closure in the as-treated analysis using data from all three randomized controlled trials (OR 0.62; 95% CI 0.41–0.94; P = .02).³¹ Although paradoxical embolism through the PFO as the mechanism of stroke has been questioned, this finding suggests that actual closure of a PFO may protect against further events, presumably by preventing paradoxical embolism.

Different closure devices have different side effects. The incidence of atrial fibrillation with the CardioSEAL STARFlex device (NMT Medical) is higher than with medical therapy (used in the CLOSURE trial¹⁵), whereas this risk was not statistically significantly increased in the PC trial¹⁶ and RESPECT,¹⁷ which used the Amplatzer device.

Benefit in those with atrial septal aneurysm?

Percutaneous closure has been shown to be safe and effective in patients with PFO and atrial septal aneurysm.³⁶ There was some benefit of closure over medical therapy in a subgroup analysis from RESPECT in these patients, with a HR of 0.19 (95% CI 0.04–0.87, P = .02),¹⁷ although this was not seen in either CLOSURE 1 or the PC trial.

WHAT ARE THE RISKS OF PERCUTANEOUS CLOSURE?

Minor complications of percutaneous closure include bleeding, atrial arrhythmias, device embolization and fracture, and complications related to vascular access. Major complications include hemorrhage requiring transfusion, need for surgery, cardiac tamponade, pulmonary embolism, and death.

The cumulative rate of major complications in 10 observational studies was 1.5%, and the rate of minor complications was 7.9%.³⁷ The RESPECT investigators reported a serious adverse event in 4.2% of patients (ranging in severity from chest tightness to cardiac tamponade).¹⁷

Another possible consequence of percutaneous closure is the need for chronic anticoagulation because of the increased risk of postprocedural atrial fibrillation seen in meta-analyses,^29,31,32 though this may be device-specific.³²

Percutaneous closure was considered successful—ie, to have nearly or completely eliminated shunting of blood through the defect—at 6 months of follow-up in 95.9% of patients in the PC trial, 93.5% in RESPECT, and 86.1% in CLOSURE 1.^15–17

WHAT SHOULD WE BE DOING IN DAILY PRACTICE?

Give aspirin. Aspirin is effective in secondary stroke prevention, and data suggest that patients with PFO and cryptogenic stroke who receive aspirin therapy alone have a similar risk of recurrent events as patients without PFO.

Give warfarin if indicated. Evidence is insufficient to recommend vitamin K antagonist therapy in all patients with PFO and cryptogenic stroke. However, coexisting conditions that warrant anticoagulation must be taken into account.

Individualize. Given the lack of evidence to definitively guide management of patients with cryptogenic stroke and PFO, we need to individualize our approach, taking into account patient preferences, bleeding risk, ability to tolerate procedures, and the likelihood that the PFO is at fault.

No definitive answer on PFO closure. The most recent data suggest that closure may be beneficial, but key questions remain: Who will benefit? And what is the ideal medical therapy? Optimal management will only be established by the continued enrollment of appropriate patients into ongoing clinical trials.

Another question is whether it is possible to perform a randomized controlled trial with enough patients to definitively prove whether percutaneous closure is superior to medical therapy. Recent experience would suggest not.

In the meantime, we have some guidance from the American Heart Association and the American Stroke Association Council on Stroke³⁸ based on the limited evidence available.

Consider patient preference. The physician should present the options to the patient in a balanced manner to enable him or her to make an informed decision. Patients can also be encouraged to seek additional information at websites such as www.stroke.org and www.nlm.nih.gov.

Referral to an interventional cardiologist for evaluation for closure is reasonable in patients with recurrent stroke, medication failure, complicated atrial septal anatomy such as PFO with aneurysm or large shunt, concurrent thromboembolic disease, or contraindications to anticoagulation.

MORE WORK NEEDED

Areas for further study include further identifying the characteristics of patients with PFO and cryptogenic stroke that might indicate who would benefit from percutaneous closure, elucidating the mechanism of stroke in these patients, and determining whether routine stroke evaluation should include echocardiography with a bubble study if there is no change in management based on the finding of PFO.³⁹

Your patient has had an ischemic stroke, and so far you have found no obvious cause such as atrial fibrillation or carotid disease. Should you look for a patent foramen ovale (PFO)? And if you find it, what should you do?

See related commentary

This scenario continues to challenge primary care physicians and subspecialists and requires an understanding of the relationship between PFO and cryptogenic stroke, as well as familiarity with current data on the safety and effectiveness of the management options. PFO is known to be associated with cryptogenic stroke, but many questions remain, including:

• How can we tell if PFO is a culprit (“pathologic”) or an innocent bystander (“incidental”) in a patient who has had a cryptogenic stroke?
• Should stroke patients receive different medical therapy if they have a PFO? In particular, should they receive warfarin in addition to aspirin? And what about the novel oral anticoagulants?
• Which patients should undergo percutaneous closure of the PFO?
• Should we even be looking for PFO in stroke patients at this point, if we cannot say with certainty what we should do if we find it?

WHY IS THIS IMPORTANT?

Cerebrovascular disease is common and costly. The estimated yearly incidence of stroke in the United States is 795,000 events, at a cost of nearly $30 billion.¹ The incidence of stroke in Europe is more than 1 million annually.²

During the diagnostic evaluation of stroke or transient ischemic attack (TIA), PFO is occasionally discovered incidentally by echocardiography. The management decisions that follow often fall to the primary care physician, who must decipher the conflicting data currently available and explain the options to the patient.

Although reviews have been published on this subject,³ several newer key trials and data on risk stratification warrant consideration.

DEFINITIONS

PFO is the failure of the septum primum to fuse with the septum secundum, so that a communication remains between the atria (Figure 1). The diagnosis is commonly made by echocardiography, when agitated saline is injected into the venous system and bubbles can be seen in the left atrium within three to five cardiac cycles (see video).

Atrial septal aneurysm is loosely defined as a septal excursion or bulging of at least 10 to 15 mm into the left and right atria during the cardiac cycle (Figure 2). The combination of PFO and atrial septal aneurysm may be more of a risk factor for stroke than PFO alone (see discussion below).

Cryptogenic stroke. The diagnostic workup of stroke fails to elucidate a clear cause in up to 40% of cases, which are thus called cryptogenic.⁴ The workup varies, but typically includes a search for a cardioembolic source and for atherosclerotic disease. Embolic sources are evaluated for by electrocardiography, transthoracic echocardiography, and possibly imaging of the aortic arch. Evaluation for atherosclerotic disease of the intracranial and extracranial arteries includes magnetic resonance angiography or, if that is unavailable, computed tomographic angiography or carotid Doppler ultrasonography. If no source is found, long-term cardiac monitoring may be used to detect paroxysmal atrial fibrillation, which may be more common than previously thought.

PFO AND CRYPTOGENIC STROKE ARE COMMON

As noted, there are approximately 800,000 strokes every year in the United States. If 25% to 40% of them are cryptogenic (the true prevalence warrants more evaluation),^4,5 then 200,000 to 320,000 strokes are cryptogenic.

Autopsy studies indicate that 25% of the general population have a PFO, and if the prevalence is the same in people with cryptogenic stroke, that would equal 80,000 people with both cryptogenic stroke and PFO every year. However, the prevalence of PFO in patients with cryptogenic stroke appears to be significantly higher than in the general population.⁶ Although these numbers are crude estimates, they provide some insight into the prevalence of this clinical presentation.

HOW ARE CRYPTOGENIC STROKE AND PFO RELATED?

The exact relationship between PFO and cryptogenic stroke is unknown, although cases have been reported of thrombus in transit through a PFO, supporting paradoxical embolism as the plausible cause in stroke patients with PFO.^7–9

There is clear evidence that the two conditions are associated by more than chance. Homma and Sacco⁶ reported that, in several studies, 93 (46%) of 202 patients under age 55 with cryptogenic stroke had PFOs, compared with 29 (11%) of 271 controls (P < .05 in all studies).⁶

In their evaluation of 23 case-control studies, Alsheikh-Ali et al¹⁰ found that the summary odds ratio (OR) for PFO in cryptogenic stroke vs PFO in control patients was 2.9 (95% confidence interval [CI] 2.1–4), largely driven by an OR of 5.1 (3.3–7.8) in those under age 55. Through Bayesian probability theory, this correlated with only a 33% probability that PFO in a patient with cryptogenic stroke was an innocent bystander rather than the culprit.¹⁰

IS PFO A RISK FACTOR FOR STROKE?

One of the more puzzling aspects of the relationship of PFO to cryptogenic stroke is that despite a clear association, there is little evidence that the relationship is causal.

Di Tullio et al¹¹ followed 1,100 people who had no history of stroke and found that the risk of a first stroke in those with a PFO was not significantly higher than in those without a PFO, regardless of age, sex, or ethnic or racial group. At 80 months, the hazard ratio of stroke in people who had a PFO was 1.64 (95% CI 0.87–3.09).¹¹ The findings were similar at 11 years, with a hazard ratio of 1.10 (95% CI 0.64–1.91).¹²

A prospective study of 585 patients found a similar risk of stroke in those with and without a PFO, with a hazard ratio of 1.46 (95% CI 0.74–2.88; P = .28).¹³

These prospective trials suggest that although previous studies have found a higher prevalence of PFOs in patients with cryptogenic stroke than in patients without stroke, there appears to be very little if any increased risk from baseline for a first stroke or TIA.

The lack of statistical significance in these trials should be interpreted with some caution, as a small increased risk is difficult to show if the event rate is low (approximately 10% of patients had events over 11 years in the study by Di Tullio et al¹²).

HOW DO WE KNOW IF A PFO IS A CULPRIT OR BYSTANDER?

Unfortunately, this is largely unanswered, though experts have suggested that echocardiographic features of the PFO, radiographic characteristics of the stroke, and clinical features of the patient may provide useful information.

‘High-risk’ features on echocardiography

Certain features of PFO may portend a high risk of cerebrovascular events. Both right-to-left shunting at rest and septal hypermobility were found in one study¹⁴ to be more common in patients with a PFO who had a stroke or TIA than in patients with a PFO but no cerebrovascular events. Also, patients who had these features and had a stroke had a higher risk of recurrence than stroke patients without these features (12.5% vs 4.3%, P = .05).¹⁴

Septal hypermobility and shunting at rest are easily diagnosed by echocardiography, and detecting these “high-risk” features would be useful if they could identify patients who would benefit from special therapy, such as percutaneous closure of the PFO.

Unfortunately, when investigators looked at these features in subgroup analysis of the major randomized controlled trials of percutaneous closure vs medical therapy, the results were mixed.

CLOSURE 1 (the Evaluation of the STARFlex Septal Closure System in Patients With a Stroke and/or Transient Ischemic Attack Due to Presumed Paradoxical Embolism Through a Patent Foramen Ovale)¹⁵ found percutaneous closure to be no better than medical therapy, regardless of shunt size or the presence of atrial septal aneurysm.

Similarly, the PC trial (Clinical Trial Comparing Percutaneous Closure of Patent Foramen Ovale Using the Amplatzer PFO Occluder With Medical Treatment in Patients With Cryptogenic Embolism)¹⁶ found no statistically significant benefit of closure in those with atrial septal aneurysm.

In contrast, the RESPECT trial (Randomized Evaluation of Recurrent Stroke Comparing PFO Closure to Established Current Standard of Care Treatment)¹⁷ showed percutaneous closure to be beneficial in patients with atrial septal aneurysm or large shunt.

Radiographic characteristics of the stroke

Another area of interest in trying to identify culprit PFOs is the radiographic characteristics of the stroke.

In a study comparing patients with stroke related to atrial fibrillation vs patients with cryptogenic stroke and a known PFO, those in the latter group were more likely to have a single cortical infarction (34.2% vs 3.1%; P < .001) or multiple small scattered lesions (23.1% vs 5.9%; P < .01).¹⁸ Similarly, in a large database of patients with cryptogenic stroke and known PFO status, a superficially located stroke was associated with the presence of PFO (OR 1.54; P < .0001).¹⁹

Although these findings do not tell us with certainty that a patient’s PFO was the cause of his or her stroke, they provide guidance when dealing with the uncertainty of how to manage a patient with PFO. They may be useful in clinical practice, for example, when discussing treatment options with a young patient with cryptogenic stroke who has no risk factors and a superficial single infarct and who is found to have a PFO with a right-to-left shunting at rest.

Patient characteristics

Kent et al²⁰ developed a 10-point index (the RoPE score) in an attempt to assign a probability to whether a stroke was PFO-related. Points were assigned for patients who were younger, who had a cortical stroke on neuroimaging, and who did not have diabetes, hypertension, smoking, or prior stroke or TIA. Patients with cryptogenic stroke with a higher RoPE score were more likely to have a PFO and thus had a higher likelihood that the index event was related to PFO. Of note, the patients with the highest likelihood of PFO-related stroke were the least likely to have a recurrence (RoPE score of 9 to 10; PFO-attributable fraction 88%; estimated 2-year recurrence rate 2%; 95% CI 0%–4%), whereas those with a low RoPE score have more traditional risk factors for stroke and thus are more likely to have a recurrence (RoPE 0 to 3; estimated 2-year recurrence rate 20%; 95% CI 12%–28%).²⁰

Again, this sheds light on a difficulty faced by randomized controlled trials: the patients who may benefit from closure of a PFO may very well be those with the lowest recurrence rates without intervention.

The RoPE index was examined in an attempt to validate previously described morphologic criteria of “high-risk” PFO,²¹ though none of the previously described “high-risk” echocardiographic features (large physiologic size, hypermobile septum, shunt at rest) were more common in the group with presumed PFO-attributable stroke (RoPE score > 6). This underscores the difficulty of distinguishing pathologic PFO from incidental PFO.

KEY TREATMENT CONSIDERATIONS FOR SECONDARY PREVENTION

Given the complicated relationship between PFO and cryptogenic stroke, there has been much debate over management strategies. The three options are surgical closure, percutaneous closure with a device, and medical therapy. The goal of all three is to prevent the recurrence of stroke or TIA.

Surgical closure has largely been supplanted by percutaneous closure, but is still done in specific situations such as when a PFO is found incidentally on transesophageal echocardiography during surgery for another cardiac condition. The data on such cases²² tend to support the argument that asymptomatic PFOs in the general population have a relatively benign natural history.

Thus, the two key questions about management that warrant discussion are: is anticoagulation superior to antiplatelet therapy? And is percutaneous closure superior to medical management?

Anticoagulant vs antiplatelet therapy

Whether to treat with aspirin or with a vitamin K antagonist has been a subject of debate, although there is no strong evidence to suggest that anticoagulation is superior to antiplatelet therapy.

