The term cardiovascular implantable electronic device (CIED) includes both permanent pacemakers and implantable cardioverter-defibrillators. These devices are being implanted in more people every year.¹ They have also become increasingly sophisticated, with newer devices capable of both pacing and cardioversion-defibrillation functions.² Patients receiving these devices are also increasingly older and have more comorbid conditions.^3,4 As more CIEDs are placed in older and sicker patients, infections of these devices can be expected to be encountered with increasing frequency.

See related article

In this issue of the Cleveland Clinic Journal of Medicine, Dababneh and Sohail⁵ review CIED infections and provide a stepwise approach to their diagnosis and treatment.

HOW THE DEVICES BECOME INFECTED

CIEDs can become infected during implantation, in which case the infection presents early on, usually with pocket manifestations, or by secondary hematogenous seeding, in which case the infection generally presents with endovascular manifestations. Dababneh and Sohail have elegantly outlined the risk factors that predispose to infection of these devices.

If there are no early complications, patients generally do well with these devices. However, many patients do fine with their first device but develop a pocket infection when the pulse generator is changed because of battery depletion or other reasons. When patients with a CIED develop bacteremia as a complication of a vascular catheter infection or other infection, particularly with Staphylococcus aureus, they are at increased risk of having the intravascular portion of their device seeded.

PATIENTS MAY NOT APPEAR VERY ILL AT PRESENTATION

Dababneh and Sohail divide the clinical presentations of CIED infection into two broad categories: pocket infection and endovascular infection with an intact pocket. This is a useful categorization, as it provides a clue to pathogenesis.

As the authors point out, most patients with CIED infection present first to their primary care physician when they develop symptoms. An understanding of this infection by primary care physicians will allow for early recognition and more timely treatment, thus avoiding unnecessary complications.

Patients with pocket infection may not appear ill, but this should not lead a clinician away from the diagnosis. A pocket hematoma is an important differential diagnosis in the early postoperative period after device implantation or pulse generator change, and it may be difficult to decide if pocket changes are from an uninfected hematoma or from an infection.

Patients with endovascular infection are more likely to have systemic symptoms such as fever, fatigue, and malaise. However, absence of systemic features does not necessarily exclude endovascular infection.

BLOOD CULTURES AND TEE ARE KEY DIAGNOSTIC TESTS

All patients with suspected CIED infection should have at least two sets of blood cultures checked, even if they appear to be reasonably well. If there is any suspicion of endovascular infection, echocardiography should be performed.

Transesophageal echocardiography (TEE) is far superior to transthoracic echocardiography (TTE) for detecting lead vegetations.⁶ TEE should be carefully performed whenever endovascular infection is suspected, including all patients with positive blood cultures and all patients with systemic signs and symptoms.

Purulent drainage should be cultured, and when the device is removed, cultures of lead tips and pocket tissue should be done as well.

TREATMENT USUALLY REQUIRES COMPLETE DEVICE REMOVAL

A superficial infection in the early postoperative period may respond to antibiotic therapy alone. But in all other patients, the device must be removed to cure the infection. In referral centers, it is not unusual to see patients who have been referred after having been treated with antibiotics for weeks and sometimes months in the mistaken belief that the infection would be cured with antibiotics alone.

In some patients presenting with only pocket findings in the early postoperative period, it may be difficult indeed to tell if there is pocket infection. In such patients, it is not necessary to make a hasty decision to remove the device, but it is important to monitor them closely until the presence or absence of infection becomes clear. Also, erosion of the device through the skin represents pocket infection even if the patient appears otherwise healthy.

When removing the device, it is necessary to remove the generator and all leads to treat the infection effectively.

If patients are device-dependent, it is usually safe to place a new device with the new pulse generator pocket in a different location from the infected one a few days after the infected device is removed.

AREAS OF UNCERTAINTY AND CHALLENGE

Although there is no controversy about the need for complete removal of infected devices in order to effect a cure, the appropriate duration of antibiotic therapy after device removal is less clear. Dababneh and Sohail provide a useful algorithm to help with this decision. Patients usually need a new device to replace the infected one and there is a legitimate reason for concern about undertreating, since one would not want the new device to become infected because of inadequate antibiotic therapy. When endovascular infection is suspected or documented, patients are probably best treated as they would be for infective endocarditis.

Difficulties arise when patients with a CIED develop bacteremia with no echocardiographic evidence of device infection. Finding the source of bacteremia is very important because a diagnosis of CIED infection indicates that the device has to be removed. When there is a clear alternative explanation for the bacteremia, the CIED does not have to be removed. The type of bacterium helps clinicians to gauge the likelihood of CIED infection and to decide on the appropriate course of action. These cases should always be managed in conjunction with an infectious disease specialist and a cardiac electrophysiologist.

Another concern is secondary seeding of an uninfected CIED caused by bacteremia from another source. This concern is particularly acute with S aureus bacteremia. When patients with a CIED and S aureus bacteremia have been studied, endovascular CIED infection was documented in about half, although only a few had evidence of pocket inflammation.^7,8 This suggests that the devices were seeded via the endovascular route.

Medical procedures such as dialysis and total parenteral nutrition require frequent intravascular access—often facilitated by leaving an indwelling vascular catheter in place. Frequent entry into the intravascular compartment puts patients at substantial risk of bloodstream infection, and in patients with a CIED this can be complicated by device infection. In patients with a CIED and an indwelling vascular catheter who develop bacteremia, determining the source of the bacteremia is particularly challenging, as is the treatment. Thus, preventing endovascular infection in such patients is extremely desirable, but there are no easy solutions.

PLACING CIED INFECTIONS IN PERSPECTIVE

The vast majority of patients with a CIED never develop a device infection. Those unfortunate enough to have a CIED infection have little choice other than to have the device removed, but those diagnosed early and treated appropriately generally do well. The development of CIEDs has been an important advance in the practice of cardiac electrophysiology. An appropriate understanding of CIED infection and its treatment will help optimize the diagnosis and management of this complication when it does occur.

The term cardiovascular implantable electronic device (CIED) includes both permanent pacemakers and implantable cardioverter-defibrillators. These devices are being implanted in more people every year.¹ They have also become increasingly sophisticated, with newer devices capable of both pacing and cardioversion-defibrillation functions.² Patients receiving these devices are also increasingly older and have more comorbid conditions.^3,4 As more CIEDs are placed in older and sicker patients, infections of these devices can be expected to be encountered with increasing frequency.

See related article

In this issue of the Cleveland Clinic Journal of Medicine, Dababneh and Sohail⁵ review CIED infections and provide a stepwise approach to their diagnosis and treatment.

HOW THE DEVICES BECOME INFECTED

CIEDs can become infected during implantation, in which case the infection presents early on, usually with pocket manifestations, or by secondary hematogenous seeding, in which case the infection generally presents with endovascular manifestations. Dababneh and Sohail have elegantly outlined the risk factors that predispose to infection of these devices.

If there are no early complications, patients generally do well with these devices. However, many patients do fine with their first device but develop a pocket infection when the pulse generator is changed because of battery depletion or other reasons. When patients with a CIED develop bacteremia as a complication of a vascular catheter infection or other infection, particularly with Staphylococcus aureus, they are at increased risk of having the intravascular portion of their device seeded.

PATIENTS MAY NOT APPEAR VERY ILL AT PRESENTATION

Dababneh and Sohail divide the clinical presentations of CIED infection into two broad categories: pocket infection and endovascular infection with an intact pocket. This is a useful categorization, as it provides a clue to pathogenesis.

As the authors point out, most patients with CIED infection present first to their primary care physician when they develop symptoms. An understanding of this infection by primary care physicians will allow for early recognition and more timely treatment, thus avoiding unnecessary complications.

Patients with pocket infection may not appear ill, but this should not lead a clinician away from the diagnosis. A pocket hematoma is an important differential diagnosis in the early postoperative period after device implantation or pulse generator change, and it may be difficult to decide if pocket changes are from an uninfected hematoma or from an infection.

Patients with endovascular infection are more likely to have systemic symptoms such as fever, fatigue, and malaise. However, absence of systemic features does not necessarily exclude endovascular infection.

BLOOD CULTURES AND TEE ARE KEY DIAGNOSTIC TESTS

All patients with suspected CIED infection should have at least two sets of blood cultures checked, even if they appear to be reasonably well. If there is any suspicion of endovascular infection, echocardiography should be performed.

Transesophageal echocardiography (TEE) is far superior to transthoracic echocardiography (TTE) for detecting lead vegetations.⁶ TEE should be carefully performed whenever endovascular infection is suspected, including all patients with positive blood cultures and all patients with systemic signs and symptoms.

Purulent drainage should be cultured, and when the device is removed, cultures of lead tips and pocket tissue should be done as well.

TREATMENT USUALLY REQUIRES COMPLETE DEVICE REMOVAL

A superficial infection in the early postoperative period may respond to antibiotic therapy alone. But in all other patients, the device must be removed to cure the infection. In referral centers, it is not unusual to see patients who have been referred after having been treated with antibiotics for weeks and sometimes months in the mistaken belief that the infection would be cured with antibiotics alone.

In some patients presenting with only pocket findings in the early postoperative period, it may be difficult indeed to tell if there is pocket infection. In such patients, it is not necessary to make a hasty decision to remove the device, but it is important to monitor them closely until the presence or absence of infection becomes clear. Also, erosion of the device through the skin represents pocket infection even if the patient appears otherwise healthy.

When removing the device, it is necessary to remove the generator and all leads to treat the infection effectively.

If patients are device-dependent, it is usually safe to place a new device with the new pulse generator pocket in a different location from the infected one a few days after the infected device is removed.

AREAS OF UNCERTAINTY AND CHALLENGE

Although there is no controversy about the need for complete removal of infected devices in order to effect a cure, the appropriate duration of antibiotic therapy after device removal is less clear. Dababneh and Sohail provide a useful algorithm to help with this decision. Patients usually need a new device to replace the infected one and there is a legitimate reason for concern about undertreating, since one would not want the new device to become infected because of inadequate antibiotic therapy. When endovascular infection is suspected or documented, patients are probably best treated as they would be for infective endocarditis.

Difficulties arise when patients with a CIED develop bacteremia with no echocardiographic evidence of device infection. Finding the source of bacteremia is very important because a diagnosis of CIED infection indicates that the device has to be removed. When there is a clear alternative explanation for the bacteremia, the CIED does not have to be removed. The type of bacterium helps clinicians to gauge the likelihood of CIED infection and to decide on the appropriate course of action. These cases should always be managed in conjunction with an infectious disease specialist and a cardiac electrophysiologist.