The concern that aspirin alone is insufficient in some patients stems from a study by Mas et al,²³ who followed 581 patients with cryptogenic stroke who had a PFO alone, a PFO with an atrial septal aneurysm, or neither. The rate of stroke recurrence at 4 years on aspirin therapy was 2.3% in those with a PFO alone, 15.2% in those with a PFO with an atrial septal aneurysm, and 4.2% in those with neither.

Many have concluded that aspirin therapy does not sufficiently protect those with both PFO and atrial septal aneurysm, given the high recurrence rate in this group. This might lead to the suggestion that anticoagulation could be of benefit in these patients.

However, the Patent Foramen Ovale in Cryptogenic Stroke Study (PiCSS)²⁴ and the Spanish Multicenter Study Into Right-to-Left Shunt in Cryptogenic Stroke (CODICIA)²⁵ found similar recurrence rates in patients with PFO and atrial septal aneurysm compared with those with only PFO. In these two studies, recurrence rates were similar regardless of whether patients were taking aspirin or warfarin.

In a study that followed 140 consecutive patients with both stroke and PFO, those treated in a nonrandomized fashion with antiplatelet agents had no difference in the recurrence rate compared with those treated with anticoagulation.²⁶

Although uncertainty remains because no head-to-head randomized controlled trial has been done, some patients with PFO have other indications for anticoagulation, most commonly atrial fibrillation and venous thromboembolic disease.

There are currently no data on the use of novel oral anticoagulants in this setting.

Is percutaneous closure better than medical therapy?

When cryptogenic stroke is treated with antiplatelet therapy or anticoagulation therapy, the recurrence rate is the same whether or not the patient has a PFO.^23–25 The belief that medical therapy offers adequate secondary protection is supported by a meta-analysis of 15 studies that found no increased risk of recurrent ischemic events in those with a PFO on medical therapy (antiplatelet or anticoagulant) vs those without a PFO (relative risk 1.1, 95% CI 0.8–1.5).²⁷

Despite the conflicting evidence, percutaneous closure of PFO is still performed, mostly on a case-by-case basis. This has been supported by an apparent benefit in observational studies.

A systematic review of 52 single-arm studies and 7 comparative nonrandomized studies of patients with PFO and cryptogenic stroke found the rate of recurrent stroke to be 0.36 per 100 person-years with percutaneous closure vs 2.53 per 100 person-years with medical therapy.²⁸ However, three long-awaited randomized controlled trials (CLOSURE 1, the PC trial, and RESPECT) failed to show a significant reduction in primary end points with percutaneous closure vs standard medical therapy.^15–17

These trials had several limitations: event rates were low, medical therapy varied by provider, and enrollment was slowed by out-of-study percutaneous closure in patients perceived to be at high risk (though, as discussed above, high risk is difficult to determine).

Intention-to-treat analysis in RESPECT showed no benefit from percutaneous closure, but a favorable outcome was noted with closure in as-treated analysis (HR 0.27; 95% CI 0.1–0.75; P = .007) and per-protocol analysis (HR 0.37; 95% CI 0.14–0.96; P = .03) of the 980 randomized patients.¹⁷ This suggests some benefit, as does the CLOSURE 1 trial, in which 3 of the 12 recurrent strokes in the percutaneous closure group occurred before the device was implanted.¹⁵

The low event rates in these studies prompted several meta-analyses.^29–35 However, only two suggested a benefit of percutaneous closure over medical therapy. In one recent meta-analysis,²⁹ observational study data suggested benefit from percutaneous closure, whereas three randomized controlled trials failed to show a statistically significant benefit.

The conclusions of the meta-analyses must be interpreted with caution because of inherent differences in the randomized controlled trials, including the closure device used, inclusion criteria, study end points, and variations in medical therapy.

Devices differ

A meta-analysis by Khan et al³⁵ showed a benefit of percutaneous closure when evaluating only studies using the Amplatzer PFO occluder (AGA Medical), as in RESPECT and the PC trial.³⁵ As data accumulate, it is important to remember that there are differences between devices. Ongoing trials continue to investigate the Amplatzer device (NCT01550588) and the GORE HELEX Septal Occluder/GORE Septal Occluder (Gore Medical) (NCT00738894).

In another meta-analysis, Pineda et al³¹ found a benefit with closure in the as-treated analysis using data from all three randomized controlled trials (OR 0.62; 95% CI 0.41–0.94; P = .02).³¹ Although paradoxical embolism through the PFO as the mechanism of stroke has been questioned, this finding suggests that actual closure of a PFO may protect against further events, presumably by preventing paradoxical embolism.

Different closure devices have different side effects. The incidence of atrial fibrillation with the CardioSEAL STARFlex device (NMT Medical) is higher than with medical therapy (used in the CLOSURE trial¹⁵), whereas this risk was not statistically significantly increased in the PC trial¹⁶ and RESPECT,¹⁷ which used the Amplatzer device.

Benefit in those with atrial septal aneurysm?

Percutaneous closure has been shown to be safe and effective in patients with PFO and atrial septal aneurysm.³⁶ There was some benefit of closure over medical therapy in a subgroup analysis from RESPECT in these patients, with a HR of 0.19 (95% CI 0.04–0.87, P = .02),¹⁷ although this was not seen in either CLOSURE 1 or the PC trial.

WHAT ARE THE RISKS OF PERCUTANEOUS CLOSURE?

Minor complications of percutaneous closure include bleeding, atrial arrhythmias, device embolization and fracture, and complications related to vascular access. Major complications include hemorrhage requiring transfusion, need for surgery, cardiac tamponade, pulmonary embolism, and death.

The cumulative rate of major complications in 10 observational studies was 1.5%, and the rate of minor complications was 7.9%.³⁷ The RESPECT investigators reported a serious adverse event in 4.2% of patients (ranging in severity from chest tightness to cardiac tamponade).¹⁷

Another possible consequence of percutaneous closure is the need for chronic anticoagulation because of the increased risk of postprocedural atrial fibrillation seen in meta-analyses,^29,31,32 though this may be device-specific.³²

Percutaneous closure was considered successful—ie, to have nearly or completely eliminated shunting of blood through the defect—at 6 months of follow-up in 95.9% of patients in the PC trial, 93.5% in RESPECT, and 86.1% in CLOSURE 1.^15–17

WHAT SHOULD WE BE DOING IN DAILY PRACTICE?

Give aspirin. Aspirin is effective in secondary stroke prevention, and data suggest that patients with PFO and cryptogenic stroke who receive aspirin therapy alone have a similar risk of recurrent events as patients without PFO.

Give warfarin if indicated. Evidence is insufficient to recommend vitamin K antagonist therapy in all patients with PFO and cryptogenic stroke. However, coexisting conditions that warrant anticoagulation must be taken into account.

Individualize. Given the lack of evidence to definitively guide management of patients with cryptogenic stroke and PFO, we need to individualize our approach, taking into account patient preferences, bleeding risk, ability to tolerate procedures, and the likelihood that the PFO is at fault.

No definitive answer on PFO closure. The most recent data suggest that closure may be beneficial, but key questions remain: Who will benefit? And what is the ideal medical therapy? Optimal management will only be established by the continued enrollment of appropriate patients into ongoing clinical trials.

Another question is whether it is possible to perform a randomized controlled trial with enough patients to definitively prove whether percutaneous closure is superior to medical therapy. Recent experience would suggest not.

In the meantime, we have some guidance from the American Heart Association and the American Stroke Association Council on Stroke³⁸ based on the limited evidence available.

Consider patient preference. The physician should present the options to the patient in a balanced manner to enable him or her to make an informed decision. Patients can also be encouraged to seek additional information at websites such as www.stroke.org and www.nlm.nih.gov.

Referral to an interventional cardiologist for evaluation for closure is reasonable in patients with recurrent stroke, medication failure, complicated atrial septal anatomy such as PFO with aneurysm or large shunt, concurrent thromboembolic disease, or contraindications to anticoagulation.

MORE WORK NEEDED

Areas for further study include further identifying the characteristics of patients with PFO and cryptogenic stroke that might indicate who would benefit from percutaneous closure, elucidating the mechanism of stroke in these patients, and determining whether routine stroke evaluation should include echocardiography with a bubble study if there is no change in management based on the finding of PFO.³⁹

Your patient has had an ischemic stroke, and so far you have found no obvious cause such as atrial fibrillation or carotid disease. Should you look for a patent foramen ovale (PFO)? And if you find it, what should you do?

See related commentary

This scenario continues to challenge primary care physicians and subspecialists and requires an understanding of the relationship between PFO and cryptogenic stroke, as well as familiarity with current data on the safety and effectiveness of the management options. PFO is known to be associated with cryptogenic stroke, but many questions remain, including:

• How can we tell if PFO is a culprit (“pathologic”) or an innocent bystander (“incidental”) in a patient who has had a cryptogenic stroke?
• Should stroke patients receive different medical therapy if they have a PFO? In particular, should they receive warfarin in addition to aspirin? And what about the novel oral anticoagulants?
• Which patients should undergo percutaneous closure of the PFO?
• Should we even be looking for PFO in stroke patients at this point, if we cannot say with certainty what we should do if we find it?

WHY IS THIS IMPORTANT?

Cerebrovascular disease is common and costly. The estimated yearly incidence of stroke in the United States is 795,000 events, at a cost of nearly $30 billion.¹ The incidence of stroke in Europe is more than 1 million annually.²

During the diagnostic evaluation of stroke or transient ischemic attack (TIA), PFO is occasionally discovered incidentally by echocardiography. The management decisions that follow often fall to the primary care physician, who must decipher the conflicting data currently available and explain the options to the patient.

Although reviews have been published on this subject,³ several newer key trials and data on risk stratification warrant consideration.

DEFINITIONS

PFO is the failure of the septum primum to fuse with the septum secundum, so that a communication remains between the atria (Figure 1). The diagnosis is commonly made by echocardiography, when agitated saline is injected into the venous system and bubbles can be seen in the left atrium within three to five cardiac cycles (see video).

Atrial septal aneurysm is loosely defined as a septal excursion or bulging of at least 10 to 15 mm into the left and right atria during the cardiac cycle (Figure 2). The combination of PFO and atrial septal aneurysm may be more of a risk factor for stroke than PFO alone (see discussion below).

Cryptogenic stroke. The diagnostic workup of stroke fails to elucidate a clear cause in up to 40% of cases, which are thus called cryptogenic.⁴ The workup varies, but typically includes a search for a cardioembolic source and for atherosclerotic disease. Embolic sources are evaluated for by electrocardiography, transthoracic echocardiography, and possibly imaging of the aortic arch. Evaluation for atherosclerotic disease of the intracranial and extracranial arteries includes magnetic resonance angiography or, if that is unavailable, computed tomographic angiography or carotid Doppler ultrasonography. If no source is found, long-term cardiac monitoring may be used to detect paroxysmal atrial fibrillation, which may be more common than previously thought.

PFO AND CRYPTOGENIC STROKE ARE COMMON

As noted, there are approximately 800,000 strokes every year in the United States. If 25% to 40% of them are cryptogenic (the true prevalence warrants more evaluation),^4,5 then 200,000 to 320,000 strokes are cryptogenic.

Autopsy studies indicate that 25% of the general population have a PFO, and if the prevalence is the same in people with cryptogenic stroke, that would equal 80,000 people with both cryptogenic stroke and PFO every year. However, the prevalence of PFO in patients with cryptogenic stroke appears to be significantly higher than in the general population.⁶ Although these numbers are crude estimates, they provide some insight into the prevalence of this clinical presentation.

HOW ARE CRYPTOGENIC STROKE AND PFO RELATED?

The exact relationship between PFO and cryptogenic stroke is unknown, although cases have been reported of thrombus in transit through a PFO, supporting paradoxical embolism as the plausible cause in stroke patients with PFO.^7–9

There is clear evidence that the two conditions are associated by more than chance. Homma and Sacco⁶ reported that, in several studies, 93 (46%) of 202 patients under age 55 with cryptogenic stroke had PFOs, compared with 29 (11%) of 271 controls (P < .05 in all studies).⁶

In their evaluation of 23 case-control studies, Alsheikh-Ali et al¹⁰ found that the summary odds ratio (OR) for PFO in cryptogenic stroke vs PFO in control patients was 2.9 (95% confidence interval [CI] 2.1–4), largely driven by an OR of 5.1 (3.3–7.8) in those under age 55. Through Bayesian probability theory, this correlated with only a 33% probability that PFO in a patient with cryptogenic stroke was an innocent bystander rather than the culprit.¹⁰

IS PFO A RISK FACTOR FOR STROKE?

One of the more puzzling aspects of the relationship of PFO to cryptogenic stroke is that despite a clear association, there is little evidence that the relationship is causal.

Di Tullio et al¹¹ followed 1,100 people who had no history of stroke and found that the risk of a first stroke in those with a PFO was not significantly higher than in those without a PFO, regardless of age, sex, or ethnic or racial group. At 80 months, the hazard ratio of stroke in people who had a PFO was 1.64 (95% CI 0.87–3.09).¹¹ The findings were similar at 11 years, with a hazard ratio of 1.10 (95% CI 0.64–1.91).¹²

A prospective study of 585 patients found a similar risk of stroke in those with and without a PFO, with a hazard ratio of 1.46 (95% CI 0.74–2.88; P = .28).¹³

These prospective trials suggest that although previous studies have found a higher prevalence of PFOs in patients with cryptogenic stroke than in patients without stroke, there appears to be very little if any increased risk from baseline for a first stroke or TIA.

The lack of statistical significance in these trials should be interpreted with some caution, as a small increased risk is difficult to show if the event rate is low (approximately 10% of patients had events over 11 years in the study by Di Tullio et al¹²).

HOW DO WE KNOW IF A PFO IS A CULPRIT OR BYSTANDER?

Unfortunately, this is largely unanswered, though experts have suggested that echocardiographic features of the PFO, radiographic characteristics of the stroke, and clinical features of the patient may provide useful information.

‘High-risk’ features on echocardiography

Certain features of PFO may portend a high risk of cerebrovascular events. Both right-to-left shunting at rest and septal hypermobility were found in one study¹⁴ to be more common in patients with a PFO who had a stroke or TIA than in patients with a PFO but no cerebrovascular events. Also, patients who had these features and had a stroke had a higher risk of recurrence than stroke patients without these features (12.5% vs 4.3%, P = .05).¹⁴

Septal hypermobility and shunting at rest are easily diagnosed by echocardiography, and detecting these “high-risk” features would be useful if they could identify patients who would benefit from special therapy, such as percutaneous closure of the PFO.

Unfortunately, when investigators looked at these features in subgroup analysis of the major randomized controlled trials of percutaneous closure vs medical therapy, the results were mixed.

CLOSURE 1 (the Evaluation of the STARFlex Septal Closure System in Patients With a Stroke and/or Transient Ischemic Attack Due to Presumed Paradoxical Embolism Through a Patent Foramen Ovale)¹⁵ found percutaneous closure to be no better than medical therapy, regardless of shunt size or the presence of atrial septal aneurysm.

Similarly, the PC trial (Clinical Trial Comparing Percutaneous Closure of Patent Foramen Ovale Using the Amplatzer PFO Occluder With Medical Treatment in Patients With Cryptogenic Embolism)¹⁶ found no statistically significant benefit of closure in those with atrial septal aneurysm.

In contrast, the RESPECT trial (Randomized Evaluation of Recurrent Stroke Comparing PFO Closure to Established Current Standard of Care Treatment)¹⁷ showed percutaneous closure to be beneficial in patients with atrial septal aneurysm or large shunt.

Radiographic characteristics of the stroke

Another area of interest in trying to identify culprit PFOs is the radiographic characteristics of the stroke.

In a study comparing patients with stroke related to atrial fibrillation vs patients with cryptogenic stroke and a known PFO, those in the latter group were more likely to have a single cortical infarction (34.2% vs 3.1%; P < .001) or multiple small scattered lesions (23.1% vs 5.9%; P < .01).¹⁸ Similarly, in a large database of patients with cryptogenic stroke and known PFO status, a superficially located stroke was associated with the presence of PFO (OR 1.54; P < .0001).¹⁹

Although these findings do not tell us with certainty that a patient’s PFO was the cause of his or her stroke, they provide guidance when dealing with the uncertainty of how to manage a patient with PFO. They may be useful in clinical practice, for example, when discussing treatment options with a young patient with cryptogenic stroke who has no risk factors and a superficial single infarct and who is found to have a PFO with a right-to-left shunting at rest.

Patient characteristics

Kent et al²⁰ developed a 10-point index (the RoPE score) in an attempt to assign a probability to whether a stroke was PFO-related. Points were assigned for patients who were younger, who had a cortical stroke on neuroimaging, and who did not have diabetes, hypertension, smoking, or prior stroke or TIA. Patients with cryptogenic stroke with a higher RoPE score were more likely to have a PFO and thus had a higher likelihood that the index event was related to PFO. Of note, the patients with the highest likelihood of PFO-related stroke were the least likely to have a recurrence (RoPE score of 9 to 10; PFO-attributable fraction 88%; estimated 2-year recurrence rate 2%; 95% CI 0%–4%), whereas those with a low RoPE score have more traditional risk factors for stroke and thus are more likely to have a recurrence (RoPE 0 to 3; estimated 2-year recurrence rate 20%; 95% CI 12%–28%).²⁰

Again, this sheds light on a difficulty faced by randomized controlled trials: the patients who may benefit from closure of a PFO may very well be those with the lowest recurrence rates without intervention.

The RoPE index was examined in an attempt to validate previously described morphologic criteria of “high-risk” PFO,²¹ though none of the previously described “high-risk” echocardiographic features (large physiologic size, hypermobile septum, shunt at rest) were more common in the group with presumed PFO-attributable stroke (RoPE score > 6). This underscores the difficulty of distinguishing pathologic PFO from incidental PFO.

KEY TREATMENT CONSIDERATIONS FOR SECONDARY PREVENTION

Given the complicated relationship between PFO and cryptogenic stroke, there has been much debate over management strategies. The three options are surgical closure, percutaneous closure with a device, and medical therapy. The goal of all three is to prevent the recurrence of stroke or TIA.

Surgical closure has largely been supplanted by percutaneous closure, but is still done in specific situations such as when a PFO is found incidentally on transesophageal echocardiography during surgery for another cardiac condition. The data on such cases²² tend to support the argument that asymptomatic PFOs in the general population have a relatively benign natural history.

Thus, the two key questions about management that warrant discussion are: is anticoagulation superior to antiplatelet therapy? And is percutaneous closure superior to medical management?

Anticoagulant vs antiplatelet therapy

Whether to treat with aspirin or with a vitamin K antagonist has been a subject of debate, although there is no strong evidence to suggest that anticoagulation is superior to antiplatelet therapy.

The concern that aspirin alone is insufficient in some patients stems from a study by Mas et al,²³ who followed 581 patients with cryptogenic stroke who had a PFO alone, a PFO with an atrial septal aneurysm, or neither. The rate of stroke recurrence at 4 years on aspirin therapy was 2.3% in those with a PFO alone, 15.2% in those with a PFO with an atrial septal aneurysm, and 4.2% in those with neither.

Many have concluded that aspirin therapy does not sufficiently protect those with both PFO and atrial septal aneurysm, given the high recurrence rate in this group. This might lead to the suggestion that anticoagulation could be of benefit in these patients.

However, the Patent Foramen Ovale in Cryptogenic Stroke Study (PiCSS)²⁴ and the Spanish Multicenter Study Into Right-to-Left Shunt in Cryptogenic Stroke (CODICIA)²⁵ found similar recurrence rates in patients with PFO and atrial septal aneurysm compared with those with only PFO. In these two studies, recurrence rates were similar regardless of whether patients were taking aspirin or warfarin.

In a study that followed 140 consecutive patients with both stroke and PFO, those treated in a nonrandomized fashion with antiplatelet agents had no difference in the recurrence rate compared with those treated with anticoagulation.²⁶

Although uncertainty remains because no head-to-head randomized controlled trial has been done, some patients with PFO have other indications for anticoagulation, most commonly atrial fibrillation and venous thromboembolic disease.

There are currently no data on the use of novel oral anticoagulants in this setting.

Is percutaneous closure better than medical therapy?

When cryptogenic stroke is treated with antiplatelet therapy or anticoagulation therapy, the recurrence rate is the same whether or not the patient has a PFO.^23–25 The belief that medical therapy offers adequate secondary protection is supported by a meta-analysis of 15 studies that found no increased risk of recurrent ischemic events in those with a PFO on medical therapy (antiplatelet or anticoagulant) vs those without a PFO (relative risk 1.1, 95% CI 0.8–1.5).²⁷

Despite the conflicting evidence, percutaneous closure of PFO is still performed, mostly on a case-by-case basis. This has been supported by an apparent benefit in observational studies.

A systematic review of 52 single-arm studies and 7 comparative nonrandomized studies of patients with PFO and cryptogenic stroke found the rate of recurrent stroke to be 0.36 per 100 person-years with percutaneous closure vs 2.53 per 100 person-years with medical therapy.²⁸ However, three long-awaited randomized controlled trials (CLOSURE 1, the PC trial, and RESPECT) failed to show a significant reduction in primary end points with percutaneous closure vs standard medical therapy.^15–17

These trials had several limitations: event rates were low, medical therapy varied by provider, and enrollment was slowed by out-of-study percutaneous closure in patients perceived to be at high risk (though, as discussed above, high risk is difficult to determine).

Intention-to-treat analysis in RESPECT showed no benefit from percutaneous closure, but a favorable outcome was noted with closure in as-treated analysis (HR 0.27; 95% CI 0.1–0.75; P = .007) and per-protocol analysis (HR 0.37; 95% CI 0.14–0.96; P = .03) of the 980 randomized patients.¹⁷ This suggests some benefit, as does the CLOSURE 1 trial, in which 3 of the 12 recurrent strokes in the percutaneous closure group occurred before the device was implanted.¹⁵

The low event rates in these studies prompted several meta-analyses.^29–35 However, only two suggested a benefit of percutaneous closure over medical therapy. In one recent meta-analysis,²⁹ observational study data suggested benefit from percutaneous closure, whereas three randomized controlled trials failed to show a statistically significant benefit.

The conclusions of the meta-analyses must be interpreted with caution because of inherent differences in the randomized controlled trials, including the closure device used, inclusion criteria, study end points, and variations in medical therapy.

Devices differ

A meta-analysis by Khan et al³⁵ showed a benefit of percutaneous closure when evaluating only studies using the Amplatzer PFO occluder (AGA Medical), as in RESPECT and the PC trial.³⁵ As data accumulate, it is important to remember that there are differences between devices. Ongoing trials continue to investigate the Amplatzer device (NCT01550588) and the GORE HELEX Septal Occluder/GORE Septal Occluder (Gore Medical) (NCT00738894).

In another meta-analysis, Pineda et al³¹ found a benefit with closure in the as-treated analysis using data from all three randomized controlled trials (OR 0.62; 95% CI 0.41–0.94; P = .02).³¹ Although paradoxical embolism through the PFO as the mechanism of stroke has been questioned, this finding suggests that actual closure of a PFO may protect against further events, presumably by preventing paradoxical embolism.

Different closure devices have different side effects. The incidence of atrial fibrillation with the CardioSEAL STARFlex device (NMT Medical) is higher than with medical therapy (used in the CLOSURE trial¹⁵), whereas this risk was not statistically significantly increased in the PC trial¹⁶ and RESPECT,¹⁷ which used the Amplatzer device.

Benefit in those with atrial septal aneurysm?

Percutaneous closure has been shown to be safe and effective in patients with PFO and atrial septal aneurysm.³⁶ There was some benefit of closure over medical therapy in a subgroup analysis from RESPECT in these patients, with a HR of 0.19 (95% CI 0.04–0.87, P = .02),¹⁷ although this was not seen in either CLOSURE 1 or the PC trial.

WHAT ARE THE RISKS OF PERCUTANEOUS CLOSURE?

Minor complications of percutaneous closure include bleeding, atrial arrhythmias, device embolization and fracture, and complications related to vascular access. Major complications include hemorrhage requiring transfusion, need for surgery, cardiac tamponade, pulmonary embolism, and death.

The cumulative rate of major complications in 10 observational studies was 1.5%, and the rate of minor complications was 7.9%.³⁷ The RESPECT investigators reported a serious adverse event in 4.2% of patients (ranging in severity from chest tightness to cardiac tamponade).¹⁷

Another possible consequence of percutaneous closure is the need for chronic anticoagulation because of the increased risk of postprocedural atrial fibrillation seen in meta-analyses,^29,31,32 though this may be device-specific.³²

Percutaneous closure was considered successful—ie, to have nearly or completely eliminated shunting of blood through the defect—at 6 months of follow-up in 95.9% of patients in the PC trial, 93.5% in RESPECT, and 86.1% in CLOSURE 1.^15–17

WHAT SHOULD WE BE DOING IN DAILY PRACTICE?

Give aspirin. Aspirin is effective in secondary stroke prevention, and data suggest that patients with PFO and cryptogenic stroke who receive aspirin therapy alone have a similar risk of recurrent events as patients without PFO.

Give warfarin if indicated. Evidence is insufficient to recommend vitamin K antagonist therapy in all patients with PFO and cryptogenic stroke. However, coexisting conditions that warrant anticoagulation must be taken into account.

Individualize. Given the lack of evidence to definitively guide management of patients with cryptogenic stroke and PFO, we need to individualize our approach, taking into account patient preferences, bleeding risk, ability to tolerate procedures, and the likelihood that the PFO is at fault.

No definitive answer on PFO closure. The most recent data suggest that closure may be beneficial, but key questions remain: Who will benefit? And what is the ideal medical therapy? Optimal management will only be established by the continued enrollment of appropriate patients into ongoing clinical trials.

Another question is whether it is possible to perform a randomized controlled trial with enough patients to definitively prove whether percutaneous closure is superior to medical therapy. Recent experience would suggest not.

In the meantime, we have some guidance from the American Heart Association and the American Stroke Association Council on Stroke³⁸ based on the limited evidence available.

Consider patient preference. The physician should present the options to the patient in a balanced manner to enable him or her to make an informed decision. Patients can also be encouraged to seek additional information at websites such as www.stroke.org and www.nlm.nih.gov.

Referral to an interventional cardiologist for evaluation for closure is reasonable in patients with recurrent stroke, medication failure, complicated atrial septal anatomy such as PFO with aneurysm or large shunt, concurrent thromboembolic disease, or contraindications to anticoagulation.

MORE WORK NEEDED

Areas for further study include further identifying the characteristics of patients with PFO and cryptogenic stroke that might indicate who would benefit from percutaneous closure, elucidating the mechanism of stroke in these patients, and determining whether routine stroke evaluation should include echocardiography with a bubble study if there is no change in management based on the finding of PFO.³⁹

The article by Roth and Alli in this issue describes in depth more than 10 years of research that addresses the question, Should we close a patent foramen ovale (PFO) to prevent recurrent cryptogenic stroke?

See related article

There is no longer any doubt that PFO can be the pathway for thrombus from the venous circulation to go from the right atrium to the left atrium, bypassing the pulmonary capillary filtration bed, and entering the arterial side to produce a stroke, myocardial infarction, or peripheral embolus. Two questions remain: What should we do to prevent another episode? And is percutaneous closure of a PFO with the current devices preferable to medical therapy?

How much do we know about the risks and benefits of closure of PFO? I maintain that we know a great deal about interatrial shunt and paradoxical embolism as a cause of cryptogenic stroke. Prospective randomized clinical trials now give us data with which we can provide appropriate direction to our patients. Percutaneous closure is no longer an “experimental procedure,” as insurance companies claim. The experiment has been done, and the only issue is how one interprets the data from the randomized clinical trials.

The review by Roth and Alli comprehensively describes the observational studies, as well as the three randomized clinical trials done to determine whether PFO closure is preferable to medical therapy to prevent recurrent stroke in patients who have already had one cryptogenic stroke. If we understand some of the subtleties and differences between the trials, we can reach an appropriate conclusion as to what to recommend to our patients.

A review of 10 reports of transcatheter closure of PFO vs six reports of medical therapy for cryptogenic stroke showed a range of rates of recurrent stroke at 1 year—between 0% and 4.9% for transcatheter closure, and between 3.8% and 12% for medical therapy.¹

These numbers are important because they were used to estimate the number of patients that would be necessary to study in a randomized clinical trial to demonstrate a benefit of PFO closure vs medical therapy. Unlike most studies of new devices, the PFO closure trials were done in an environment in which patients could get their PFO closed with other devices that were already approved by the US Food and Drug Administration (FDA) for closure of an atrial septal defect. This ability of patients to obtain PFO closure outside of the trial with an off-label device meant that the patients who agreed to be randomized tended to have lower risk for recurrence than patients studied in the observational populations. From a practical standpoint, this meant that the event rate in the patients who participated in the randomized clinical trials (1.7% per year) was lower than predicted from the observational studies.^2,3

Another way of saying this is that the randomized clinical trials were underpowered to answer the question. A common way of dealing with this problem is to combine the results of different studies in a meta-analysis. This makes sense if the studies are assessing the same thing. This is not the case with the PFO closure trials. Although the topic of percutaneous PFO closure vs medical therapy was the same, the devices used were different.