Another concern is secondary seeding of an uninfected CIED caused by bacteremia from another source. This concern is particularly acute with S aureus bacteremia. When patients with a CIED and S aureus bacteremia have been studied, endovascular CIED infection was documented in about half, although only a few had evidence of pocket inflammation.^7,8 This suggests that the devices were seeded via the endovascular route.

Medical procedures such as dialysis and total parenteral nutrition require frequent intravascular access—often facilitated by leaving an indwelling vascular catheter in place. Frequent entry into the intravascular compartment puts patients at substantial risk of bloodstream infection, and in patients with a CIED this can be complicated by device infection. In patients with a CIED and an indwelling vascular catheter who develop bacteremia, determining the source of the bacteremia is particularly challenging, as is the treatment. Thus, preventing endovascular infection in such patients is extremely desirable, but there are no easy solutions.

PLACING CIED INFECTIONS IN PERSPECTIVE

The vast majority of patients with a CIED never develop a device infection. Those unfortunate enough to have a CIED infection have little choice other than to have the device removed, but those diagnosed early and treated appropriately generally do well. The development of CIEDs has been an important advance in the practice of cardiac electrophysiology. An appropriate understanding of CIED infection and its treatment will help optimize the diagnosis and management of this complication when it does occur.

The term cardiovascular implantable electronic device (CIED) includes both permanent pacemakers and implantable cardioverter-defibrillators. These devices are being implanted in more people every year.¹ They have also become increasingly sophisticated, with newer devices capable of both pacing and cardioversion-defibrillation functions.² Patients receiving these devices are also increasingly older and have more comorbid conditions.^3,4 As more CIEDs are placed in older and sicker patients, infections of these devices can be expected to be encountered with increasing frequency.

See related article

In this issue of the Cleveland Clinic Journal of Medicine, Dababneh and Sohail⁵ review CIED infections and provide a stepwise approach to their diagnosis and treatment.

HOW THE DEVICES BECOME INFECTED

CIEDs can become infected during implantation, in which case the infection presents early on, usually with pocket manifestations, or by secondary hematogenous seeding, in which case the infection generally presents with endovascular manifestations. Dababneh and Sohail have elegantly outlined the risk factors that predispose to infection of these devices.

If there are no early complications, patients generally do well with these devices. However, many patients do fine with their first device but develop a pocket infection when the pulse generator is changed because of battery depletion or other reasons. When patients with a CIED develop bacteremia as a complication of a vascular catheter infection or other infection, particularly with Staphylococcus aureus, they are at increased risk of having the intravascular portion of their device seeded.

PATIENTS MAY NOT APPEAR VERY ILL AT PRESENTATION

Dababneh and Sohail divide the clinical presentations of CIED infection into two broad categories: pocket infection and endovascular infection with an intact pocket. This is a useful categorization, as it provides a clue to pathogenesis.

As the authors point out, most patients with CIED infection present first to their primary care physician when they develop symptoms. An understanding of this infection by primary care physicians will allow for early recognition and more timely treatment, thus avoiding unnecessary complications.

Patients with pocket infection may not appear ill, but this should not lead a clinician away from the diagnosis. A pocket hematoma is an important differential diagnosis in the early postoperative period after device implantation or pulse generator change, and it may be difficult to decide if pocket changes are from an uninfected hematoma or from an infection.

Patients with endovascular infection are more likely to have systemic symptoms such as fever, fatigue, and malaise. However, absence of systemic features does not necessarily exclude endovascular infection.

BLOOD CULTURES AND TEE ARE KEY DIAGNOSTIC TESTS

All patients with suspected CIED infection should have at least two sets of blood cultures checked, even if they appear to be reasonably well. If there is any suspicion of endovascular infection, echocardiography should be performed.

Transesophageal echocardiography (TEE) is far superior to transthoracic echocardiography (TTE) for detecting lead vegetations.⁶ TEE should be carefully performed whenever endovascular infection is suspected, including all patients with positive blood cultures and all patients with systemic signs and symptoms.

Purulent drainage should be cultured, and when the device is removed, cultures of lead tips and pocket tissue should be done as well.

TREATMENT USUALLY REQUIRES COMPLETE DEVICE REMOVAL

A superficial infection in the early postoperative period may respond to antibiotic therapy alone. But in all other patients, the device must be removed to cure the infection. In referral centers, it is not unusual to see patients who have been referred after having been treated with antibiotics for weeks and sometimes months in the mistaken belief that the infection would be cured with antibiotics alone.

In some patients presenting with only pocket findings in the early postoperative period, it may be difficult indeed to tell if there is pocket infection. In such patients, it is not necessary to make a hasty decision to remove the device, but it is important to monitor them closely until the presence or absence of infection becomes clear. Also, erosion of the device through the skin represents pocket infection even if the patient appears otherwise healthy.

When removing the device, it is necessary to remove the generator and all leads to treat the infection effectively.

If patients are device-dependent, it is usually safe to place a new device with the new pulse generator pocket in a different location from the infected one a few days after the infected device is removed.

AREAS OF UNCERTAINTY AND CHALLENGE

Although there is no controversy about the need for complete removal of infected devices in order to effect a cure, the appropriate duration of antibiotic therapy after device removal is less clear. Dababneh and Sohail provide a useful algorithm to help with this decision. Patients usually need a new device to replace the infected one and there is a legitimate reason for concern about undertreating, since one would not want the new device to become infected because of inadequate antibiotic therapy. When endovascular infection is suspected or documented, patients are probably best treated as they would be for infective endocarditis.

Difficulties arise when patients with a CIED develop bacteremia with no echocardiographic evidence of device infection. Finding the source of bacteremia is very important because a diagnosis of CIED infection indicates that the device has to be removed. When there is a clear alternative explanation for the bacteremia, the CIED does not have to be removed. The type of bacterium helps clinicians to gauge the likelihood of CIED infection and to decide on the appropriate course of action. These cases should always be managed in conjunction with an infectious disease specialist and a cardiac electrophysiologist.

Another concern is secondary seeding of an uninfected CIED caused by bacteremia from another source. This concern is particularly acute with S aureus bacteremia. When patients with a CIED and S aureus bacteremia have been studied, endovascular CIED infection was documented in about half, although only a few had evidence of pocket inflammation.^7,8 This suggests that the devices were seeded via the endovascular route.

Medical procedures such as dialysis and total parenteral nutrition require frequent intravascular access—often facilitated by leaving an indwelling vascular catheter in place. Frequent entry into the intravascular compartment puts patients at substantial risk of bloodstream infection, and in patients with a CIED this can be complicated by device infection. In patients with a CIED and an indwelling vascular catheter who develop bacteremia, determining the source of the bacteremia is particularly challenging, as is the treatment. Thus, preventing endovascular infection in such patients is extremely desirable, but there are no easy solutions.

PLACING CIED INFECTIONS IN PERSPECTIVE

The vast majority of patients with a CIED never develop a device infection. Those unfortunate enough to have a CIED infection have little choice other than to have the device removed, but those diagnosed early and treated appropriately generally do well. The development of CIEDs has been an important advance in the practice of cardiac electrophysiology. An appropriate understanding of CIED infection and its treatment will help optimize the diagnosis and management of this complication when it does occur.

These days, an increasing number of people are receiving permanent pacemakers, implantable cardioverter-defibrillators, endovascular devices, and cardiac resynchronization therapy devices—collectively called cardiovascular implantable electronic devices (CIEDs). One reason for this upswing is that these devices have been approved for more indications, such as sick sinus syndrome, third-degree heart block, atrial fibrillation, life-threatening ventricular arrhythmias, survival of sudden cardiac death, and advanced congestive heart failure. Another reason is that the population is getting older, and therefore more people need these devices.

See related editorial

Although the use of a CIED is associated with a lower risk of death and a better quality of life, CIED-related infection can eclipse some of these benefits for their recipients. Historically reported rates of infections range from 0% to 19.9%.¹ However, recent data point to a disturbing trend: infection rates are rising faster than implantation rates.²

Besides causing morbidity and even death, infection is also associated with significant financial cost for patients and third-party payers. The estimated average cost of combined medical and surgical treatment of CIED-related infection ranges from $25,000 for permanent pacemakers to $50,000 for implantable cardioverter-defibrillators.^3,4

Although cardiologists and cardiac surgeons are the ones who implant these devices, most patients receive their routine outpatient care from a primary care physician, who can be a general internist, a family physician, or other specialist. Moreover, many patients with device infection are admitted to hospital internal medicine services for various diagnoses requiring inpatient care. Therefore, an internist, a family physician, or a hospitalist may be the first physician to respond to a suspected or confirmed device infection. Knowledge of the clinical manifestations and the initial steps in evaluation and management is essential for optimal care.

These complex infections pose challenges, which we will illustrate by presenting a case of CIED-related infection and reviewing key elements of diagnosis and management.

AN ILLUSTRATIVE CASE

A 60-year-old man had a permanent pacemaker implanted 3 months ago because of third-degree heart block; he now presents to his primary care physician with increasing pain, swelling, and erythema at the site of his pacemaker pocket. He has a history of type 2 diabetes mellitus, stage 3 chronic kidney disease, and coronary artery disease.

The symptoms started 2 weeks ago and have slowly progressed, prompting him to seek medical care. He is quite anxious and wants to know if he needs to arrange an emergency consultation with his cardiologist.

IMPORTANT CLINICAL QUESTIONS

This presentation raises several important questions:

• What should be the next step in his evaluation?
• Which laboratory tests should be done?
• Should he be admitted to the hospital, or can he be managed as an outpatient?
• Should he be started empirically on antibiotics? If so, which antibiotics? Or is it better to wait?
• When should an infectious disease specialist be consulted?
• Should the device be removed, and if so, all of it or which components?
• How long should antibiotics be given?

We will provide evidence-based answers to these questions in the discussions below.

PATHOGENESIS AND RISK FACTORS FOR DEVICE INFECTION

The first step in understanding the clinical manifestations of CIED-related infections is to grasp their pathogenesis. Risk factors for device infection have been evaluated in several studies.¹

Several factors interact in the inception and evolution of these infections, some related to the care in the perioperative period, some to the device, some to the host, and some to the causative microorganism.⁵ Although any one of these may play a predominant role in a given patient, most patients have a combination.