In the CLOSURE trial (Evaluation of the STARFlex Septal Closure System in Patients With a Stroke and/or Transient Ischemic Attack Due to Presumed Paradoxical Embolism Through a Patent Foramen Ovale),³ the device used was the STARFlex, which is no longer produced—and for good reasons. It is not as effective as the Amplatzer or Helex devices in completely closing the right-to-left shunt produced by a PFO. In addition, the CardioSEAL or STARFlex device increases the risk of atrial fibrillation, which was seen in 6% of the treated patients.³ This was the major cause of recurrent stroke in the CLOSURE trial. The CardioSEAL STARFlex device was also more thrombogenic.

In the RESPECT trial (Randomized Evaluation of Recurrent Stroke Comparing PFO Closure to Established Current Standard of Care Treatment),² which used the Amplatzer PFO closure device, there was no increased incidence of atrial fibrillation in the device group compared with the control group. Therefore, it is not appropriate to combine the results of the CLOSURE trial with the results of the RESPECT trial and PC trial,⁴ both of which used the Amplatzer device.

Our patients want to know what the potential risks and benefits will be if they get their PFO closed with a specific device. They don’t want to know the average risk between two different devices.

However, if you do a meta-analysis of the RESPECT and PC trials, which used the same Amplatzer PFO occluder device, and combine the number of patients studied to increase the statistical power, then the benefit of PFO closure is significant even with an intention-to-treat analysis. By combining the two studies that assessed the same device, you reach a completely different interpretation than if you do a meta-analysis including the CLOSURE trial, which showed no benefit.

The medical community should not uncritically accept meta-analysis methodology. It is a marvelous case example of how scientific methods can be inappropriately used and two diametrically opposed conclusions reached if the meta-analysis combines two different types of devices vs a meta-analysis of just the Amplatzer device.

If we combine the numbers from the RESPECT and PC trials, there were 23 strokes in 691 patients (3.3%) in the medical groups and 10 strokes in 703 patients (1.4%) who underwent PFO closure. By chi square analysis of this intention-to-treat protocol, PFO closure provides a statistically significant reduction in preventing recurrent stroke (95% confidence interval 0.20–0.89, P = .02).

From the patient’s perspective, what is important is this: If I get my PFO closed with an Amplatzer PFO occluder device, what are the risks of the procedure, and what are the potential benefits compared with medical therapy? We can now answer that question definitively. I tell my patients, “The risks of the procedure are remarkably low (about 1%) in experienced hands, and the benefit is that your risk of recurrent stroke will be reduced 73%² compared with medical therapy.” In the RESPECT Trial, the as-treated cohort consisted of 958 patients with 21 primary end-point events (5 in the closure group and 16 in the medical-therapy group). The rate of the primary end point was 0.39 events per 100 patient-years in the closure group vs 1.45 events per 100 patient-years in the medical-therapy group (hazard ratio 0.27; 95% confidence interval 0.10–0.75; P = .007).

Not all cryptogenic strokes in people who have a PFO are caused by paradoxical embolism. PFO may be an innocent bystander. In addition, not all people who have a paradoxical embolism will have a recurrent stroke. For example, if a young woman presents with a PFO and stroke, is it possible that she can prevent another stroke just by stopping her birth-control pills and not have her PFO closed? What is the risk of recurrent stroke if she were to become pregnant? We do not know the answers to these questions.

Your patients do not want to wait to find out if they are going to have another stroke. The meta-analysis of the randomized clinical trials for paradoxical embolism demonstrates that the closure devices are safe and effective. The FDA should approve the Amplatzer PFO occluder with an indication to prevent recurrent stroke in patients with PFO and an initial cryptogenic event.

The article by Roth and Alli in this issue describes in depth more than 10 years of research that addresses the question, Should we close a patent foramen ovale (PFO) to prevent recurrent cryptogenic stroke?

See related article

There is no longer any doubt that PFO can be the pathway for thrombus from the venous circulation to go from the right atrium to the left atrium, bypassing the pulmonary capillary filtration bed, and entering the arterial side to produce a stroke, myocardial infarction, or peripheral embolus. Two questions remain: What should we do to prevent another episode? And is percutaneous closure of a PFO with the current devices preferable to medical therapy?

How much do we know about the risks and benefits of closure of PFO? I maintain that we know a great deal about interatrial shunt and paradoxical embolism as a cause of cryptogenic stroke. Prospective randomized clinical trials now give us data with which we can provide appropriate direction to our patients. Percutaneous closure is no longer an “experimental procedure,” as insurance companies claim. The experiment has been done, and the only issue is how one interprets the data from the randomized clinical trials.

The review by Roth and Alli comprehensively describes the observational studies, as well as the three randomized clinical trials done to determine whether PFO closure is preferable to medical therapy to prevent recurrent stroke in patients who have already had one cryptogenic stroke. If we understand some of the subtleties and differences between the trials, we can reach an appropriate conclusion as to what to recommend to our patients.

A review of 10 reports of transcatheter closure of PFO vs six reports of medical therapy for cryptogenic stroke showed a range of rates of recurrent stroke at 1 year—between 0% and 4.9% for transcatheter closure, and between 3.8% and 12% for medical therapy.¹

These numbers are important because they were used to estimate the number of patients that would be necessary to study in a randomized clinical trial to demonstrate a benefit of PFO closure vs medical therapy. Unlike most studies of new devices, the PFO closure trials were done in an environment in which patients could get their PFO closed with other devices that were already approved by the US Food and Drug Administration (FDA) for closure of an atrial septal defect. This ability of patients to obtain PFO closure outside of the trial with an off-label device meant that the patients who agreed to be randomized tended to have lower risk for recurrence than patients studied in the observational populations. From a practical standpoint, this meant that the event rate in the patients who participated in the randomized clinical trials (1.7% per year) was lower than predicted from the observational studies.^2,3

Another way of saying this is that the randomized clinical trials were underpowered to answer the question. A common way of dealing with this problem is to combine the results of different studies in a meta-analysis. This makes sense if the studies are assessing the same thing. This is not the case with the PFO closure trials. Although the topic of percutaneous PFO closure vs medical therapy was the same, the devices used were different.

In the CLOSURE trial (Evaluation of the STARFlex Septal Closure System in Patients With a Stroke and/or Transient Ischemic Attack Due to Presumed Paradoxical Embolism Through a Patent Foramen Ovale),³ the device used was the STARFlex, which is no longer produced—and for good reasons. It is not as effective as the Amplatzer or Helex devices in completely closing the right-to-left shunt produced by a PFO. In addition, the CardioSEAL or STARFlex device increases the risk of atrial fibrillation, which was seen in 6% of the treated patients.³ This was the major cause of recurrent stroke in the CLOSURE trial. The CardioSEAL STARFlex device was also more thrombogenic.

In the RESPECT trial (Randomized Evaluation of Recurrent Stroke Comparing PFO Closure to Established Current Standard of Care Treatment),² which used the Amplatzer PFO closure device, there was no increased incidence of atrial fibrillation in the device group compared with the control group. Therefore, it is not appropriate to combine the results of the CLOSURE trial with the results of the RESPECT trial and PC trial,⁴ both of which used the Amplatzer device.

Our patients want to know what the potential risks and benefits will be if they get their PFO closed with a specific device. They don’t want to know the average risk between two different devices.

However, if you do a meta-analysis of the RESPECT and PC trials, which used the same Amplatzer PFO occluder device, and combine the number of patients studied to increase the statistical power, then the benefit of PFO closure is significant even with an intention-to-treat analysis. By combining the two studies that assessed the same device, you reach a completely different interpretation than if you do a meta-analysis including the CLOSURE trial, which showed no benefit.

The medical community should not uncritically accept meta-analysis methodology. It is a marvelous case example of how scientific methods can be inappropriately used and two diametrically opposed conclusions reached if the meta-analysis combines two different types of devices vs a meta-analysis of just the Amplatzer device.

If we combine the numbers from the RESPECT and PC trials, there were 23 strokes in 691 patients (3.3%) in the medical groups and 10 strokes in 703 patients (1.4%) who underwent PFO closure. By chi square analysis of this intention-to-treat protocol, PFO closure provides a statistically significant reduction in preventing recurrent stroke (95% confidence interval 0.20–0.89, P = .02).

From the patient’s perspective, what is important is this: If I get my PFO closed with an Amplatzer PFO occluder device, what are the risks of the procedure, and what are the potential benefits compared with medical therapy? We can now answer that question definitively. I tell my patients, “The risks of the procedure are remarkably low (about 1%) in experienced hands, and the benefit is that your risk of recurrent stroke will be reduced 73%² compared with medical therapy.” In the RESPECT Trial, the as-treated cohort consisted of 958 patients with 21 primary end-point events (5 in the closure group and 16 in the medical-therapy group). The rate of the primary end point was 0.39 events per 100 patient-years in the closure group vs 1.45 events per 100 patient-years in the medical-therapy group (hazard ratio 0.27; 95% confidence interval 0.10–0.75; P = .007).

Not all cryptogenic strokes in people who have a PFO are caused by paradoxical embolism. PFO may be an innocent bystander. In addition, not all people who have a paradoxical embolism will have a recurrent stroke. For example, if a young woman presents with a PFO and stroke, is it possible that she can prevent another stroke just by stopping her birth-control pills and not have her PFO closed? What is the risk of recurrent stroke if she were to become pregnant? We do not know the answers to these questions.

Your patients do not want to wait to find out if they are going to have another stroke. The meta-analysis of the randomized clinical trials for paradoxical embolism demonstrates that the closure devices are safe and effective. The FDA should approve the Amplatzer PFO occluder with an indication to prevent recurrent stroke in patients with PFO and an initial cryptogenic event.

The article by Roth and Alli in this issue describes in depth more than 10 years of research that addresses the question, Should we close a patent foramen ovale (PFO) to prevent recurrent cryptogenic stroke?

See related article

There is no longer any doubt that PFO can be the pathway for thrombus from the venous circulation to go from the right atrium to the left atrium, bypassing the pulmonary capillary filtration bed, and entering the arterial side to produce a stroke, myocardial infarction, or peripheral embolus. Two questions remain: What should we do to prevent another episode? And is percutaneous closure of a PFO with the current devices preferable to medical therapy?

How much do we know about the risks and benefits of closure of PFO? I maintain that we know a great deal about interatrial shunt and paradoxical embolism as a cause of cryptogenic stroke. Prospective randomized clinical trials now give us data with which we can provide appropriate direction to our patients. Percutaneous closure is no longer an “experimental procedure,” as insurance companies claim. The experiment has been done, and the only issue is how one interprets the data from the randomized clinical trials.

The review by Roth and Alli comprehensively describes the observational studies, as well as the three randomized clinical trials done to determine whether PFO closure is preferable to medical therapy to prevent recurrent stroke in patients who have already had one cryptogenic stroke. If we understand some of the subtleties and differences between the trials, we can reach an appropriate conclusion as to what to recommend to our patients.

A review of 10 reports of transcatheter closure of PFO vs six reports of medical therapy for cryptogenic stroke showed a range of rates of recurrent stroke at 1 year—between 0% and 4.9% for transcatheter closure, and between 3.8% and 12% for medical therapy.¹

These numbers are important because they were used to estimate the number of patients that would be necessary to study in a randomized clinical trial to demonstrate a benefit of PFO closure vs medical therapy. Unlike most studies of new devices, the PFO closure trials were done in an environment in which patients could get their PFO closed with other devices that were already approved by the US Food and Drug Administration (FDA) for closure of an atrial septal defect. This ability of patients to obtain PFO closure outside of the trial with an off-label device meant that the patients who agreed to be randomized tended to have lower risk for recurrence than patients studied in the observational populations. From a practical standpoint, this meant that the event rate in the patients who participated in the randomized clinical trials (1.7% per year) was lower than predicted from the observational studies.^2,3

Another way of saying this is that the randomized clinical trials were underpowered to answer the question. A common way of dealing with this problem is to combine the results of different studies in a meta-analysis. This makes sense if the studies are assessing the same thing. This is not the case with the PFO closure trials. Although the topic of percutaneous PFO closure vs medical therapy was the same, the devices used were different.

In the CLOSURE trial (Evaluation of the STARFlex Septal Closure System in Patients With a Stroke and/or Transient Ischemic Attack Due to Presumed Paradoxical Embolism Through a Patent Foramen Ovale),³ the device used was the STARFlex, which is no longer produced—and for good reasons. It is not as effective as the Amplatzer or Helex devices in completely closing the right-to-left shunt produced by a PFO. In addition, the CardioSEAL or STARFlex device increases the risk of atrial fibrillation, which was seen in 6% of the treated patients.³ This was the major cause of recurrent stroke in the CLOSURE trial. The CardioSEAL STARFlex device was also more thrombogenic.

In the RESPECT trial (Randomized Evaluation of Recurrent Stroke Comparing PFO Closure to Established Current Standard of Care Treatment),² which used the Amplatzer PFO closure device, there was no increased incidence of atrial fibrillation in the device group compared with the control group. Therefore, it is not appropriate to combine the results of the CLOSURE trial with the results of the RESPECT trial and PC trial,⁴ both of which used the Amplatzer device.

Our patients want to know what the potential risks and benefits will be if they get their PFO closed with a specific device. They don’t want to know the average risk between two different devices.

However, if you do a meta-analysis of the RESPECT and PC trials, which used the same Amplatzer PFO occluder device, and combine the number of patients studied to increase the statistical power, then the benefit of PFO closure is significant even with an intention-to-treat analysis. By combining the two studies that assessed the same device, you reach a completely different interpretation than if you do a meta-analysis including the CLOSURE trial, which showed no benefit.

The medical community should not uncritically accept meta-analysis methodology. It is a marvelous case example of how scientific methods can be inappropriately used and two diametrically opposed conclusions reached if the meta-analysis combines two different types of devices vs a meta-analysis of just the Amplatzer device.

If we combine the numbers from the RESPECT and PC trials, there were 23 strokes in 691 patients (3.3%) in the medical groups and 10 strokes in 703 patients (1.4%) who underwent PFO closure. By chi square analysis of this intention-to-treat protocol, PFO closure provides a statistically significant reduction in preventing recurrent stroke (95% confidence interval 0.20–0.89, P = .02).

From the patient’s perspective, what is important is this: If I get my PFO closed with an Amplatzer PFO occluder device, what are the risks of the procedure, and what are the potential benefits compared with medical therapy? We can now answer that question definitively. I tell my patients, “The risks of the procedure are remarkably low (about 1%) in experienced hands, and the benefit is that your risk of recurrent stroke will be reduced 73%² compared with medical therapy.” In the RESPECT Trial, the as-treated cohort consisted of 958 patients with 21 primary end-point events (5 in the closure group and 16 in the medical-therapy group). The rate of the primary end point was 0.39 events per 100 patient-years in the closure group vs 1.45 events per 100 patient-years in the medical-therapy group (hazard ratio 0.27; 95% confidence interval 0.10–0.75; P = .007).