Perioperative factors that may contribute to a higher risk of infection include device revision; use of temporary pacing leads before placement of the permanent device; lack of antibiotic prophylaxis before implantation; longer operative time; operative inexperience; development of postoperative pocket hematoma; and factors such as diabetes mellitus and long-term use of corticosteroids and other immunosuppressive drugs that impair wound healing at the generator pocket.^6–11

Device factors. Abdominal generator placement, use of epicardial leads, and complexity of the device play a significant role.^6,12,13 In general, implantable cardioverter-defibrillators and cardiac resynchronization therapy devices have higher rates of infection than permanent pacemakers.^2,14

Host factors. Diseases and conditions that predispose to bloodstream infection may result in hematogenous seeding of the device and its leads and are associated with a higher risk of late-onset infection. These include an implanted central venous catheter (for hemodialysis or other long-term access), a distant focus of primary infection (such as pneumonia and skin and soft-tissue infections), and invasive procedures unrelated to the CIED.^10,15

In general, contamination at the time of surgery leads to early-onset infection (ie, within weeks to months of implantation), whereas hematogenous seeding is a predominant factor in most patients with late-onset infection.¹⁶

STAPHYLOCOCCI ARE THE MOST COMMON CAUSE

A key to making an accurate diagnosis and determining the appropriate empiric antibiotic therapy is to understand the microbiology of device infections.

Regardless of the clinical presentation, staphylococci are the predominant organisms responsible for both early- and late-onset infections.^17,18 These include Staphylococcus aureus and coagulase-negative staphylococci. Depending on where the implanting hospital is located and where the organism was acquired (in the community or in the hospital), up to 50% of these staphylococci may be methicillin-resistant,^17,18 a fact that necessitates using vancomycin for empiric coverage until the pathogen is identified and its susceptibility is known.

Gram-negative or polymicrobial CIED infections are infrequent. However, empiric gram-negative coverage should be considered for patients who present with systemic signs of infection, in whom delaying adequate coverage could jeopardize the successful outcome of infection treatment.

Fungal and mycobacterial infections of cardiac devices are exceedingly uncommon, mainly occurring in immunocompromised patients.

CLINICAL MANIFESTATIONS OF CARDIOVASCULAR DEVICE INFECTION

The clinical presentations of CIED-related infection can be broadly categorized into two groups: generator pocket infection and endovascular infection with an intact pocket.^17,18

Generator pocket infection

Most patients with a pocket infection present with inflammatory changes at the device generator site. Usual signs and symptoms include pain, erythema, swelling, and serosanguinous or purulent drainage from the pocket.

Patients with a pocket infection generally present within weeks to months of implantation, as the predominant mechanism of pocket infection is contamination of the generator or leads during implantation. However, occasionally, pocket infection caused by indolent organisms such as Propionibacterium, Corynebacterium, and certain species of coagulase-negative staphylococci can present more than 1 year after implantation. Hematogenous seeding of the device pocket, as a result of bacteremia from a distant primary focus, is infrequent except in cases of S aureus bloodstream infection.¹⁹

Endovascular infection with an intact pocket

A subset of patients with CIED-related infections, mostly late-onset infections, present only with systemic signs and symptoms without inflammatory changes at the generator pocket.^16–18 Most of these patients have multiple comorbid conditions and likely acquire the infection via hematogenous seeding of transvenous device leads from a distant focus of primary infection, such as a skin or soft-tissue infection, pneumonia, bacteremia arising from an implanted long-term central venous catheter, or bloodstream infection secondary to an invasive procedure unrelated to the CIED.

Most patients with an endovascular device infection have positive blood cultures at presentation. However, occasionally, blood cultures may be negative. The main reason for negative blood cultures in this setting is the use of empiric antibiotic therapy before blood cultures are drawn.

Endovascular device infections are further complicated by the formation of infected vegetations on the leads or cardiac valves in up to one-fourth of cases.^{16–18,20,21} This complication poses additional challenges in management, such as choosing the appropriate lead extraction technique, the waiting time before implanting a replacement device, and the optimal length of parenteral antimicrobial therapy. Many of these decisions are beyond the realm of internal medicine practice and are best managed by consultation with an infectious disease specialist and a cardiologist.

DIAGNOSIS OF INFECTION AND ASSOCIATED COMPLICATIONS

The clinical diagnosis of pocket infection is usually quite straightforward. However, occasionally, an early postoperative pocket hematoma can mimic pocket infection, and distinguishing these two may be difficult. Close collaboration between an internist, cardiologist, and infectious-disease specialist and careful observation of the patient may help to avoid a premature and incorrect diagnosis of pocket infection and unnecessary removal of the device in this scenario.

While diagnosing a pocket infection may be simple, an accurate and timely diagnosis of endovascular infection with an intact pocket can be challenging, especially if echocardiography shows no conclusive evidence of involvement of the device leads. Even when the infection is limited to the generator pocket, attempts to isolate causative pathogens may be hampered if empiric antibiotic therapy is started before culture samples are obtained from the pocket and from the blood.

Complete blood count with differential cell count.

Electrolyte and serum creatinine concentrations.

Inflammatory markers, including erythrocyte sedimentation rate and C-reactive protein concentration.

Swabs for bacterial cultures should be sent if there is purulent drainage from the generator pocket. This can be done in the office before referral to the emergency department or a tertiary care center for inpatient admission. If the pocket appears swollen or fluctuant, needle aspiration should be avoided, as it can introduce organisms and cause contamination.⁵

Two sets of peripheral blood cultures should be obtained. If the patient has an implanted central venous catheter, blood cultures via each catheter port should also be obtained, as they may help to pinpoint the source of bloodstream infection in cases in which blood culture results are positive.

TEE should also be performed in patients with systemic signs and symptoms (such as fever, chills, malaise, dyspnea, hypotension, or peripheral stigmata of endocarditis) or abnormal test results (leukocytosis, elevated inflammatory markers, or evidence of pulmonary emboli on imaging), even if blood cultures are negative. Similarly, TEE should also be considered in patients in whom blood cultures may be negative as a result of previous antimicrobial therapy.

If a decision is made to remove the device (see below), intraoperative pocket tissue and lead-tip cultures should be sent for Gram staining and bacterial culture. Fungal and mycobacterial cultures may be necessary in immunocompromised hosts, or if Gram staining and bacterial cultures from pocket tissue samples are negative. Caution must be exercised when interpreting the results of lead-tip cultures, as lead tips may become contaminated while being pulled through an infected pocket during removal.^20,22

This approach should lead to an accurate diagnosis of CIED-related infection and associated complications in most patients. However, the diagnosis may remain elusive if results of blood cultures are positive but the pocket is intact and there is no echocardiographic evidence of lead or valve involvement. This is especially true in cases of S aureus bacteremia, in which positive blood cultures may be the sole manifestation of underlying device infection.^19,23 Factors associated with higher odds of underlying device infection in this scenario include bacteremia lasting more than 24 hours, prosthetic valves, bacteremia within 3 months of device implantation, and no alternative focus of bacteremia.¹²

Evidence is emerging that underlying device infection should also be considered in patients with bloodstream infection with coagulase-negative staphylococci in the setting of an implanted device.²⁴ On the other hand, seeding of device leads with gram-negative organisms is infrequent, and routine imaging of intracardiac leads is not necessary in cases of gram-negative bacteremia.²⁵

In our opinion, cases of bacteremia in which underlying occult device infection is a concern are best managed by consultation with an infectious disease specialist.

A STEPWISE APPROACH TO MANAGING DEVICE INFECTION

Should antibiotics be started empirically?

The first step in managing CIED-related infection is to decide whether empiric antibiotic therapy should be started immediately once infection is suspected or if it is prudent to wait until the culture results are available.

In our opinion, if the infection is limited to the generator pocket, it is reasonable to wait until immediately before surgery to maximize the culture yield from pocket tissue samples. An exception to this rule is when systemic signs or symptoms are present, in which case delaying antibiotic therapy could jeopardize the outcome (FIGURE 2). In such cases, empiric antibiotic therapy can be started once two sets of peripheral blood samples for cultures have been obtained.

Which antibiotics should be given empirically?

Because gram-positive organisms, namely coagulase-negative staphylococci and S aureus, are the causative pathogens in most cases of CIED-related infection, empiric antibiotic therapy should provide adequate coverage for these organisms. Because methicillin resistance is quite prevalent in staphylococci, we routinely use vancomycin (Vancocin) for empiric coverage. In patients who are allergic to vancomycin or cannot tolerate it, daptomycin (Cubicin) is an alternative.

Empiric gram-negative coverage is generally reserved for patients who present with systemic signs and symptoms, in whom delaying adequate coverage could have untoward consequences. We routinely use cefepime (Maxipime) for empiric gram-negative coverage in our institution. Other beta-lactam agents that provide coverage for gram-negative bacilli, especially Pseudomonas, are also appropriate in this setting.

Should the device be removed?

Superficial infection of the wound or incision site (eg, stitch abscess) early after implantation can be managed by conservative antibiotic therapy without removing the device. However, complete removal of the device system, including intracardiac leads, is necessary in all other presentations of device infection, even if the infection appears limited to the generator pocket.^5,12 Leaving the device in place or removing parts of the device is associated with persistent or relapsed infection and is not advisable.^17,26

Leaving the device in place may be necessary in extenuating circumstances, eg, if surgery would be too risky for the patient or if the patient refuses device removal or has a short life expectancy. In these cases, lifelong suppressive antibiotic therapy should be prescribed after an initial course of parenteral antibiotics.²⁷ Antibiotic choices for long-term suppressive therapy should be guided by antimicrobial susceptibility testing and consultation with an infectious disease specialist.

How should the leads be removed?

Leads are extracted percutaneously in most cases. Percutaneous extraction is generally considered safe even in cases in which infection is complicated by lead vegetations, which raises concern about pulmonary embolization of detached vegetation fragments during extraction.^5,20

Thoracotomy is generally reserved for patients who have cardiac complications (such as a cardiac abscess or the need to replace cardiac valves) or in whom attempts to extract the leads percutaneously are unsuccessful.

Details of the removal procedure and choice of extraction technique are beyond the scope of this paper and are best left to the discretion of the treating cardiologist or cardiac surgeon. Because of the potential for complications during percutaneous device removal, such as laceration of the superior vena cava or cardiac tamponade, the patient should be referred to a high-volume center where cardiothoracic intervention can be provided on an emergency basis if needed.

How long should antibiotic therapy go on?

An algorithm for deciding the duration of antibiotic therapy is shown in Figure 3. These guidelines, first published in 2007,¹⁷ were adopted by the American Heart Association in its updated statement on the management of CIED-related infections.⁵ However, it should be noted that these guidelines are not based on randomized clinical trials; rather, they represent expert opinion based on published series of patients with CIED-related infections.

In general, cases of device erosion or pocket infection can be treated with 1 to 2 weeks of appropriate antibiotic therapy based on antimicrobial susceptibility testing. However, cases of bloodstream infection require 2 to 4 weeks of antibiotic therapy—or sometimes even longer if associated complications are present, such as septic thrombosis, endocarditis, or osteomyelitis.

We favor parenteral antibiotics for the entire course of treatment. However, patients can be discharged from the hospital once the bloodstream infection has cleared, and the antibiotic course can be completed on an outpatient basis.

Outpatient antimicrobial monitoring

We recommend adherence to the Infectious Diseases Society of America’s guidelines for monitoring outpatient parenteral antimicrobial therapy.²⁸

At discharge from the hospital, patients should be instructed to promptly call their primary care physician if they have a fever or notice inflammatory changes at the pocket site. If the patient reports such symptoms, repeat blood cultures should be ordered, and the patient should be monitored closely for signs of a relapse of infection.

A routine follow-up visit should be arranged at 2 weeks and at the end of parenteral antibiotic therapy (for patients receiving therapy for 4 weeks or longer) to make sure the infection has resolved.

When should a new device be implanted?

Before deciding when a new device should be implanted, one should carefully assess whether the patient still needs one. Studies indicate that up to 30% of patients may no longer require a cardiac device.^17,18

However, we believe that removal of drains and closure of the old pocket are not necessary before implanting a new device in a different location (usually the contralateral pectoral area). Exceptions to this general principle are cases of valvular endocarditis, in which a minimum of 2 weeks is recommended between removal of an infected device (plus clearance of bloodstream infection) and implantation of a new device.

OUTCOMES OF INFECTION

Despite improvements in our understanding of how to manage CIED-related infection, the rates of morbidity and death remain significant.

The outcome, in part, depends on the clinical presentation and the patient’s comorbid conditions. In general, the death rate in patients with a pocket infection is less than 5%. However, in patients with endovascular infection, it may be as high as 20%.^16–18 Other factors that affect the outcome include complications such as septic thrombosis, valvular endocarditis, or osteomyelitis; complications during device extraction; the need for open heart surgery; and the overall health of the patient.

Complete removal of the device system is a requisite for successful outcome, and the risk of death tends to be higher if only part of the infected CIED system is extracted.²⁶

STRATEGIES TO PREVENT DEVICE INFECTION

Preventive efforts should focus on strategies to minimize the chances of contamination of the generator, leads, and pocket during implantation.²⁹ Patients who are known to be colonized with methicillin-resistant S aureus may benefit from decolonization programs, which should include nasal application of mupirocin (Bactroban) ointment preoperatively.³⁰ In addition, use of chlorhexidine for surgical-site antisepsis has been shown to reduce the risk of surgical site infection.³¹

Moreover, all patients should receive antibiotic prophylaxis before implantation of a CIED.^32,33 Most institutions use a first-generation cephalosporin, such as cefazolin (Ancef), for this purpose.³⁴ However, the increasing rate of methicillin resistance in staphylococci has led to the routine use of vancomycin for preoperative prophylaxis at some centers.¹⁸

Regardless of the antibiotic chosen for prophylaxis, protocols that ensure that all patients receive an appropriate antibiotic at the appropriate time are a key determinant in the success of these infection-control programs.

These days, an increasing number of people are receiving permanent pacemakers, implantable cardioverter-defibrillators, endovascular devices, and cardiac resynchronization therapy devices—collectively called cardiovascular implantable electronic devices (CIEDs). One reason for this upswing is that these devices have been approved for more indications, such as sick sinus syndrome, third-degree heart block, atrial fibrillation, life-threatening ventricular arrhythmias, survival of sudden cardiac death, and advanced congestive heart failure. Another reason is that the population is getting older, and therefore more people need these devices.

See related editorial

Although the use of a CIED is associated with a lower risk of death and a better quality of life, CIED-related infection can eclipse some of these benefits for their recipients. Historically reported rates of infections range from 0% to 19.9%.¹ However, recent data point to a disturbing trend: infection rates are rising faster than implantation rates.²

Besides causing morbidity and even death, infection is also associated with significant financial cost for patients and third-party payers. The estimated average cost of combined medical and surgical treatment of CIED-related infection ranges from $25,000 for permanent pacemakers to $50,000 for implantable cardioverter-defibrillators.^3,4

Although cardiologists and cardiac surgeons are the ones who implant these devices, most patients receive their routine outpatient care from a primary care physician, who can be a general internist, a family physician, or other specialist. Moreover, many patients with device infection are admitted to hospital internal medicine services for various diagnoses requiring inpatient care. Therefore, an internist, a family physician, or a hospitalist may be the first physician to respond to a suspected or confirmed device infection. Knowledge of the clinical manifestations and the initial steps in evaluation and management is essential for optimal care.

These complex infections pose challenges, which we will illustrate by presenting a case of CIED-related infection and reviewing key elements of diagnosis and management.

AN ILLUSTRATIVE CASE

A 60-year-old man had a permanent pacemaker implanted 3 months ago because of third-degree heart block; he now presents to his primary care physician with increasing pain, swelling, and erythema at the site of his pacemaker pocket. He has a history of type 2 diabetes mellitus, stage 3 chronic kidney disease, and coronary artery disease.

The symptoms started 2 weeks ago and have slowly progressed, prompting him to seek medical care. He is quite anxious and wants to know if he needs to arrange an emergency consultation with his cardiologist.

IMPORTANT CLINICAL QUESTIONS

This presentation raises several important questions:

• What should be the next step in his evaluation?
• Which laboratory tests should be done?
• Should he be admitted to the hospital, or can he be managed as an outpatient?
• Should he be started empirically on antibiotics? If so, which antibiotics? Or is it better to wait?
• When should an infectious disease specialist be consulted?
• Should the device be removed, and if so, all of it or which components?
• How long should antibiotics be given?

We will provide evidence-based answers to these questions in the discussions below.

PATHOGENESIS AND RISK FACTORS FOR DEVICE INFECTION

The first step in understanding the clinical manifestations of CIED-related infections is to grasp their pathogenesis. Risk factors for device infection have been evaluated in several studies.¹

Several factors interact in the inception and evolution of these infections, some related to the care in the perioperative period, some to the device, some to the host, and some to the causative microorganism.⁵ Although any one of these may play a predominant role in a given patient, most patients have a combination.

Perioperative factors that may contribute to a higher risk of infection include device revision; use of temporary pacing leads before placement of the permanent device; lack of antibiotic prophylaxis before implantation; longer operative time; operative inexperience; development of postoperative pocket hematoma; and factors such as diabetes mellitus and long-term use of corticosteroids and other immunosuppressive drugs that impair wound healing at the generator pocket.^6–11

Device factors. Abdominal generator placement, use of epicardial leads, and complexity of the device play a significant role.^6,12,13 In general, implantable cardioverter-defibrillators and cardiac resynchronization therapy devices have higher rates of infection than permanent pacemakers.^2,14

Host factors. Diseases and conditions that predispose to bloodstream infection may result in hematogenous seeding of the device and its leads and are associated with a higher risk of late-onset infection. These include an implanted central venous catheter (for hemodialysis or other long-term access), a distant focus of primary infection (such as pneumonia and skin and soft-tissue infections), and invasive procedures unrelated to the CIED.^10,15

In general, contamination at the time of surgery leads to early-onset infection (ie, within weeks to months of implantation), whereas hematogenous seeding is a predominant factor in most patients with late-onset infection.¹⁶

STAPHYLOCOCCI ARE THE MOST COMMON CAUSE

A key to making an accurate diagnosis and determining the appropriate empiric antibiotic therapy is to understand the microbiology of device infections.

Regardless of the clinical presentation, staphylococci are the predominant organisms responsible for both early- and late-onset infections.^17,18 These include Staphylococcus aureus and coagulase-negative staphylococci. Depending on where the implanting hospital is located and where the organism was acquired (in the community or in the hospital), up to 50% of these staphylococci may be methicillin-resistant,^17,18 a fact that necessitates using vancomycin for empiric coverage until the pathogen is identified and its susceptibility is known.

Gram-negative or polymicrobial CIED infections are infrequent. However, empiric gram-negative coverage should be considered for patients who present with systemic signs of infection, in whom delaying adequate coverage could jeopardize the successful outcome of infection treatment.

Fungal and mycobacterial infections of cardiac devices are exceedingly uncommon, mainly occurring in immunocompromised patients.

CLINICAL MANIFESTATIONS OF CARDIOVASCULAR DEVICE INFECTION

The clinical presentations of CIED-related infection can be broadly categorized into two groups: generator pocket infection and endovascular infection with an intact pocket.^17,18

Generator pocket infection

Most patients with a pocket infection present with inflammatory changes at the device generator site. Usual signs and symptoms include pain, erythema, swelling, and serosanguinous or purulent drainage from the pocket.

Patients with a pocket infection generally present within weeks to months of implantation, as the predominant mechanism of pocket infection is contamination of the generator or leads during implantation. However, occasionally, pocket infection caused by indolent organisms such as Propionibacterium, Corynebacterium, and certain species of coagulase-negative staphylococci can present more than 1 year after implantation. Hematogenous seeding of the device pocket, as a result of bacteremia from a distant primary focus, is infrequent except in cases of S aureus bloodstream infection.¹⁹

Endovascular infection with an intact pocket

A subset of patients with CIED-related infections, mostly late-onset infections, present only with systemic signs and symptoms without inflammatory changes at the generator pocket.^16–18 Most of these patients have multiple comorbid conditions and likely acquire the infection via hematogenous seeding of transvenous device leads from a distant focus of primary infection, such as a skin or soft-tissue infection, pneumonia, bacteremia arising from an implanted long-term central venous catheter, or bloodstream infection secondary to an invasive procedure unrelated to the CIED.

Most patients with an endovascular device infection have positive blood cultures at presentation. However, occasionally, blood cultures may be negative. The main reason for negative blood cultures in this setting is the use of empiric antibiotic therapy before blood cultures are drawn.

Endovascular device infections are further complicated by the formation of infected vegetations on the leads or cardiac valves in up to one-fourth of cases.^{16–18,20,21} This complication poses additional challenges in management, such as choosing the appropriate lead extraction technique, the waiting time before implanting a replacement device, and the optimal length of parenteral antimicrobial therapy. Many of these decisions are beyond the realm of internal medicine practice and are best managed by consultation with an infectious disease specialist and a cardiologist.