Not all cryptogenic strokes in people who have a PFO are caused by paradoxical embolism. PFO may be an innocent bystander. In addition, not all people who have a paradoxical embolism will have a recurrent stroke. For example, if a young woman presents with a PFO and stroke, is it possible that she can prevent another stroke just by stopping her birth-control pills and not have her PFO closed? What is the risk of recurrent stroke if she were to become pregnant? We do not know the answers to these questions.

Your patients do not want to wait to find out if they are going to have another stroke. The meta-analysis of the randomized clinical trials for paradoxical embolism demonstrates that the closure devices are safe and effective. The FDA should approve the Amplatzer PFO occluder with an indication to prevent recurrent stroke in patients with PFO and an initial cryptogenic event.

A 48-year-old man with gout, multiple sclerosis, and previously treated methicillin-resistant Staphylococcus aureus (MRSA) infection presented to the emergency room with pain and significant swelling at the site of a dog bite on his left forearm. He had been bitten 2 weeks earlier by a friend’s dog, and the bite had punctured the skin. He also had red streaking on the skin of the left arm from the wrist to the elbow, and he reported feeling “feverish” and having night sweats.

At first, the bite had seemed to improve, then swelling and pain had developed and increased. He reported this to his primary care physician, along with the information that he had previously had an anaphylactic reaction to penicillin and a cephalosporin. His physician, considering a penicillin allergy, started him on ciprofloxacin (Cipro) plus clindamycin (Cleocin). The patient took this for 5 days, but without improvement. The appearance of the red streaking on his left forearm prompted his presentation to our emergency room.

ORGANISMS IN DOG BITES

1. Which is the most common cause of infected dog bite?

• Pasteurella canis
• Streptococci and S aureus
• Erysipelothrix rhusiopathiae
• Capnocytophaga canimorsus
• Eikenella corrodens

Streptococci (50%) and S aureus (20% to 40%) are the organisms most commonly responsible for infected dog bites, as they are for other skin and soft-tissue infections.¹P canis is unique to dog bite infections but accounts for only 18%.²E rhusiopathiae is an unusual isolate from cat and dog bites and is more commonly isolated from the mouths of fish and aquatic mammals. C canimorsus is a normal inhabitant of the oral cavity of dogs and cats but an unusual cause of wound infection from a dog bite. It is notable for sepsis and central nervous system infections uniquely associated with veterinarians, dog owners, kennel workers, and mail carriers.³E corrodens infection is more common with human bites.⁴

THE EVALUATION BEGINS

On examination, the patient had marked edema of the left forearm and pain in the joints of the left hand. His temperature was 100.2°F (37.9°C). Because of the duration and severity of symptoms, the examining physician was concerned about septic arthritis of the wrist, and the patient was admitted to the hospital.

In the hospital, our patient was thermodynamically stable without documented fever or chills. There was no open wound to culture, and blood cultures were negative. Marked edema and joint involvement raised suspicion of erysipeloid. This “cousin” of erysipelas often involves the underlying joint, is associated with edema, and produces systemic manifestations of fever and arthralgia.

Radiography of the left forearm and hand demonstrated multiple foci of demineralization within the carpal bones and proximal radius, attributed to disuse. Magnetic resonance imaging (MRI) the next day showed multiple bone infarcts in the carpal bones and the distal radius, with synovitis and fluid in the carpal joints and without adjacent osteomyelitis. Fluid was also seen in the soft tissues in the ulnar aspect of the left wrist, and tenosynovitis involving the flexor carpi radialis tendon was noted.

Arthrocentesis of his left radiocarpal joint produced synovial fluid negative for crystals and negative on Gram stain; the fluid was also sent for culture. The patient’s tetanus immunization was current, and the dog was known to have been immunized against rabies.

ANTIBIOTICS FOR INFECTED DOG BITES

2. Which antibiotic regimen would you choose for this patient?

• Oral amoxicillin and clavulanate
• Meropenem
• Vancomycin, clindamycin, aztreonam
• Clindamycin and levofloxacin
• Clindamycin and trimethoprim-sulfamethoxazole

Oral amoxicillin and clavulanate (Augmentin) is a judicious choice for prophylactic treatment of deep bites in the early stages of infection. However, our patient’s wound was no longer in the early stages of infection, and he had a history of an adverse reaction to penicillin.

Meropenem (Merrem IV) cross-reacts minimally with penicillin allergy and is reported to be safe in patients with a history of anaphylactic reactions to penicillin,⁵ but overuse of carbapenems has led to the development of carbapenem-resistant strains of Klebsiella, Stenotrophomonas, and Acinetobacter organisms.

Given the rise of MRSA infections and the common involvement of staphylococci, streptococci, and anaerobic bacteria in complicated dog bites, the combination of vancomycin and clindamycin is a good choice, and aztreonam (Azactam) would add empiric coverage of gram-negative enteric organisms.

Levofloxacin (Levaquin) also covers gramnegative enteric organisms, but Fusobacterium canifelinum, a common anaerobe in the oral flora of dogs and cats, is intrinsically resistant to fluoroquinolones.

Clindamycin and levofloxacin would be a good step-down oral regimen. Pasteurella multocida has variable sensitivity to the commonly used agents dicloxacillin (Dynapen), cephalexin (Keflex), macrolides, and clindamycin, but it is a less likely pathogen at this late stage and could be covered with levofloxacin alone.

C canimorsus is resistant to trimethoprim-sulfamethoxazole (Bactrim) and cephalexin, but is well covered by clindamycin.⁶

CASE CONTINUED

Our patient was started on intravenous vancomycin, clindamycin, and aztreonam for coverage of dog-mouth flora. Blood cultures and cultures of synovial fluid of the left wrist were negative. Vancomycin was discontinued after 48 hours when blood cultures did not grow staphylococcal organisms, but clindamycin and aztreonam were continued for a total of 8 days to treat possible infection with anaerobic and gram-negative enteric pathogens.

To test for autonomic dysfunction, a plastic pen case drawn lightly across each forearm revealed a loss of tactile adherence (ie, areas where moist, sweaty skin impeded the movement of the pen case) on the affected forearm, a sign of underlying nerve injury. The affected forearm was sensitive to light touch, with pain out of proportion to the stimulus.

ARRIVING AT THE DIAGNOSIS

Based on the wide distribution of inflammation, autonomic dysfunction (shown by differences in temperature and sweating), radiographic evidence of demineralization, hyperesthesia, and lack of improvement in pain and swelling after two courses of antibiotics, the patient’s clinical course was determined to be consistent with complex regional pain syndrome type 1, previously referred to as reflex sympathetic dystrophy.

Symptoms of complex regional pain syndrome traditionally include pain, regional edema, joint stiffness, muscular atrophy, vasomotor disturbances (causing temperature variability and erythema), regional diaphoresis, and localized skeletal demineralization on radiography.

Complex regional pain syndrome type 1 occurs as regional pain and inflammation as an excessive sympathetic reaction to an often minor insult, without nerve injury. When the syndrome occurs in a patient with obvious partial nerve injury, it is categorized as type 2 (formerly known as causalgia). The two types are clinically indistinguishable and are not uncommon. About 10% of all patients with complex regional pain syndrome have obvious nerve injury (complex regional pain syndrome type 2). In a study of 109 patients with Colles fracture, 25% developed symptoms of complex regional pain syndrome.⁷

Complex regional pain syndrome is difficult to diagnose, as it resembles many other ailments, such as gout, infection, bone tumor, stress fracture, and arthritis. Its pathophysiology is poorly understood, but it is believed to result from a “short circuit” in the reflex arc between somatic afferent sensory fibers and autonomic sympathetic efferent fibers, and this is thought to explain the increased sympathetic stimulation.

Although the pathophysiology is likely the same in type 1 and type 2, electromyography with a nerve conduction study is a reliable way to detect nerve damage and thus distinguish between the two types of complex regional pain syndrome.⁸

Our understanding of this syndrome is evolving. A recent study using sensory testing showed that 33% of patients with type 1 had combinations of increased and decreased thresholds for the detection of thermal, vibratory, and mechanical stimuli in the distribution of discrete peripheral nerves, suggesting that the patients actually had type 2.⁹

CONFIRMING COMPLEX REGIONAL PAIN SYNDROME TYPE 1

3. Which of the following is the best way to confirm complex regional pain syndrome type 1?

• Erythrocyte sedimentation rate, C-reactive protein, and complete blood cell count
• Plain radiography of the hand and forearm
• Three-phase technetium bone scan
• The Budapest diagnostic criteria
• MRI
• Autonomic testing

Complex regional pain syndrome type 1 is a clinical diagnosis. Diagnostic studies lack sensitivity and specificity but may confirm complex regional pain syndrome type 1 or rule out other diagnoses. The Budapest diagnostic criteria¹⁰ (Table 1) may be the best way to confirm this diagnosis. The criteria are as follows: continuing pain disproportionate to an inciting event, coupled with three of four symptoms plus at least one sign from sensory, vasomotor, sudomotor, and motor-trophic categories.

Laboratory tests are not helpful because acute-phase reactants and blood counts remain normal in these patients.

Plain radiography is not sensitive in early diagnosis, but at 2 weeks it may show patchy areas of osteopenia in adjacent bones throughout the region, as well as subsequent diffuse demineralization.

Three-phase bone scanning is more sensitive than plain radiography, with 75% of patients showing regional disparities in blood flow in early sequences and increased bone uptake in the later sequences.

MRI is a sensitive early test, as it better defines focal areas of bone loss and increased T2 bone signal in adjacent bone, as well as early soft-tissue changes. Computed tomography does not show early specific changes in muscle, tendon, or bone and so is not recommended.

THE EVALUATION CONTINUES

The admitting diagnosis was septic arthritis, and our patient underwent computed tomography, which showed focal demineralization that could have represented bone infarcts or infection, confounding the diagnosis of complex regional pain syndrome.

Autonomic nerve testing can help distinguish complex regional pain syndrome from other disorders. Complex regional pain syndrome is characterized by increased sympathetic activity and results in increased sweat output. Autonomic testing includes resting sweat output, resting skin temperature, and quantitative sudomotor axon reflex testing. In one study, an increase in resting sweat output used in conjunction with quantitative sudomotor axon reflex testing predicted the diagnosis of complex regional pain syndrome with a specificity of 98%.¹¹ However, autonomic testing is limited to academic centers and is not readily available.

TREATING COMPLEX REGIONAL PAIN SYNDROME TYPE 1

4. Which is the best first-line therapy for complex regional pain syndrome type 1?

• Stellate ganglion nerve block
• Occupational therapy to splint the wrist and forearm
• Oral corticosteroids
• Physical therapy to prevent loss of joint motion
• Tricyclic antidepressant drugs (eg, amitriptyline), pregabalin, and bisphosphonates

Physical therapy should be started early in all patients, with range-of-motion exercises to prevent contracture and enhance mobility.

Stellate ganglion nerve block has been used to counter severe sympathetic hyperactivity, but it also may aggravate symptoms of complex regional pain syndrome and so remains a controversial treatment.

Immobilization and splinting should be avoided, as this will augment edema, pain, and contracture of joints.

Corticosteroids do not shorten the course or assuage symptoms and may increase edema.

Amitriptyline (Elavil) and pregabalin (Lyrica) have been used successfully to counter extended courses of allodynia and hyperalgesia. Bisphosphonates may decrease bone loss and pain and may be needed should the course be complicated by myositis ossificans, a form of dystrophic bone formation in juxtaposed tendon and muscle related to neuroactivation of fibroblasts and osteoblasts.

THE COURSE OF COMPLEX REGIONAL PAIN SYNDROME

Traditionally, type 1 was divided into three stages—an early inflammatory stage, a dystrophic stage, and a late atrophic stage.¹² Although there is no evidence to support a consistent three-stage evolution, the severity of symptoms may help determine the best approach to management.¹³

Patients initially exhibit burning or throbbing pain, diffuse aching, sensitivity to touch or cold (allodynia), localized edema, and vasomotor disturbances of variable intensity that may produce altered color and temperature. Topical capsaicin cream; a tricyclic antidepressant; an anticonvulsant such as gabapentin (Neurontin), pregabalin, or lamotrigine (Lamictal); or a nonsteroidal anti-inflammatory drug should be tried first. Some of these treatments are poorly tolerated in elderly patients. If pain persists, nasal calcitonin may be added. Trigger-point injections with an anesthetic or glucocorticoid may be tried.

The management of early complex regional pain syndrome is sometimes supplemented with systemic corticosteroids, but reviews of randomized controlled trials have failed to show efficacy.¹⁴

Later in the course, patients may suffer persistent soft-tissue edema, accompanied by thickening of the skin and periarticular soft tissues, muscle wasting, and the skin changes of brawny edema. Regional blockade of sympathetic ganglions, epidural administration of clonidine, implantable peripheral nerve stimulators, and spinal cord stimulators have all been applied by experts in pain management and may provide benefit. Progression of the syndrome may include cyanosis, mottling, increased sweating, abnormal hair growth, and diffuse swelling in nonarticular tissue.

It is always acceptable to refer to an experienced pain management specialist, and a multidisciplinary approach is recommended at the outset.¹²

OUR PATIENT’S CARE CONTINUED

Our patient’s forearm and wrist were placed in a sling to keep his left arm elevated when active. This helped control sympathetic vascular edema and throbbing pain. Physical therapy with range-of-motion exercises prevented contracture.

He was discharged home on limited oxycodone as needed, with close follow-up by his primary care physician to monitor his pain symptoms. The pain and swelling slowly improved over the next 2 months, but he suffered a fall, twisting his left wrist. This minor injury was followed by more intense pain and swelling of the forearm, hand, and wrist.

COMORBIDITIES

5. Which of the following statements about conditions associated with complex regional pain syndrome most likely applies to our patient?

• Gout is likely following minor trauma
• Minor trauma or surgical bone biopsy may reactivate complex regional pain syndrome
• Septic hip arthritis due to MRSA may have reemerged and seeded the wrist
• Patients with multiple sclerosis have a propensity for complex regional pain syndrome
• Complex regional pain syndrome type 1 begets type 2

Gout does follow minor injury, but our patient’s uric acid was well controlled on allopurinol (Zyloprim), and gout is unlikely to be causing polyarticular swelling of the hand, wrist, and forearm.

Minor trauma, sometimes inconsequential enough to have been completely forgotten, may either initiate complex regional pain syndrome or, as seen here, reactivate it. Bone changes seen on MRI sometimes trigger surgical bone biopsy, only to reactivate the dysesthesia and sympathetic vascular reaction. Surgery should be avoided. Trauma and surgery are causative rather than associative comorbidities.