DIAGNOSIS OF INFECTION AND ASSOCIATED COMPLICATIONS

The clinical diagnosis of pocket infection is usually quite straightforward. However, occasionally, an early postoperative pocket hematoma can mimic pocket infection, and distinguishing these two may be difficult. Close collaboration between an internist, cardiologist, and infectious-disease specialist and careful observation of the patient may help to avoid a premature and incorrect diagnosis of pocket infection and unnecessary removal of the device in this scenario.

While diagnosing a pocket infection may be simple, an accurate and timely diagnosis of endovascular infection with an intact pocket can be challenging, especially if echocardiography shows no conclusive evidence of involvement of the device leads. Even when the infection is limited to the generator pocket, attempts to isolate causative pathogens may be hampered if empiric antibiotic therapy is started before culture samples are obtained from the pocket and from the blood.

Complete blood count with differential cell count.

Electrolyte and serum creatinine concentrations.

Inflammatory markers, including erythrocyte sedimentation rate and C-reactive protein concentration.

Swabs for bacterial cultures should be sent if there is purulent drainage from the generator pocket. This can be done in the office before referral to the emergency department or a tertiary care center for inpatient admission. If the pocket appears swollen or fluctuant, needle aspiration should be avoided, as it can introduce organisms and cause contamination.⁵

Two sets of peripheral blood cultures should be obtained. If the patient has an implanted central venous catheter, blood cultures via each catheter port should also be obtained, as they may help to pinpoint the source of bloodstream infection in cases in which blood culture results are positive.

TEE should also be performed in patients with systemic signs and symptoms (such as fever, chills, malaise, dyspnea, hypotension, or peripheral stigmata of endocarditis) or abnormal test results (leukocytosis, elevated inflammatory markers, or evidence of pulmonary emboli on imaging), even if blood cultures are negative. Similarly, TEE should also be considered in patients in whom blood cultures may be negative as a result of previous antimicrobial therapy.

If a decision is made to remove the device (see below), intraoperative pocket tissue and lead-tip cultures should be sent for Gram staining and bacterial culture. Fungal and mycobacterial cultures may be necessary in immunocompromised hosts, or if Gram staining and bacterial cultures from pocket tissue samples are negative. Caution must be exercised when interpreting the results of lead-tip cultures, as lead tips may become contaminated while being pulled through an infected pocket during removal.^20,22

This approach should lead to an accurate diagnosis of CIED-related infection and associated complications in most patients. However, the diagnosis may remain elusive if results of blood cultures are positive but the pocket is intact and there is no echocardiographic evidence of lead or valve involvement. This is especially true in cases of S aureus bacteremia, in which positive blood cultures may be the sole manifestation of underlying device infection.^19,23 Factors associated with higher odds of underlying device infection in this scenario include bacteremia lasting more than 24 hours, prosthetic valves, bacteremia within 3 months of device implantation, and no alternative focus of bacteremia.¹²

Evidence is emerging that underlying device infection should also be considered in patients with bloodstream infection with coagulase-negative staphylococci in the setting of an implanted device.²⁴ On the other hand, seeding of device leads with gram-negative organisms is infrequent, and routine imaging of intracardiac leads is not necessary in cases of gram-negative bacteremia.²⁵

In our opinion, cases of bacteremia in which underlying occult device infection is a concern are best managed by consultation with an infectious disease specialist.

A STEPWISE APPROACH TO MANAGING DEVICE INFECTION

Should antibiotics be started empirically?

The first step in managing CIED-related infection is to decide whether empiric antibiotic therapy should be started immediately once infection is suspected or if it is prudent to wait until the culture results are available.

In our opinion, if the infection is limited to the generator pocket, it is reasonable to wait until immediately before surgery to maximize the culture yield from pocket tissue samples. An exception to this rule is when systemic signs or symptoms are present, in which case delaying antibiotic therapy could jeopardize the outcome (FIGURE 2). In such cases, empiric antibiotic therapy can be started once two sets of peripheral blood samples for cultures have been obtained.

Which antibiotics should be given empirically?

Because gram-positive organisms, namely coagulase-negative staphylococci and S aureus, are the causative pathogens in most cases of CIED-related infection, empiric antibiotic therapy should provide adequate coverage for these organisms. Because methicillin resistance is quite prevalent in staphylococci, we routinely use vancomycin (Vancocin) for empiric coverage. In patients who are allergic to vancomycin or cannot tolerate it, daptomycin (Cubicin) is an alternative.

Empiric gram-negative coverage is generally reserved for patients who present with systemic signs and symptoms, in whom delaying adequate coverage could have untoward consequences. We routinely use cefepime (Maxipime) for empiric gram-negative coverage in our institution. Other beta-lactam agents that provide coverage for gram-negative bacilli, especially Pseudomonas, are also appropriate in this setting.

Should the device be removed?

Superficial infection of the wound or incision site (eg, stitch abscess) early after implantation can be managed by conservative antibiotic therapy without removing the device. However, complete removal of the device system, including intracardiac leads, is necessary in all other presentations of device infection, even if the infection appears limited to the generator pocket.^5,12 Leaving the device in place or removing parts of the device is associated with persistent or relapsed infection and is not advisable.^17,26

Leaving the device in place may be necessary in extenuating circumstances, eg, if surgery would be too risky for the patient or if the patient refuses device removal or has a short life expectancy. In these cases, lifelong suppressive antibiotic therapy should be prescribed after an initial course of parenteral antibiotics.²⁷ Antibiotic choices for long-term suppressive therapy should be guided by antimicrobial susceptibility testing and consultation with an infectious disease specialist.

How should the leads be removed?

Leads are extracted percutaneously in most cases. Percutaneous extraction is generally considered safe even in cases in which infection is complicated by lead vegetations, which raises concern about pulmonary embolization of detached vegetation fragments during extraction.^5,20

Thoracotomy is generally reserved for patients who have cardiac complications (such as a cardiac abscess or the need to replace cardiac valves) or in whom attempts to extract the leads percutaneously are unsuccessful.

Details of the removal procedure and choice of extraction technique are beyond the scope of this paper and are best left to the discretion of the treating cardiologist or cardiac surgeon. Because of the potential for complications during percutaneous device removal, such as laceration of the superior vena cava or cardiac tamponade, the patient should be referred to a high-volume center where cardiothoracic intervention can be provided on an emergency basis if needed.

How long should antibiotic therapy go on?

An algorithm for deciding the duration of antibiotic therapy is shown in Figure 3. These guidelines, first published in 2007,¹⁷ were adopted by the American Heart Association in its updated statement on the management of CIED-related infections.⁵ However, it should be noted that these guidelines are not based on randomized clinical trials; rather, they represent expert opinion based on published series of patients with CIED-related infections.

In general, cases of device erosion or pocket infection can be treated with 1 to 2 weeks of appropriate antibiotic therapy based on antimicrobial susceptibility testing. However, cases of bloodstream infection require 2 to 4 weeks of antibiotic therapy—or sometimes even longer if associated complications are present, such as septic thrombosis, endocarditis, or osteomyelitis.

We favor parenteral antibiotics for the entire course of treatment. However, patients can be discharged from the hospital once the bloodstream infection has cleared, and the antibiotic course can be completed on an outpatient basis.

Outpatient antimicrobial monitoring

We recommend adherence to the Infectious Diseases Society of America’s guidelines for monitoring outpatient parenteral antimicrobial therapy.²⁸

At discharge from the hospital, patients should be instructed to promptly call their primary care physician if they have a fever or notice inflammatory changes at the pocket site. If the patient reports such symptoms, repeat blood cultures should be ordered, and the patient should be monitored closely for signs of a relapse of infection.

A routine follow-up visit should be arranged at 2 weeks and at the end of parenteral antibiotic therapy (for patients receiving therapy for 4 weeks or longer) to make sure the infection has resolved.

When should a new device be implanted?

Before deciding when a new device should be implanted, one should carefully assess whether the patient still needs one. Studies indicate that up to 30% of patients may no longer require a cardiac device.^17,18

However, we believe that removal of drains and closure of the old pocket are not necessary before implanting a new device in a different location (usually the contralateral pectoral area). Exceptions to this general principle are cases of valvular endocarditis, in which a minimum of 2 weeks is recommended between removal of an infected device (plus clearance of bloodstream infection) and implantation of a new device.

OUTCOMES OF INFECTION

Despite improvements in our understanding of how to manage CIED-related infection, the rates of morbidity and death remain significant.

The outcome, in part, depends on the clinical presentation and the patient’s comorbid conditions. In general, the death rate in patients with a pocket infection is less than 5%. However, in patients with endovascular infection, it may be as high as 20%.^16–18 Other factors that affect the outcome include complications such as septic thrombosis, valvular endocarditis, or osteomyelitis; complications during device extraction; the need for open heart surgery; and the overall health of the patient.

Complete removal of the device system is a requisite for successful outcome, and the risk of death tends to be higher if only part of the infected CIED system is extracted.²⁶

STRATEGIES TO PREVENT DEVICE INFECTION

Preventive efforts should focus on strategies to minimize the chances of contamination of the generator, leads, and pocket during implantation.²⁹ Patients who are known to be colonized with methicillin-resistant S aureus may benefit from decolonization programs, which should include nasal application of mupirocin (Bactroban) ointment preoperatively.³⁰ In addition, use of chlorhexidine for surgical-site antisepsis has been shown to reduce the risk of surgical site infection.³¹

Moreover, all patients should receive antibiotic prophylaxis before implantation of a CIED.^32,33 Most institutions use a first-generation cephalosporin, such as cefazolin (Ancef), for this purpose.³⁴ However, the increasing rate of methicillin resistance in staphylococci has led to the routine use of vancomycin for preoperative prophylaxis at some centers.¹⁸

Regardless of the antibiotic chosen for prophylaxis, protocols that ensure that all patients receive an appropriate antibiotic at the appropriate time are a key determinant in the success of these infection-control programs.

These days, an increasing number of people are receiving permanent pacemakers, implantable cardioverter-defibrillators, endovascular devices, and cardiac resynchronization therapy devices—collectively called cardiovascular implantable electronic devices (CIEDs). One reason for this upswing is that these devices have been approved for more indications, such as sick sinus syndrome, third-degree heart block, atrial fibrillation, life-threatening ventricular arrhythmias, survival of sudden cardiac death, and advanced congestive heart failure. Another reason is that the population is getting older, and therefore more people need these devices.