Sepsis due to MRSA after total hip arthroplasty may be reactivated, especially in the setting of immunosuppressive treatment. But the diffuse bone changes seen in multiple carpal, radial, and ulnar bones suggest generalized vascular and sympathetic disarray, most consistent with complex regional pain syndrome type 1.

AN ASSOCIATION WITH MULTIPLE SCLEROSIS?

Multiple sclerosis and other central neuropathic conditions such as stroke are associated with complex regional pain syndrome type 1.^15,16

A hypothetical cause for the higher prevalence of complex regional pain syndrome in patients with multiple sclerosis may be demyelination resulting in aberrant signaling and overreaction to distal pain receptors. Demyelination of neurons within the autonomic or spinothalamic tracts potentially increases susceptibility to development of the pain syndrome.

Our patient had an apparent stimulus for the development of the syndrome, ie, the initial dog bite, and the wrist injury later may have caused peripheral nerve injury. Such injury may lead to release of vasodilatory neuropeptides including substance P from stimulated cutaneous nerves with cell bodies in the dorsal root ganglia. Excessive vasodilation and increased vascular permeability result in the affected limb becoming edematous and causing cutaneous nerves to be further activated. Stimulated cutaneous neurons normally have an inhibitory influence on sympathetic activity at the level of entry of the dorsal root ganglia in the cord. In complex regional pain syndrome, this inhibition is lost, resulting in a hyperactive somatosympathetic reflex.¹⁷ Underlying multiple sclerosis may have contributed to the loss of inhibition by the cutaneous nerves on the sympathetic system.

CASE CONCLUDED

We continued to closely follow this patient, who was on a self-directed program of physical therapy. One year after the original dog bite, the complex regional pain syndrome had completely resolved.

A 48-year-old man with gout, multiple sclerosis, and previously treated methicillin-resistant Staphylococcus aureus (MRSA) infection presented to the emergency room with pain and significant swelling at the site of a dog bite on his left forearm. He had been bitten 2 weeks earlier by a friend’s dog, and the bite had punctured the skin. He also had red streaking on the skin of the left arm from the wrist to the elbow, and he reported feeling “feverish” and having night sweats.

At first, the bite had seemed to improve, then swelling and pain had developed and increased. He reported this to his primary care physician, along with the information that he had previously had an anaphylactic reaction to penicillin and a cephalosporin. His physician, considering a penicillin allergy, started him on ciprofloxacin (Cipro) plus clindamycin (Cleocin). The patient took this for 5 days, but without improvement. The appearance of the red streaking on his left forearm prompted his presentation to our emergency room.

ORGANISMS IN DOG BITES

1. Which is the most common cause of infected dog bite?

• Pasteurella canis
• Streptococci and S aureus
• Erysipelothrix rhusiopathiae
• Capnocytophaga canimorsus
• Eikenella corrodens

Streptococci (50%) and S aureus (20% to 40%) are the organisms most commonly responsible for infected dog bites, as they are for other skin and soft-tissue infections.¹P canis is unique to dog bite infections but accounts for only 18%.²E rhusiopathiae is an unusual isolate from cat and dog bites and is more commonly isolated from the mouths of fish and aquatic mammals. C canimorsus is a normal inhabitant of the oral cavity of dogs and cats but an unusual cause of wound infection from a dog bite. It is notable for sepsis and central nervous system infections uniquely associated with veterinarians, dog owners, kennel workers, and mail carriers.³E corrodens infection is more common with human bites.⁴

THE EVALUATION BEGINS

On examination, the patient had marked edema of the left forearm and pain in the joints of the left hand. His temperature was 100.2°F (37.9°C). Because of the duration and severity of symptoms, the examining physician was concerned about septic arthritis of the wrist, and the patient was admitted to the hospital.

In the hospital, our patient was thermodynamically stable without documented fever or chills. There was no open wound to culture, and blood cultures were negative. Marked edema and joint involvement raised suspicion of erysipeloid. This “cousin” of erysipelas often involves the underlying joint, is associated with edema, and produces systemic manifestations of fever and arthralgia.

Radiography of the left forearm and hand demonstrated multiple foci of demineralization within the carpal bones and proximal radius, attributed to disuse. Magnetic resonance imaging (MRI) the next day showed multiple bone infarcts in the carpal bones and the distal radius, with synovitis and fluid in the carpal joints and without adjacent osteomyelitis. Fluid was also seen in the soft tissues in the ulnar aspect of the left wrist, and tenosynovitis involving the flexor carpi radialis tendon was noted.

Arthrocentesis of his left radiocarpal joint produced synovial fluid negative for crystals and negative on Gram stain; the fluid was also sent for culture. The patient’s tetanus immunization was current, and the dog was known to have been immunized against rabies.

ANTIBIOTICS FOR INFECTED DOG BITES

2. Which antibiotic regimen would you choose for this patient?

• Oral amoxicillin and clavulanate
• Meropenem
• Vancomycin, clindamycin, aztreonam
• Clindamycin and levofloxacin
• Clindamycin and trimethoprim-sulfamethoxazole

Oral amoxicillin and clavulanate (Augmentin) is a judicious choice for prophylactic treatment of deep bites in the early stages of infection. However, our patient’s wound was no longer in the early stages of infection, and he had a history of an adverse reaction to penicillin.

Meropenem (Merrem IV) cross-reacts minimally with penicillin allergy and is reported to be safe in patients with a history of anaphylactic reactions to penicillin,⁵ but overuse of carbapenems has led to the development of carbapenem-resistant strains of Klebsiella, Stenotrophomonas, and Acinetobacter organisms.

Given the rise of MRSA infections and the common involvement of staphylococci, streptococci, and anaerobic bacteria in complicated dog bites, the combination of vancomycin and clindamycin is a good choice, and aztreonam (Azactam) would add empiric coverage of gram-negative enteric organisms.

Levofloxacin (Levaquin) also covers gramnegative enteric organisms, but Fusobacterium canifelinum, a common anaerobe in the oral flora of dogs and cats, is intrinsically resistant to fluoroquinolones.

Clindamycin and levofloxacin would be a good step-down oral regimen. Pasteurella multocida has variable sensitivity to the commonly used agents dicloxacillin (Dynapen), cephalexin (Keflex), macrolides, and clindamycin, but it is a less likely pathogen at this late stage and could be covered with levofloxacin alone.

C canimorsus is resistant to trimethoprim-sulfamethoxazole (Bactrim) and cephalexin, but is well covered by clindamycin.⁶

CASE CONTINUED

Our patient was started on intravenous vancomycin, clindamycin, and aztreonam for coverage of dog-mouth flora. Blood cultures and cultures of synovial fluid of the left wrist were negative. Vancomycin was discontinued after 48 hours when blood cultures did not grow staphylococcal organisms, but clindamycin and aztreonam were continued for a total of 8 days to treat possible infection with anaerobic and gram-negative enteric pathogens.

To test for autonomic dysfunction, a plastic pen case drawn lightly across each forearm revealed a loss of tactile adherence (ie, areas where moist, sweaty skin impeded the movement of the pen case) on the affected forearm, a sign of underlying nerve injury. The affected forearm was sensitive to light touch, with pain out of proportion to the stimulus.

ARRIVING AT THE DIAGNOSIS

Based on the wide distribution of inflammation, autonomic dysfunction (shown by differences in temperature and sweating), radiographic evidence of demineralization, hyperesthesia, and lack of improvement in pain and swelling after two courses of antibiotics, the patient’s clinical course was determined to be consistent with complex regional pain syndrome type 1, previously referred to as reflex sympathetic dystrophy.

Symptoms of complex regional pain syndrome traditionally include pain, regional edema, joint stiffness, muscular atrophy, vasomotor disturbances (causing temperature variability and erythema), regional diaphoresis, and localized skeletal demineralization on radiography.

Complex regional pain syndrome type 1 occurs as regional pain and inflammation as an excessive sympathetic reaction to an often minor insult, without nerve injury. When the syndrome occurs in a patient with obvious partial nerve injury, it is categorized as type 2 (formerly known as causalgia). The two types are clinically indistinguishable and are not uncommon. About 10% of all patients with complex regional pain syndrome have obvious nerve injury (complex regional pain syndrome type 2). In a study of 109 patients with Colles fracture, 25% developed symptoms of complex regional pain syndrome.⁷

Complex regional pain syndrome is difficult to diagnose, as it resembles many other ailments, such as gout, infection, bone tumor, stress fracture, and arthritis. Its pathophysiology is poorly understood, but it is believed to result from a “short circuit” in the reflex arc between somatic afferent sensory fibers and autonomic sympathetic efferent fibers, and this is thought to explain the increased sympathetic stimulation.

Although the pathophysiology is likely the same in type 1 and type 2, electromyography with a nerve conduction study is a reliable way to detect nerve damage and thus distinguish between the two types of complex regional pain syndrome.⁸

Our understanding of this syndrome is evolving. A recent study using sensory testing showed that 33% of patients with type 1 had combinations of increased and decreased thresholds for the detection of thermal, vibratory, and mechanical stimuli in the distribution of discrete peripheral nerves, suggesting that the patients actually had type 2.⁹

CONFIRMING COMPLEX REGIONAL PAIN SYNDROME TYPE 1

3. Which of the following is the best way to confirm complex regional pain syndrome type 1?

• Erythrocyte sedimentation rate, C-reactive protein, and complete blood cell count
• Plain radiography of the hand and forearm
• Three-phase technetium bone scan
• The Budapest diagnostic criteria
• MRI
• Autonomic testing

Complex regional pain syndrome type 1 is a clinical diagnosis. Diagnostic studies lack sensitivity and specificity but may confirm complex regional pain syndrome type 1 or rule out other diagnoses. The Budapest diagnostic criteria¹⁰ (Table 1) may be the best way to confirm this diagnosis. The criteria are as follows: continuing pain disproportionate to an inciting event, coupled with three of four symptoms plus at least one sign from sensory, vasomotor, sudomotor, and motor-trophic categories.

Laboratory tests are not helpful because acute-phase reactants and blood counts remain normal in these patients.

Plain radiography is not sensitive in early diagnosis, but at 2 weeks it may show patchy areas of osteopenia in adjacent bones throughout the region, as well as subsequent diffuse demineralization.

Three-phase bone scanning is more sensitive than plain radiography, with 75% of patients showing regional disparities in blood flow in early sequences and increased bone uptake in the later sequences.

MRI is a sensitive early test, as it better defines focal areas of bone loss and increased T2 bone signal in adjacent bone, as well as early soft-tissue changes. Computed tomography does not show early specific changes in muscle, tendon, or bone and so is not recommended.

THE EVALUATION CONTINUES

The admitting diagnosis was septic arthritis, and our patient underwent computed tomography, which showed focal demineralization that could have represented bone infarcts or infection, confounding the diagnosis of complex regional pain syndrome.

Autonomic nerve testing can help distinguish complex regional pain syndrome from other disorders. Complex regional pain syndrome is characterized by increased sympathetic activity and results in increased sweat output. Autonomic testing includes resting sweat output, resting skin temperature, and quantitative sudomotor axon reflex testing. In one study, an increase in resting sweat output used in conjunction with quantitative sudomotor axon reflex testing predicted the diagnosis of complex regional pain syndrome with a specificity of 98%.¹¹ However, autonomic testing is limited to academic centers and is not readily available.

TREATING COMPLEX REGIONAL PAIN SYNDROME TYPE 1

4. Which is the best first-line therapy for complex regional pain syndrome type 1?

• Stellate ganglion nerve block
• Occupational therapy to splint the wrist and forearm
• Oral corticosteroids
• Physical therapy to prevent loss of joint motion
• Tricyclic antidepressant drugs (eg, amitriptyline), pregabalin, and bisphosphonates

Physical therapy should be started early in all patients, with range-of-motion exercises to prevent contracture and enhance mobility.

Stellate ganglion nerve block has been used to counter severe sympathetic hyperactivity, but it also may aggravate symptoms of complex regional pain syndrome and so remains a controversial treatment.

Immobilization and splinting should be avoided, as this will augment edema, pain, and contracture of joints.

Corticosteroids do not shorten the course or assuage symptoms and may increase edema.

Amitriptyline (Elavil) and pregabalin (Lyrica) have been used successfully to counter extended courses of allodynia and hyperalgesia. Bisphosphonates may decrease bone loss and pain and may be needed should the course be complicated by myositis ossificans, a form of dystrophic bone formation in juxtaposed tendon and muscle related to neuroactivation of fibroblasts and osteoblasts.

THE COURSE OF COMPLEX REGIONAL PAIN SYNDROME

Traditionally, type 1 was divided into three stages—an early inflammatory stage, a dystrophic stage, and a late atrophic stage.¹² Although there is no evidence to support a consistent three-stage evolution, the severity of symptoms may help determine the best approach to management.¹³

Patients initially exhibit burning or throbbing pain, diffuse aching, sensitivity to touch or cold (allodynia), localized edema, and vasomotor disturbances of variable intensity that may produce altered color and temperature. Topical capsaicin cream; a tricyclic antidepressant; an anticonvulsant such as gabapentin (Neurontin), pregabalin, or lamotrigine (Lamictal); or a nonsteroidal anti-inflammatory drug should be tried first. Some of these treatments are poorly tolerated in elderly patients. If pain persists, nasal calcitonin may be added. Trigger-point injections with an anesthetic or glucocorticoid may be tried.

The management of early complex regional pain syndrome is sometimes supplemented with systemic corticosteroids, but reviews of randomized controlled trials have failed to show efficacy.¹⁴

Later in the course, patients may suffer persistent soft-tissue edema, accompanied by thickening of the skin and periarticular soft tissues, muscle wasting, and the skin changes of brawny edema. Regional blockade of sympathetic ganglions, epidural administration of clonidine, implantable peripheral nerve stimulators, and spinal cord stimulators have all been applied by experts in pain management and may provide benefit. Progression of the syndrome may include cyanosis, mottling, increased sweating, abnormal hair growth, and diffuse swelling in nonarticular tissue.

It is always acceptable to refer to an experienced pain management specialist, and a multidisciplinary approach is recommended at the outset.¹²

OUR PATIENT’S CARE CONTINUED

Our patient’s forearm and wrist were placed in a sling to keep his left arm elevated when active. This helped control sympathetic vascular edema and throbbing pain. Physical therapy with range-of-motion exercises prevented contracture.

He was discharged home on limited oxycodone as needed, with close follow-up by his primary care physician to monitor his pain symptoms. The pain and swelling slowly improved over the next 2 months, but he suffered a fall, twisting his left wrist. This minor injury was followed by more intense pain and swelling of the forearm, hand, and wrist.

COMORBIDITIES

5. Which of the following statements about conditions associated with complex regional pain syndrome most likely applies to our patient?