See related editorial

Although the use of a CIED is associated with a lower risk of death and a better quality of life, CIED-related infection can eclipse some of these benefits for their recipients. Historically reported rates of infections range from 0% to 19.9%.¹ However, recent data point to a disturbing trend: infection rates are rising faster than implantation rates.²

Besides causing morbidity and even death, infection is also associated with significant financial cost for patients and third-party payers. The estimated average cost of combined medical and surgical treatment of CIED-related infection ranges from $25,000 for permanent pacemakers to $50,000 for implantable cardioverter-defibrillators.^3,4

Although cardiologists and cardiac surgeons are the ones who implant these devices, most patients receive their routine outpatient care from a primary care physician, who can be a general internist, a family physician, or other specialist. Moreover, many patients with device infection are admitted to hospital internal medicine services for various diagnoses requiring inpatient care. Therefore, an internist, a family physician, or a hospitalist may be the first physician to respond to a suspected or confirmed device infection. Knowledge of the clinical manifestations and the initial steps in evaluation and management is essential for optimal care.

These complex infections pose challenges, which we will illustrate by presenting a case of CIED-related infection and reviewing key elements of diagnosis and management.

AN ILLUSTRATIVE CASE

A 60-year-old man had a permanent pacemaker implanted 3 months ago because of third-degree heart block; he now presents to his primary care physician with increasing pain, swelling, and erythema at the site of his pacemaker pocket. He has a history of type 2 diabetes mellitus, stage 3 chronic kidney disease, and coronary artery disease.

The symptoms started 2 weeks ago and have slowly progressed, prompting him to seek medical care. He is quite anxious and wants to know if he needs to arrange an emergency consultation with his cardiologist.

IMPORTANT CLINICAL QUESTIONS

This presentation raises several important questions:

• What should be the next step in his evaluation?
• Which laboratory tests should be done?
• Should he be admitted to the hospital, or can he be managed as an outpatient?
• Should he be started empirically on antibiotics? If so, which antibiotics? Or is it better to wait?
• When should an infectious disease specialist be consulted?
• Should the device be removed, and if so, all of it or which components?
• How long should antibiotics be given?

We will provide evidence-based answers to these questions in the discussions below.

PATHOGENESIS AND RISK FACTORS FOR DEVICE INFECTION

The first step in understanding the clinical manifestations of CIED-related infections is to grasp their pathogenesis. Risk factors for device infection have been evaluated in several studies.¹

Several factors interact in the inception and evolution of these infections, some related to the care in the perioperative period, some to the device, some to the host, and some to the causative microorganism.⁵ Although any one of these may play a predominant role in a given patient, most patients have a combination.

Perioperative factors that may contribute to a higher risk of infection include device revision; use of temporary pacing leads before placement of the permanent device; lack of antibiotic prophylaxis before implantation; longer operative time; operative inexperience; development of postoperative pocket hematoma; and factors such as diabetes mellitus and long-term use of corticosteroids and other immunosuppressive drugs that impair wound healing at the generator pocket.^6–11

Device factors. Abdominal generator placement, use of epicardial leads, and complexity of the device play a significant role.^6,12,13 In general, implantable cardioverter-defibrillators and cardiac resynchronization therapy devices have higher rates of infection than permanent pacemakers.^2,14

Host factors. Diseases and conditions that predispose to bloodstream infection may result in hematogenous seeding of the device and its leads and are associated with a higher risk of late-onset infection. These include an implanted central venous catheter (for hemodialysis or other long-term access), a distant focus of primary infection (such as pneumonia and skin and soft-tissue infections), and invasive procedures unrelated to the CIED.^10,15

In general, contamination at the time of surgery leads to early-onset infection (ie, within weeks to months of implantation), whereas hematogenous seeding is a predominant factor in most patients with late-onset infection.¹⁶

STAPHYLOCOCCI ARE THE MOST COMMON CAUSE

A key to making an accurate diagnosis and determining the appropriate empiric antibiotic therapy is to understand the microbiology of device infections.

Regardless of the clinical presentation, staphylococci are the predominant organisms responsible for both early- and late-onset infections.^17,18 These include Staphylococcus aureus and coagulase-negative staphylococci. Depending on where the implanting hospital is located and where the organism was acquired (in the community or in the hospital), up to 50% of these staphylococci may be methicillin-resistant,^17,18 a fact that necessitates using vancomycin for empiric coverage until the pathogen is identified and its susceptibility is known.

Gram-negative or polymicrobial CIED infections are infrequent. However, empiric gram-negative coverage should be considered for patients who present with systemic signs of infection, in whom delaying adequate coverage could jeopardize the successful outcome of infection treatment.

Fungal and mycobacterial infections of cardiac devices are exceedingly uncommon, mainly occurring in immunocompromised patients.

CLINICAL MANIFESTATIONS OF CARDIOVASCULAR DEVICE INFECTION

The clinical presentations of CIED-related infection can be broadly categorized into two groups: generator pocket infection and endovascular infection with an intact pocket.^17,18

Generator pocket infection

Most patients with a pocket infection present with inflammatory changes at the device generator site. Usual signs and symptoms include pain, erythema, swelling, and serosanguinous or purulent drainage from the pocket.

Patients with a pocket infection generally present within weeks to months of implantation, as the predominant mechanism of pocket infection is contamination of the generator or leads during implantation. However, occasionally, pocket infection caused by indolent organisms such as Propionibacterium, Corynebacterium, and certain species of coagulase-negative staphylococci can present more than 1 year after implantation. Hematogenous seeding of the device pocket, as a result of bacteremia from a distant primary focus, is infrequent except in cases of S aureus bloodstream infection.¹⁹

Endovascular infection with an intact pocket

A subset of patients with CIED-related infections, mostly late-onset infections, present only with systemic signs and symptoms without inflammatory changes at the generator pocket.^16–18 Most of these patients have multiple comorbid conditions and likely acquire the infection via hematogenous seeding of transvenous device leads from a distant focus of primary infection, such as a skin or soft-tissue infection, pneumonia, bacteremia arising from an implanted long-term central venous catheter, or bloodstream infection secondary to an invasive procedure unrelated to the CIED.

Most patients with an endovascular device infection have positive blood cultures at presentation. However, occasionally, blood cultures may be negative. The main reason for negative blood cultures in this setting is the use of empiric antibiotic therapy before blood cultures are drawn.

Endovascular device infections are further complicated by the formation of infected vegetations on the leads or cardiac valves in up to one-fourth of cases.^{16–18,20,21} This complication poses additional challenges in management, such as choosing the appropriate lead extraction technique, the waiting time before implanting a replacement device, and the optimal length of parenteral antimicrobial therapy. Many of these decisions are beyond the realm of internal medicine practice and are best managed by consultation with an infectious disease specialist and a cardiologist.

DIAGNOSIS OF INFECTION AND ASSOCIATED COMPLICATIONS

The clinical diagnosis of pocket infection is usually quite straightforward. However, occasionally, an early postoperative pocket hematoma can mimic pocket infection, and distinguishing these two may be difficult. Close collaboration between an internist, cardiologist, and infectious-disease specialist and careful observation of the patient may help to avoid a premature and incorrect diagnosis of pocket infection and unnecessary removal of the device in this scenario.

While diagnosing a pocket infection may be simple, an accurate and timely diagnosis of endovascular infection with an intact pocket can be challenging, especially if echocardiography shows no conclusive evidence of involvement of the device leads. Even when the infection is limited to the generator pocket, attempts to isolate causative pathogens may be hampered if empiric antibiotic therapy is started before culture samples are obtained from the pocket and from the blood.

Complete blood count with differential cell count.

Electrolyte and serum creatinine concentrations.

Inflammatory markers, including erythrocyte sedimentation rate and C-reactive protein concentration.

Swabs for bacterial cultures should be sent if there is purulent drainage from the generator pocket. This can be done in the office before referral to the emergency department or a tertiary care center for inpatient admission. If the pocket appears swollen or fluctuant, needle aspiration should be avoided, as it can introduce organisms and cause contamination.⁵

Two sets of peripheral blood cultures should be obtained. If the patient has an implanted central venous catheter, blood cultures via each catheter port should also be obtained, as they may help to pinpoint the source of bloodstream infection in cases in which blood culture results are positive.

TEE should also be performed in patients with systemic signs and symptoms (such as fever, chills, malaise, dyspnea, hypotension, or peripheral stigmata of endocarditis) or abnormal test results (leukocytosis, elevated inflammatory markers, or evidence of pulmonary emboli on imaging), even if blood cultures are negative. Similarly, TEE should also be considered in patients in whom blood cultures may be negative as a result of previous antimicrobial therapy.

If a decision is made to remove the device (see below), intraoperative pocket tissue and lead-tip cultures should be sent for Gram staining and bacterial culture. Fungal and mycobacterial cultures may be necessary in immunocompromised hosts, or if Gram staining and bacterial cultures from pocket tissue samples are negative. Caution must be exercised when interpreting the results of lead-tip cultures, as lead tips may become contaminated while being pulled through an infected pocket during removal.^20,22

This approach should lead to an accurate diagnosis of CIED-related infection and associated complications in most patients. However, the diagnosis may remain elusive if results of blood cultures are positive but the pocket is intact and there is no echocardiographic evidence of lead or valve involvement. This is especially true in cases of S aureus bacteremia, in which positive blood cultures may be the sole manifestation of underlying device infection.^19,23 Factors associated with higher odds of underlying device infection in this scenario include bacteremia lasting more than 24 hours, prosthetic valves, bacteremia within 3 months of device implantation, and no alternative focus of bacteremia.¹²

Evidence is emerging that underlying device infection should also be considered in patients with bloodstream infection with coagulase-negative staphylococci in the setting of an implanted device.²⁴ On the other hand, seeding of device leads with gram-negative organisms is infrequent, and routine imaging of intracardiac leads is not necessary in cases of gram-negative bacteremia.²⁵

In our opinion, cases of bacteremia in which underlying occult device infection is a concern are best managed by consultation with an infectious disease specialist.

A STEPWISE APPROACH TO MANAGING DEVICE INFECTION

Should antibiotics be started empirically?

The first step in managing CIED-related infection is to decide whether empiric antibiotic therapy should be started immediately once infection is suspected or if it is prudent to wait until the culture results are available.

In our opinion, if the infection is limited to the generator pocket, it is reasonable to wait until immediately before surgery to maximize the culture yield from pocket tissue samples. An exception to this rule is when systemic signs or symptoms are present, in which case delaying antibiotic therapy could jeopardize the outcome (FIGURE 2). In such cases, empiric antibiotic therapy can be started once two sets of peripheral blood samples for cultures have been obtained.