• Gout is likely following minor trauma
• Minor trauma or surgical bone biopsy may reactivate complex regional pain syndrome
• Septic hip arthritis due to MRSA may have reemerged and seeded the wrist
• Patients with multiple sclerosis have a propensity for complex regional pain syndrome
• Complex regional pain syndrome type 1 begets type 2

Gout does follow minor injury, but our patient’s uric acid was well controlled on allopurinol (Zyloprim), and gout is unlikely to be causing polyarticular swelling of the hand, wrist, and forearm.

Minor trauma, sometimes inconsequential enough to have been completely forgotten, may either initiate complex regional pain syndrome or, as seen here, reactivate it. Bone changes seen on MRI sometimes trigger surgical bone biopsy, only to reactivate the dysesthesia and sympathetic vascular reaction. Surgery should be avoided. Trauma and surgery are causative rather than associative comorbidities.

Sepsis due to MRSA after total hip arthroplasty may be reactivated, especially in the setting of immunosuppressive treatment. But the diffuse bone changes seen in multiple carpal, radial, and ulnar bones suggest generalized vascular and sympathetic disarray, most consistent with complex regional pain syndrome type 1.

AN ASSOCIATION WITH MULTIPLE SCLEROSIS?

Multiple sclerosis and other central neuropathic conditions such as stroke are associated with complex regional pain syndrome type 1.^15,16

A hypothetical cause for the higher prevalence of complex regional pain syndrome in patients with multiple sclerosis may be demyelination resulting in aberrant signaling and overreaction to distal pain receptors. Demyelination of neurons within the autonomic or spinothalamic tracts potentially increases susceptibility to development of the pain syndrome.

Our patient had an apparent stimulus for the development of the syndrome, ie, the initial dog bite, and the wrist injury later may have caused peripheral nerve injury. Such injury may lead to release of vasodilatory neuropeptides including substance P from stimulated cutaneous nerves with cell bodies in the dorsal root ganglia. Excessive vasodilation and increased vascular permeability result in the affected limb becoming edematous and causing cutaneous nerves to be further activated. Stimulated cutaneous neurons normally have an inhibitory influence on sympathetic activity at the level of entry of the dorsal root ganglia in the cord. In complex regional pain syndrome, this inhibition is lost, resulting in a hyperactive somatosympathetic reflex.¹⁷ Underlying multiple sclerosis may have contributed to the loss of inhibition by the cutaneous nerves on the sympathetic system.

CASE CONCLUDED

We continued to closely follow this patient, who was on a self-directed program of physical therapy. One year after the original dog bite, the complex regional pain syndrome had completely resolved.

A 48-year-old man with gout, multiple sclerosis, and previously treated methicillin-resistant Staphylococcus aureus (MRSA) infection presented to the emergency room with pain and significant swelling at the site of a dog bite on his left forearm. He had been bitten 2 weeks earlier by a friend’s dog, and the bite had punctured the skin. He also had red streaking on the skin of the left arm from the wrist to the elbow, and he reported feeling “feverish” and having night sweats.

At first, the bite had seemed to improve, then swelling and pain had developed and increased. He reported this to his primary care physician, along with the information that he had previously had an anaphylactic reaction to penicillin and a cephalosporin. His physician, considering a penicillin allergy, started him on ciprofloxacin (Cipro) plus clindamycin (Cleocin). The patient took this for 5 days, but without improvement. The appearance of the red streaking on his left forearm prompted his presentation to our emergency room.

ORGANISMS IN DOG BITES

1. Which is the most common cause of infected dog bite?

• Pasteurella canis
• Streptococci and S aureus
• Erysipelothrix rhusiopathiae
• Capnocytophaga canimorsus
• Eikenella corrodens

Streptococci (50%) and S aureus (20% to 40%) are the organisms most commonly responsible for infected dog bites, as they are for other skin and soft-tissue infections.¹P canis is unique to dog bite infections but accounts for only 18%.²E rhusiopathiae is an unusual isolate from cat and dog bites and is more commonly isolated from the mouths of fish and aquatic mammals. C canimorsus is a normal inhabitant of the oral cavity of dogs and cats but an unusual cause of wound infection from a dog bite. It is notable for sepsis and central nervous system infections uniquely associated with veterinarians, dog owners, kennel workers, and mail carriers.³E corrodens infection is more common with human bites.⁴

THE EVALUATION BEGINS

On examination, the patient had marked edema of the left forearm and pain in the joints of the left hand. His temperature was 100.2°F (37.9°C). Because of the duration and severity of symptoms, the examining physician was concerned about septic arthritis of the wrist, and the patient was admitted to the hospital.

In the hospital, our patient was thermodynamically stable without documented fever or chills. There was no open wound to culture, and blood cultures were negative. Marked edema and joint involvement raised suspicion of erysipeloid. This “cousin” of erysipelas often involves the underlying joint, is associated with edema, and produces systemic manifestations of fever and arthralgia.

Radiography of the left forearm and hand demonstrated multiple foci of demineralization within the carpal bones and proximal radius, attributed to disuse. Magnetic resonance imaging (MRI) the next day showed multiple bone infarcts in the carpal bones and the distal radius, with synovitis and fluid in the carpal joints and without adjacent osteomyelitis. Fluid was also seen in the soft tissues in the ulnar aspect of the left wrist, and tenosynovitis involving the flexor carpi radialis tendon was noted.

Arthrocentesis of his left radiocarpal joint produced synovial fluid negative for crystals and negative on Gram stain; the fluid was also sent for culture. The patient’s tetanus immunization was current, and the dog was known to have been immunized against rabies.

ANTIBIOTICS FOR INFECTED DOG BITES

2. Which antibiotic regimen would you choose for this patient?

• Oral amoxicillin and clavulanate
• Meropenem
• Vancomycin, clindamycin, aztreonam
• Clindamycin and levofloxacin
• Clindamycin and trimethoprim-sulfamethoxazole

Oral amoxicillin and clavulanate (Augmentin) is a judicious choice for prophylactic treatment of deep bites in the early stages of infection. However, our patient’s wound was no longer in the early stages of infection, and he had a history of an adverse reaction to penicillin.

Meropenem (Merrem IV) cross-reacts minimally with penicillin allergy and is reported to be safe in patients with a history of anaphylactic reactions to penicillin,⁵ but overuse of carbapenems has led to the development of carbapenem-resistant strains of Klebsiella, Stenotrophomonas, and Acinetobacter organisms.

Given the rise of MRSA infections and the common involvement of staphylococci, streptococci, and anaerobic bacteria in complicated dog bites, the combination of vancomycin and clindamycin is a good choice, and aztreonam (Azactam) would add empiric coverage of gram-negative enteric organisms.

Levofloxacin (Levaquin) also covers gramnegative enteric organisms, but Fusobacterium canifelinum, a common anaerobe in the oral flora of dogs and cats, is intrinsically resistant to fluoroquinolones.

Clindamycin and levofloxacin would be a good step-down oral regimen. Pasteurella multocida has variable sensitivity to the commonly used agents dicloxacillin (Dynapen), cephalexin (Keflex), macrolides, and clindamycin, but it is a less likely pathogen at this late stage and could be covered with levofloxacin alone.

C canimorsus is resistant to trimethoprim-sulfamethoxazole (Bactrim) and cephalexin, but is well covered by clindamycin.⁶

CASE CONTINUED

Our patient was started on intravenous vancomycin, clindamycin, and aztreonam for coverage of dog-mouth flora. Blood cultures and cultures of synovial fluid of the left wrist were negative. Vancomycin was discontinued after 48 hours when blood cultures did not grow staphylococcal organisms, but clindamycin and aztreonam were continued for a total of 8 days to treat possible infection with anaerobic and gram-negative enteric pathogens.

To test for autonomic dysfunction, a plastic pen case drawn lightly across each forearm revealed a loss of tactile adherence (ie, areas where moist, sweaty skin impeded the movement of the pen case) on the affected forearm, a sign of underlying nerve injury. The affected forearm was sensitive to light touch, with pain out of proportion to the stimulus.

ARRIVING AT THE DIAGNOSIS

Based on the wide distribution of inflammation, autonomic dysfunction (shown by differences in temperature and sweating), radiographic evidence of demineralization, hyperesthesia, and lack of improvement in pain and swelling after two courses of antibiotics, the patient’s clinical course was determined to be consistent with complex regional pain syndrome type 1, previously referred to as reflex sympathetic dystrophy.

Symptoms of complex regional pain syndrome traditionally include pain, regional edema, joint stiffness, muscular atrophy, vasomotor disturbances (causing temperature variability and erythema), regional diaphoresis, and localized skeletal demineralization on radiography.

Complex regional pain syndrome type 1 occurs as regional pain and inflammation as an excessive sympathetic reaction to an often minor insult, without nerve injury. When the syndrome occurs in a patient with obvious partial nerve injury, it is categorized as type 2 (formerly known as causalgia). The two types are clinically indistinguishable and are not uncommon. About 10% of all patients with complex regional pain syndrome have obvious nerve injury (complex regional pain syndrome type 2). In a study of 109 patients with Colles fracture, 25% developed symptoms of complex regional pain syndrome.⁷

Complex regional pain syndrome is difficult to diagnose, as it resembles many other ailments, such as gout, infection, bone tumor, stress fracture, and arthritis. Its pathophysiology is poorly understood, but it is believed to result from a “short circuit” in the reflex arc between somatic afferent sensory fibers and autonomic sympathetic efferent fibers, and this is thought to explain the increased sympathetic stimulation.

Although the pathophysiology is likely the same in type 1 and type 2, electromyography with a nerve conduction study is a reliable way to detect nerve damage and thus distinguish between the two types of complex regional pain syndrome.⁸

Our understanding of this syndrome is evolving. A recent study using sensory testing showed that 33% of patients with type 1 had combinations of increased and decreased thresholds for the detection of thermal, vibratory, and mechanical stimuli in the distribution of discrete peripheral nerves, suggesting that the patients actually had type 2.⁹

CONFIRMING COMPLEX REGIONAL PAIN SYNDROME TYPE 1

3. Which of the following is the best way to confirm complex regional pain syndrome type 1?

• Erythrocyte sedimentation rate, C-reactive protein, and complete blood cell count
• Plain radiography of the hand and forearm
• Three-phase technetium bone scan
• The Budapest diagnostic criteria
• MRI
• Autonomic testing

Complex regional pain syndrome type 1 is a clinical diagnosis. Diagnostic studies lack sensitivity and specificity but may confirm complex regional pain syndrome type 1 or rule out other diagnoses. The Budapest diagnostic criteria¹⁰ (Table 1) may be the best way to confirm this diagnosis. The criteria are as follows: continuing pain disproportionate to an inciting event, coupled with three of four symptoms plus at least one sign from sensory, vasomotor, sudomotor, and motor-trophic categories.

Laboratory tests are not helpful because acute-phase reactants and blood counts remain normal in these patients.

Plain radiography is not sensitive in early diagnosis, but at 2 weeks it may show patchy areas of osteopenia in adjacent bones throughout the region, as well as subsequent diffuse demineralization.

Three-phase bone scanning is more sensitive than plain radiography, with 75% of patients showing regional disparities in blood flow in early sequences and increased bone uptake in the later sequences.

MRI is a sensitive early test, as it better defines focal areas of bone loss and increased T2 bone signal in adjacent bone, as well as early soft-tissue changes. Computed tomography does not show early specific changes in muscle, tendon, or bone and so is not recommended.

THE EVALUATION CONTINUES

The admitting diagnosis was septic arthritis, and our patient underwent computed tomography, which showed focal demineralization that could have represented bone infarcts or infection, confounding the diagnosis of complex regional pain syndrome.

Autonomic nerve testing can help distinguish complex regional pain syndrome from other disorders. Complex regional pain syndrome is characterized by increased sympathetic activity and results in increased sweat output. Autonomic testing includes resting sweat output, resting skin temperature, and quantitative sudomotor axon reflex testing. In one study, an increase in resting sweat output used in conjunction with quantitative sudomotor axon reflex testing predicted the diagnosis of complex regional pain syndrome with a specificity of 98%.¹¹ However, autonomic testing is limited to academic centers and is not readily available.

TREATING COMPLEX REGIONAL PAIN SYNDROME TYPE 1

4. Which is the best first-line therapy for complex regional pain syndrome type 1?

• Stellate ganglion nerve block
• Occupational therapy to splint the wrist and forearm
• Oral corticosteroids
• Physical therapy to prevent loss of joint motion
• Tricyclic antidepressant drugs (eg, amitriptyline), pregabalin, and bisphosphonates

Physical therapy should be started early in all patients, with range-of-motion exercises to prevent contracture and enhance mobility.

Stellate ganglion nerve block has been used to counter severe sympathetic hyperactivity, but it also may aggravate symptoms of complex regional pain syndrome and so remains a controversial treatment.

Immobilization and splinting should be avoided, as this will augment edema, pain, and contracture of joints.

Corticosteroids do not shorten the course or assuage symptoms and may increase edema.

Amitriptyline (Elavil) and pregabalin (Lyrica) have been used successfully to counter extended courses of allodynia and hyperalgesia. Bisphosphonates may decrease bone loss and pain and may be needed should the course be complicated by myositis ossificans, a form of dystrophic bone formation in juxtaposed tendon and muscle related to neuroactivation of fibroblasts and osteoblasts.

THE COURSE OF COMPLEX REGIONAL PAIN SYNDROME

Traditionally, type 1 was divided into three stages—an early inflammatory stage, a dystrophic stage, and a late atrophic stage.¹² Although there is no evidence to support a consistent three-stage evolution, the severity of symptoms may help determine the best approach to management.¹³

Patients initially exhibit burning or throbbing pain, diffuse aching, sensitivity to touch or cold (allodynia), localized edema, and vasomotor disturbances of variable intensity that may produce altered color and temperature. Topical capsaicin cream; a tricyclic antidepressant; an anticonvulsant such as gabapentin (Neurontin), pregabalin, or lamotrigine (Lamictal); or a nonsteroidal anti-inflammatory drug should be tried first. Some of these treatments are poorly tolerated in elderly patients. If pain persists, nasal calcitonin may be added. Trigger-point injections with an anesthetic or glucocorticoid may be tried.

The management of early complex regional pain syndrome is sometimes supplemented with systemic corticosteroids, but reviews of randomized controlled trials have failed to show efficacy.¹⁴

Later in the course, patients may suffer persistent soft-tissue edema, accompanied by thickening of the skin and periarticular soft tissues, muscle wasting, and the skin changes of brawny edema. Regional blockade of sympathetic ganglions, epidural administration of clonidine, implantable peripheral nerve stimulators, and spinal cord stimulators have all been applied by experts in pain management and may provide benefit. Progression of the syndrome may include cyanosis, mottling, increased sweating, abnormal hair growth, and diffuse swelling in nonarticular tissue.