Which antibiotics should be given empirically?

Because gram-positive organisms, namely coagulase-negative staphylococci and S aureus, are the causative pathogens in most cases of CIED-related infection, empiric antibiotic therapy should provide adequate coverage for these organisms. Because methicillin resistance is quite prevalent in staphylococci, we routinely use vancomycin (Vancocin) for empiric coverage. In patients who are allergic to vancomycin or cannot tolerate it, daptomycin (Cubicin) is an alternative.

Empiric gram-negative coverage is generally reserved for patients who present with systemic signs and symptoms, in whom delaying adequate coverage could have untoward consequences. We routinely use cefepime (Maxipime) for empiric gram-negative coverage in our institution. Other beta-lactam agents that provide coverage for gram-negative bacilli, especially Pseudomonas, are also appropriate in this setting.

Should the device be removed?

Superficial infection of the wound or incision site (eg, stitch abscess) early after implantation can be managed by conservative antibiotic therapy without removing the device. However, complete removal of the device system, including intracardiac leads, is necessary in all other presentations of device infection, even if the infection appears limited to the generator pocket.^5,12 Leaving the device in place or removing parts of the device is associated with persistent or relapsed infection and is not advisable.^17,26

Leaving the device in place may be necessary in extenuating circumstances, eg, if surgery would be too risky for the patient or if the patient refuses device removal or has a short life expectancy. In these cases, lifelong suppressive antibiotic therapy should be prescribed after an initial course of parenteral antibiotics.²⁷ Antibiotic choices for long-term suppressive therapy should be guided by antimicrobial susceptibility testing and consultation with an infectious disease specialist.

How should the leads be removed?

Leads are extracted percutaneously in most cases. Percutaneous extraction is generally considered safe even in cases in which infection is complicated by lead vegetations, which raises concern about pulmonary embolization of detached vegetation fragments during extraction.^5,20

Thoracotomy is generally reserved for patients who have cardiac complications (such as a cardiac abscess or the need to replace cardiac valves) or in whom attempts to extract the leads percutaneously are unsuccessful.

Details of the removal procedure and choice of extraction technique are beyond the scope of this paper and are best left to the discretion of the treating cardiologist or cardiac surgeon. Because of the potential for complications during percutaneous device removal, such as laceration of the superior vena cava or cardiac tamponade, the patient should be referred to a high-volume center where cardiothoracic intervention can be provided on an emergency basis if needed.

How long should antibiotic therapy go on?

An algorithm for deciding the duration of antibiotic therapy is shown in Figure 3. These guidelines, first published in 2007,¹⁷ were adopted by the American Heart Association in its updated statement on the management of CIED-related infections.⁵ However, it should be noted that these guidelines are not based on randomized clinical trials; rather, they represent expert opinion based on published series of patients with CIED-related infections.

In general, cases of device erosion or pocket infection can be treated with 1 to 2 weeks of appropriate antibiotic therapy based on antimicrobial susceptibility testing. However, cases of bloodstream infection require 2 to 4 weeks of antibiotic therapy—or sometimes even longer if associated complications are present, such as septic thrombosis, endocarditis, or osteomyelitis.

We favor parenteral antibiotics for the entire course of treatment. However, patients can be discharged from the hospital once the bloodstream infection has cleared, and the antibiotic course can be completed on an outpatient basis.

Outpatient antimicrobial monitoring

We recommend adherence to the Infectious Diseases Society of America’s guidelines for monitoring outpatient parenteral antimicrobial therapy.²⁸

At discharge from the hospital, patients should be instructed to promptly call their primary care physician if they have a fever or notice inflammatory changes at the pocket site. If the patient reports such symptoms, repeat blood cultures should be ordered, and the patient should be monitored closely for signs of a relapse of infection.

A routine follow-up visit should be arranged at 2 weeks and at the end of parenteral antibiotic therapy (for patients receiving therapy for 4 weeks or longer) to make sure the infection has resolved.

When should a new device be implanted?

Before deciding when a new device should be implanted, one should carefully assess whether the patient still needs one. Studies indicate that up to 30% of patients may no longer require a cardiac device.^17,18

However, we believe that removal of drains and closure of the old pocket are not necessary before implanting a new device in a different location (usually the contralateral pectoral area). Exceptions to this general principle are cases of valvular endocarditis, in which a minimum of 2 weeks is recommended between removal of an infected device (plus clearance of bloodstream infection) and implantation of a new device.

OUTCOMES OF INFECTION

Despite improvements in our understanding of how to manage CIED-related infection, the rates of morbidity and death remain significant.

The outcome, in part, depends on the clinical presentation and the patient’s comorbid conditions. In general, the death rate in patients with a pocket infection is less than 5%. However, in patients with endovascular infection, it may be as high as 20%.^16–18 Other factors that affect the outcome include complications such as septic thrombosis, valvular endocarditis, or osteomyelitis; complications during device extraction; the need for open heart surgery; and the overall health of the patient.

Complete removal of the device system is a requisite for successful outcome, and the risk of death tends to be higher if only part of the infected CIED system is extracted.²⁶

STRATEGIES TO PREVENT DEVICE INFECTION

Preventive efforts should focus on strategies to minimize the chances of contamination of the generator, leads, and pocket during implantation.²⁹ Patients who are known to be colonized with methicillin-resistant S aureus may benefit from decolonization programs, which should include nasal application of mupirocin (Bactroban) ointment preoperatively.³⁰ In addition, use of chlorhexidine for surgical-site antisepsis has been shown to reduce the risk of surgical site infection.³¹

Moreover, all patients should receive antibiotic prophylaxis before implantation of a CIED.^32,33 Most institutions use a first-generation cephalosporin, such as cefazolin (Ancef), for this purpose.³⁴ However, the increasing rate of methicillin resistance in staphylococci has led to the routine use of vancomycin for preoperative prophylaxis at some centers.¹⁸

Regardless of the antibiotic chosen for prophylaxis, protocols that ensure that all patients receive an appropriate antibiotic at the appropriate time are a key determinant in the success of these infection-control programs.

Much of the data are cross-sectional and epidemiologic, so the direction of causation (if causation exists) cannot be established with certainty. There is a host of interwoven confounders, and many of these intersect around the patient’s weight and the presence of sleep apnea. Nevertheless, the authors explore some provocative associations.

Over the years, clinicians have increasingly recognized the myriad of comorbidities that accompany sleep apnea. We have discussed this in the Journal on several occasions since 2005. Naïvely, I have attributed many of these, particularly the cardiac complications, to downstream effects of repetitive hypoxic and hypercarbic insults, but there may be more fundamental physiologic principles in play, some linked to the affected sleep cycle and not to the apnea.

Drs. Touma and Pannain discuss some of the physiologic consequences of altered or decreased sleep cycles. Some of these are a result of disrupting the circadian release of hormones such as glucocorticoids and growth hormone, both of which can influence the body’s sensitivity to insulin’s hypoglycemic effects. The same can be said for disruption of normal sympathetic-parasympathetic nerve flow. In addition, sleep disruption affects appetite. Thinking back to residency, I recall the need to follow the admonition of one of my peers: in order to survive nights on call, never miss a meal. I still remember the (leptin-linked?) cravings after being up all night for a heavy carbohydrate-laden breakfast. Given these effects, coupled with the fatigue of sleep deprivation resulting in decreased exercise, it is easy to construct innumerable positive feedback loops contributing to the development of insulin resistance and type 2 diabetes.

So while it is a truism that sleep is good and that we all need to “recharge our batteries,” we still lack a full understanding of the complex physiology of sleep and the effects of sleep deprivation on a number of clinical conditions, from diabetes to fibromyalgia.

Recognizing the associations is a beginning. Knowing what to do about defective sleep in terms of preventing or ameliorating disease awaits appropriately controlled interventional trials—and the definition of appropriate interventions to evaluate.

Much of the data are cross-sectional and epidemiologic, so the direction of causation (if causation exists) cannot be established with certainty. There is a host of interwoven confounders, and many of these intersect around the patient’s weight and the presence of sleep apnea. Nevertheless, the authors explore some provocative associations.

Over the years, clinicians have increasingly recognized the myriad of comorbidities that accompany sleep apnea. We have discussed this in the Journal on several occasions since 2005. Naïvely, I have attributed many of these, particularly the cardiac complications, to downstream effects of repetitive hypoxic and hypercarbic insults, but there may be more fundamental physiologic principles in play, some linked to the affected sleep cycle and not to the apnea.

Drs. Touma and Pannain discuss some of the physiologic consequences of altered or decreased sleep cycles. Some of these are a result of disrupting the circadian release of hormones such as glucocorticoids and growth hormone, both of which can influence the body’s sensitivity to insulin’s hypoglycemic effects. The same can be said for disruption of normal sympathetic-parasympathetic nerve flow. In addition, sleep disruption affects appetite. Thinking back to residency, I recall the need to follow the admonition of one of my peers: in order to survive nights on call, never miss a meal. I still remember the (leptin-linked?) cravings after being up all night for a heavy carbohydrate-laden breakfast. Given these effects, coupled with the fatigue of sleep deprivation resulting in decreased exercise, it is easy to construct innumerable positive feedback loops contributing to the development of insulin resistance and type 2 diabetes.

So while it is a truism that sleep is good and that we all need to “recharge our batteries,” we still lack a full understanding of the complex physiology of sleep and the effects of sleep deprivation on a number of clinical conditions, from diabetes to fibromyalgia.

Recognizing the associations is a beginning. Knowing what to do about defective sleep in terms of preventing or ameliorating disease awaits appropriately controlled interventional trials—and the definition of appropriate interventions to evaluate.

Much of the data are cross-sectional and epidemiologic, so the direction of causation (if causation exists) cannot be established with certainty. There is a host of interwoven confounders, and many of these intersect around the patient’s weight and the presence of sleep apnea. Nevertheless, the authors explore some provocative associations.

Over the years, clinicians have increasingly recognized the myriad of comorbidities that accompany sleep apnea. We have discussed this in the Journal on several occasions since 2005. Naïvely, I have attributed many of these, particularly the cardiac complications, to downstream effects of repetitive hypoxic and hypercarbic insults, but there may be more fundamental physiologic principles in play, some linked to the affected sleep cycle and not to the apnea.