It is always acceptable to refer to an experienced pain management specialist, and a multidisciplinary approach is recommended at the outset.¹²

OUR PATIENT’S CARE CONTINUED

Our patient’s forearm and wrist were placed in a sling to keep his left arm elevated when active. This helped control sympathetic vascular edema and throbbing pain. Physical therapy with range-of-motion exercises prevented contracture.

He was discharged home on limited oxycodone as needed, with close follow-up by his primary care physician to monitor his pain symptoms. The pain and swelling slowly improved over the next 2 months, but he suffered a fall, twisting his left wrist. This minor injury was followed by more intense pain and swelling of the forearm, hand, and wrist.

COMORBIDITIES

5. Which of the following statements about conditions associated with complex regional pain syndrome most likely applies to our patient?

• Gout is likely following minor trauma
• Minor trauma or surgical bone biopsy may reactivate complex regional pain syndrome
• Septic hip arthritis due to MRSA may have reemerged and seeded the wrist
• Patients with multiple sclerosis have a propensity for complex regional pain syndrome
• Complex regional pain syndrome type 1 begets type 2

Gout does follow minor injury, but our patient’s uric acid was well controlled on allopurinol (Zyloprim), and gout is unlikely to be causing polyarticular swelling of the hand, wrist, and forearm.

Minor trauma, sometimes inconsequential enough to have been completely forgotten, may either initiate complex regional pain syndrome or, as seen here, reactivate it. Bone changes seen on MRI sometimes trigger surgical bone biopsy, only to reactivate the dysesthesia and sympathetic vascular reaction. Surgery should be avoided. Trauma and surgery are causative rather than associative comorbidities.

Sepsis due to MRSA after total hip arthroplasty may be reactivated, especially in the setting of immunosuppressive treatment. But the diffuse bone changes seen in multiple carpal, radial, and ulnar bones suggest generalized vascular and sympathetic disarray, most consistent with complex regional pain syndrome type 1.

AN ASSOCIATION WITH MULTIPLE SCLEROSIS?

Multiple sclerosis and other central neuropathic conditions such as stroke are associated with complex regional pain syndrome type 1.^15,16

A hypothetical cause for the higher prevalence of complex regional pain syndrome in patients with multiple sclerosis may be demyelination resulting in aberrant signaling and overreaction to distal pain receptors. Demyelination of neurons within the autonomic or spinothalamic tracts potentially increases susceptibility to development of the pain syndrome.

Our patient had an apparent stimulus for the development of the syndrome, ie, the initial dog bite, and the wrist injury later may have caused peripheral nerve injury. Such injury may lead to release of vasodilatory neuropeptides including substance P from stimulated cutaneous nerves with cell bodies in the dorsal root ganglia. Excessive vasodilation and increased vascular permeability result in the affected limb becoming edematous and causing cutaneous nerves to be further activated. Stimulated cutaneous neurons normally have an inhibitory influence on sympathetic activity at the level of entry of the dorsal root ganglia in the cord. In complex regional pain syndrome, this inhibition is lost, resulting in a hyperactive somatosympathetic reflex.¹⁷ Underlying multiple sclerosis may have contributed to the loss of inhibition by the cutaneous nerves on the sympathetic system.

CASE CONCLUDED

We continued to closely follow this patient, who was on a self-directed program of physical therapy. One year after the original dog bite, the complex regional pain syndrome had completely resolved.

A 51-year-old man with end-stage liver disease from alcohol abuse presented with worsening dyspnea on exertion. He had a history of ascites requiring diuretic therapy and intermittent paracentesis, as well as symptomatic hepatic hydrothorax requiring thoracentesis. Chest radiography showed a large right hydrothorax (Figure 1).

See related commentary

The patient underwent high-volume thoracentesis, and 3.2 L of clear fluid was removed. Chest radiography after the procedure revealed a right-sided pneumothorax (Figure 2, arrow). The patient was mildly short of breath and was treated with high-flow oxygen. Later the same day, his shortness of breath worsened, and repeat chest radiography showed an unchanged pneumothorax that was now complicated by reexpansion pulmonary edema after thoracentesis (Figure 3, star). The reexpansion pulmonary edema resolved by the following day, and the pneumothorax resolved after placement of a pig-tail catheter into the pleural space (Figure 4).

Iatrogenic pneumothorax after thoracentesis occurs in 6% of cases.¹ Iatrogenic reexpansion pulmonary edema after thoracentesis occurs in fewer than 1% of cases.^2,3 Simultaneous pneumothorax and reexpansion pulmonary edema arising from the same procedure appears to be extremely rare.

A 51-year-old man with end-stage liver disease from alcohol abuse presented with worsening dyspnea on exertion. He had a history of ascites requiring diuretic therapy and intermittent paracentesis, as well as symptomatic hepatic hydrothorax requiring thoracentesis. Chest radiography showed a large right hydrothorax (Figure 1).

See related commentary

The patient underwent high-volume thoracentesis, and 3.2 L of clear fluid was removed. Chest radiography after the procedure revealed a right-sided pneumothorax (Figure 2, arrow). The patient was mildly short of breath and was treated with high-flow oxygen. Later the same day, his shortness of breath worsened, and repeat chest radiography showed an unchanged pneumothorax that was now complicated by reexpansion pulmonary edema after thoracentesis (Figure 3, star). The reexpansion pulmonary edema resolved by the following day, and the pneumothorax resolved after placement of a pig-tail catheter into the pleural space (Figure 4).

Iatrogenic pneumothorax after thoracentesis occurs in 6% of cases.¹ Iatrogenic reexpansion pulmonary edema after thoracentesis occurs in fewer than 1% of cases.^2,3 Simultaneous pneumothorax and reexpansion pulmonary edema arising from the same procedure appears to be extremely rare.

A 51-year-old man with end-stage liver disease from alcohol abuse presented with worsening dyspnea on exertion. He had a history of ascites requiring diuretic therapy and intermittent paracentesis, as well as symptomatic hepatic hydrothorax requiring thoracentesis. Chest radiography showed a large right hydrothorax (Figure 1).

See related commentary

The patient underwent high-volume thoracentesis, and 3.2 L of clear fluid was removed. Chest radiography after the procedure revealed a right-sided pneumothorax (Figure 2, arrow). The patient was mildly short of breath and was treated with high-flow oxygen. Later the same day, his shortness of breath worsened, and repeat chest radiography showed an unchanged pneumothorax that was now complicated by reexpansion pulmonary edema after thoracentesis (Figure 3, star). The reexpansion pulmonary edema resolved by the following day, and the pneumothorax resolved after placement of a pig-tail catheter into the pleural space (Figure 4).

Iatrogenic pneumothorax after thoracentesis occurs in 6% of cases.¹ Iatrogenic reexpansion pulmonary edema after thoracentesis occurs in fewer than 1% of cases.^2,3 Simultaneous pneumothorax and reexpansion pulmonary edema arising from the same procedure appears to be extremely rare.

Large-volume thoracentesis is defined as the drainage of more than 1 L of fluid. Inherent in this procedure is the removal of a large amount of fluid from a cavity with a rigid wall, which leads to changes in pleural pressure and to expansion of the lung. Two specific complications occur, pneumothorax and reexpansion pulmonary edema. The images submitted for the Clinical Picture article by Drs. Apter and Aronowitz in this issue of the Journal highlight these complications.

See related article

Retrospective studies have found an association between the amount of fluid drained and the incidence of pneumothorax.^1,2 Although technical issues may account for it (eg, needle injury to the lung that leads to postprocedural pneumothorax), the available evidence suggests that it has more to do with the drainage of larger volumes than the lung can expand to fill.^3,4 That is, the patient’s lung cannot expand,⁵ so drainage creates a vacuum, and air enters the pleural space³ through the lung parenchyma, or perhaps from around the drainage catheter.

In a series of patients who underwent therapeutic thoracentesis,³ 23 (8.7%) of 265 patients had pneumothorax. Interestingly, some patients had only symptoms, some had only excessively negative pressures (< 25 cm H₂O), some had both, and some had neither. Thus, there does not seem to be a reliable sign or symptom of an unexpanding lung, but pleural manometry may help increase its detection.⁶ This technique, however, is rarely used in clinical practice.

Another consequence of therapeutic thoracentesis is reexpansion pulmonary edema. This rare condition occurs only after large-volume thoracentesis or evacuation of a moderate to large pneumothorax.⁷ The pathophysiology behind this is controversial.⁸ As with pneumothorax, a large case series did not find a correlation between volume removed or pleural pressures and reexpansion pulmonary edema.⁷ Experimental data and analysis of case series^8–10 suggest that the duration of lung collapse and the speed of drainage and negative pressure applied contribute to the development of edema. Vacuum bottles are often used to speed drainage and to contain the large amount of fluid drained. These bottles have an initial negative pressure of about −723 mm Hg (personal communication with Baxter Healthcare Product information line), which may lead to rapid changes in lung volume and perhaps to higher negative pleural pressures.

Given the risks discussed above, we believe it is appropriate to avoid vacuum bottles and instead to use the syringe and one-way valve supplied in most thoracentesis kits. Further, pleural manometry to detect changes in pressure that suggest an unexpandable lung may lead to the appropriate early termination of a planned large-volume thoracentesis.³ The complications reported by Drs. Apter and Aronowitz are relatively rare and, at this point, unpredictable; therefore, generating high-quality evidence for prediction or management will be difficult. In the meantime, understanding the physiologic changes in the lung and the pleural space when draining large effusions from the chest may help avoid double trouble.

Large-volume thoracentesis is defined as the drainage of more than 1 L of fluid. Inherent in this procedure is the removal of a large amount of fluid from a cavity with a rigid wall, which leads to changes in pleural pressure and to expansion of the lung. Two specific complications occur, pneumothorax and reexpansion pulmonary edema. The images submitted for the Clinical Picture article by Drs. Apter and Aronowitz in this issue of the Journal highlight these complications.

See related article

Retrospective studies have found an association between the amount of fluid drained and the incidence of pneumothorax.^1,2 Although technical issues may account for it (eg, needle injury to the lung that leads to postprocedural pneumothorax), the available evidence suggests that it has more to do with the drainage of larger volumes than the lung can expand to fill.^3,4 That is, the patient’s lung cannot expand,⁵ so drainage creates a vacuum, and air enters the pleural space³ through the lung parenchyma, or perhaps from around the drainage catheter.

In a series of patients who underwent therapeutic thoracentesis,³ 23 (8.7%) of 265 patients had pneumothorax. Interestingly, some patients had only symptoms, some had only excessively negative pressures (< 25 cm H₂O), some had both, and some had neither. Thus, there does not seem to be a reliable sign or symptom of an unexpanding lung, but pleural manometry may help increase its detection.⁶ This technique, however, is rarely used in clinical practice.

Another consequence of therapeutic thoracentesis is reexpansion pulmonary edema. This rare condition occurs only after large-volume thoracentesis or evacuation of a moderate to large pneumothorax.⁷ The pathophysiology behind this is controversial.⁸ As with pneumothorax, a large case series did not find a correlation between volume removed or pleural pressures and reexpansion pulmonary edema.⁷ Experimental data and analysis of case series^8–10 suggest that the duration of lung collapse and the speed of drainage and negative pressure applied contribute to the development of edema. Vacuum bottles are often used to speed drainage and to contain the large amount of fluid drained. These bottles have an initial negative pressure of about −723 mm Hg (personal communication with Baxter Healthcare Product information line), which may lead to rapid changes in lung volume and perhaps to higher negative pleural pressures.

Given the risks discussed above, we believe it is appropriate to avoid vacuum bottles and instead to use the syringe and one-way valve supplied in most thoracentesis kits. Further, pleural manometry to detect changes in pressure that suggest an unexpandable lung may lead to the appropriate early termination of a planned large-volume thoracentesis.³ The complications reported by Drs. Apter and Aronowitz are relatively rare and, at this point, unpredictable; therefore, generating high-quality evidence for prediction or management will be difficult. In the meantime, understanding the physiologic changes in the lung and the pleural space when draining large effusions from the chest may help avoid double trouble.

Large-volume thoracentesis is defined as the drainage of more than 1 L of fluid. Inherent in this procedure is the removal of a large amount of fluid from a cavity with a rigid wall, which leads to changes in pleural pressure and to expansion of the lung. Two specific complications occur, pneumothorax and reexpansion pulmonary edema. The images submitted for the Clinical Picture article by Drs. Apter and Aronowitz in this issue of the Journal highlight these complications.

See related article

Retrospective studies have found an association between the amount of fluid drained and the incidence of pneumothorax.^1,2 Although technical issues may account for it (eg, needle injury to the lung that leads to postprocedural pneumothorax), the available evidence suggests that it has more to do with the drainage of larger volumes than the lung can expand to fill.^3,4 That is, the patient’s lung cannot expand,⁵ so drainage creates a vacuum, and air enters the pleural space³ through the lung parenchyma, or perhaps from around the drainage catheter.

In a series of patients who underwent therapeutic thoracentesis,³ 23 (8.7%) of 265 patients had pneumothorax. Interestingly, some patients had only symptoms, some had only excessively negative pressures (< 25 cm H₂O), some had both, and some had neither. Thus, there does not seem to be a reliable sign or symptom of an unexpanding lung, but pleural manometry may help increase its detection.⁶ This technique, however, is rarely used in clinical practice.

Another consequence of therapeutic thoracentesis is reexpansion pulmonary edema. This rare condition occurs only after large-volume thoracentesis or evacuation of a moderate to large pneumothorax.⁷ The pathophysiology behind this is controversial.⁸ As with pneumothorax, a large case series did not find a correlation between volume removed or pleural pressures and reexpansion pulmonary edema.⁷ Experimental data and analysis of case series^8–10 suggest that the duration of lung collapse and the speed of drainage and negative pressure applied contribute to the development of edema. Vacuum bottles are often used to speed drainage and to contain the large amount of fluid drained. These bottles have an initial negative pressure of about −723 mm Hg (personal communication with Baxter Healthcare Product information line), which may lead to rapid changes in lung volume and perhaps to higher negative pleural pressures.

Given the risks discussed above, we believe it is appropriate to avoid vacuum bottles and instead to use the syringe and one-way valve supplied in most thoracentesis kits. Further, pleural manometry to detect changes in pressure that suggest an unexpandable lung may lead to the appropriate early termination of a planned large-volume thoracentesis.³ The complications reported by Drs. Apter and Aronowitz are relatively rare and, at this point, unpredictable; therefore, generating high-quality evidence for prediction or management will be difficult. In the meantime, understanding the physiologic changes in the lung and the pleural space when draining large effusions from the chest may help avoid double trouble.