Drs. Touma and Pannain discuss some of the physiologic consequences of altered or decreased sleep cycles. Some of these are a result of disrupting the circadian release of hormones such as glucocorticoids and growth hormone, both of which can influence the body’s sensitivity to insulin’s hypoglycemic effects. The same can be said for disruption of normal sympathetic-parasympathetic nerve flow. In addition, sleep disruption affects appetite. Thinking back to residency, I recall the need to follow the admonition of one of my peers: in order to survive nights on call, never miss a meal. I still remember the (leptin-linked?) cravings after being up all night for a heavy carbohydrate-laden breakfast. Given these effects, coupled with the fatigue of sleep deprivation resulting in decreased exercise, it is easy to construct innumerable positive feedback loops contributing to the development of insulin resistance and type 2 diabetes.

So while it is a truism that sleep is good and that we all need to “recharge our batteries,” we still lack a full understanding of the complex physiology of sleep and the effects of sleep deprivation on a number of clinical conditions, from diabetes to fibromyalgia.

Recognizing the associations is a beginning. Knowing what to do about defective sleep in terms of preventing or ameliorating disease awaits appropriately controlled interventional trials—and the definition of appropriate interventions to evaluate.

Adults are sleeping less and less in our society. Yet sleep is no longer thought of as strictly a restorative process for the body. The importance of sleep for metabolic function and specifically glucose homeostasis is now widely accepted, as many studies have shown a correlation between sleep deprivation or poor sleep quality and an increased risk of diabetes.

Obesity and aging are both associated with worse sleep. As the prevalence of obesity and diabetes increases, and as the number of elderly people increases, it is imperative to target sleep in the overall treatment of our patients.

In the pages that follow, we examine the evidence of a link between sleep loss (both short sleep duration and poor-quality sleep) and the risk of diabetes. (For evidence linking short sleep duration and the related problem of obesity, we invite the reader to refer to previous publications on the topic.^1,2)

SLEEP LOSS, OBESITY, AND DIABETES ARE ALL ON THE RISE

The prevalence of obesity and, consequently, of type 2 diabetes mellitus has increased alarmingly worldwide and particularly in the United States in the past few decades. Such a rapid increase cannot be explained simply by an alteration in the genetic pool; it is more likely due to environmental, socioeconomic, behavioral, and demographic factors and the interaction between genetics and these factors. Besides traditional lifestyle factors such as high-calorie diets and sedentary habits, other, nontraditional behavioral and environmental factors could be contributing to the epidemic of obesity and diabetes.

At the same time, people are sleeping less, and sleep disorders are on the rise. According to recent polls from the US Centers for Disease Control and Prevention, approximately 29% of US adults report sleeping less than 7 hours per night, and 50 to 70 million have chronic sleep and wakefulness disorders.³

The sleep curtailment of our times probably is partly self-imposed, as the pace and the opportunities of modern society place more demands on time for work and leisure activities and leave less time for sleep.

The quality of sleep has also declined as the population has aged and as the prevalence of obesity and its related sleep disorders has increased. Furthermore, patients with type 2 diabetes tend to sleep less, and to sleep poorly.^4,5 Poor sleep quality generally results in overall sleep loss.

GLUCOSE TOLERANCE HAS A CIRCADIAN RHYTHM

The human body regulates blood levels of glucose within a narrow range.

Glucose tolerance refers to the ability to maintain euglycemia by disposing of exogenous glucose via insulin-mediated and non–insulin-mediated mechanisms. Normal glucose tolerance depends on the ability of the pancreatic beta cells to produce insulin. As insulin sensitivity declines, insulin secretion increases to maintain normal glucose levels. Diabetes becomes manifest when the pancreatic beta cells fail to compensate for the decreased insulin sensitivity.

Glucose tolerance varies in a circadian rhythm, including during the different stages of sleep.

HOW SLEEP AFFECTS METABOLISM AND HORMONES

Sleep has often been thought of as a “restorative” process for the mind and the body; however, many studies have shown that it also directly affects many metabolic and hormonal processes.⁶

Sleep has five stages: rapid eye movement (REM) sleep and stages 1, 2, 3, and 4 of non-REM sleep. The deeper stages of non-REM sleep, ie, stages 3 and 4, are also known as slow-wave sleep and are thought to be the most restorative.

Additionally, the onset of slow-wave sleep is temporally associated with transient metabolic, hormonal, and neurophysiologic changes, all of which can affect glucose homeostasis. The brain uses less glucose,⁷ the pituitary gland releases more growth hormone and less corticotropin,⁸ the sympathetic nervous system is less active, and conversely, vagal tone is increased.⁹

As a result, in the first part of the night, when slow-wave sleep predominates, glucose metabolism is slower. These effects are reversed in the second part of the night, when REM sleep, stage 1, and awakening are more likely.

In view of these important changes in glucose metabolism during sleep, it is not surprising that getting less sleep or poorer sleep on a regular basis could affect overall glucose homeostasis.

SHORT SLEEP DURATION AND RISK OF DIABETES

Laboratory and epidemiologic evidence supports an association between short sleep duration (< 7 hours per night) and the risk of diabetes, and also between poor sleep quality and the risk of diabetes. We will explore putative mechanisms for these relationships.

Laboratory studies of short sleep duration and glucose metabolism

Studies in small numbers of healthy volunteers who underwent experimental sleep restriction or disruption have revealed mechanisms by which sleep loss might increase the risk of diabetes.

Kuhn et al¹⁰ performed the very first laboratory study of the effect of sleep deprivation on metabolism. Published in 1969, it showed that total sleep deprivation led to a marked increase in glucose levels.

A caution in extrapolating such results to real-life conditions is that total sleep deprivation is uncommon in humans and is inevitably followed by sleep recovery, with normalization of glucose metabolism. However, people in modern society are experiencing recurrent partial sleep deprivation, and its effect on glucose metabolism may be different.

Spiegel et al,¹¹ in landmark laboratory studies of partial sleep deprivation in healthy, lean adults, found that restricting sleep to 4 hours per night for 6 nights resulted in a 40% decrease in glucose tolerance, to levels similar to those seen in older adults with impaired glucose tolerance. This metabolic change was paralleled by an increase in the activity of the sympathetic nervous system, and both of these effects reversed with sleep recovery.

A criticism of these initial studies is that they restricted sleep to 4 hours, a restriction more severe than that seen in real life.

Nedeltcheva et al¹² more recently examined the effects of less-severe sleep curtailment (5.5 hours per night for 14 nights) in sedentary middle-aged men and women. This degree of bedtime restriction led to a decrease in glucose tolerance due to decreased insulin sensitivity in the absence of adequate beta cell compensation.

Such recurrent bedtime restriction is closer to the short sleep duration experienced by many people in everyday life, and in people at risk it may facilitate the development of insulin resistance, reduced glucose tolerance, and ultimately diabetes. Indeed, epidemiologic studies suggest that people who sleep less than 6 hours per night are at higher risk of type 2 diabetes.

Multiple cross-sectional epidemiologic studies have suggested an association between short sleep duration and diabetes, and several prospective epidemiologic studies have suggested that short sleep actually plays a causative role in diabetes.

The landmark observations of Spiegel et al¹¹ led to a number of epidemiologic studies examining the relationships between sleep duration and sleep disturbances and diabetes risk.¹³

The Sleep Heart Study¹⁴ was a large, cross-sectional, community-based study of the cardiovascular consequences of sleep-disordered breathing. The authors assessed the relationship between reported sleep duration and impaired glucose tolerance or type 2 diabetes in more than 1,400 men and women who had no history of insomnia. After adjustment for age, sex, race, body habitus, and apnea-hypopnea index, the prevalence of impaired glucose tolerance and type 2 diabetes was higher in those who reported sleeping 6 hours or less per night—or 9 hours or more per night (more below about the possible effect of too much sleep on the risk of diabetes).

The major limitations of the study were that it was cross-sectional in design, sleep duration was self-reported, the reasons for sleep curtailment were unknown, and possible confounding variables as physical activity, diet, and socioeconomic status were not measured.

Knutson et al,⁴ in our medical center, examined the association between self-reported sleep duration and sleep quality on the one hand and hemoglobin A_1c levels on the other in 161 black patients with type 2 diabetes. In patients without diabetic complications, glycemic control correlated with perceived sleep debt (calculated as the difference between self-reported actual and preferred weekday sleep duration); the authors calculated that a perceived sleep debt of 3 hours per night predicted a hemoglobin A_1c value 1.1 absolute percentage points higher than the median value. The analyses controlled for age, sex, body mass index, insulin use, and the presence of major complications; it excluded patients whose sleep was frequently disrupted by pain. The effect size was comparable to (but opposite) that of oral antidiabetic drugs. However, the direction of causality cannot be confirmed from this association, as it is possible that poor glycemic control in diabetic patients could impair their ability to achieve sufficient sleep.

To date, several major prospective studies have looked at the association between short sleep duration and sleep problems and the risk of developing type 2 diabetes in adults.

The Nurses Health Study¹⁵ followed 70,000 nondiabetic women for 10 years. Compared with nurses who slept 7 to 8 hours per 24 hours, those who slept 5 hours or less had a relative risk of diabetes of 1.34 even after controlling for many covariables, such as body mass index, shift work, hypertension, exercise, and depression.

The first National Health and Nutrition Examination Survey (NHANES I)¹⁶ examined the effect of sleep duration on the risk of incident diabetes in roughly 9,000 men and women over a period of 8 to 10 years. The statistical model included body mass index and hypertension and adjusted for physical activity, depression, alcohol consumption, ethnicity, education, marital status, and age. Findings: those who slept 5 hours or less per night were significantly more likely to develop type 2 diabetes than were those who slept 7 hours per night (odds ratio 1.57, 95% confidence interval [CI] 1.11–2.22), and so were those who slept 9 or more hours per night (odds ratio 1.57, 95% CI 1.10–2.24).

Kawakami et al¹⁷ followed 2,649 Japanese men for 8 years. Those who had difficulty going to sleep and staying asleep, which are both likely to result in shorter sleep duration, had higher age-adjusted risks of developing type 2 diabetes, with hazard ratios of 2.98 and 2.23, respectively.

Björkelund et al¹⁸ followed 6,599 nondiabetic Swedish men for an average of 15 years. Self-reported difficulty sleeping predicted the development of diabetes with an odds ratio of 1.52 even after controlling for age, body mass index at screening, changes in body mass index at follow-up, baseline glucose level, follow-up time, physical activity, family history of type 2 diabetes, smoking, social class, and alcohol intake.¹⁹

Interestingly, the authors found that the resting heart rate was higher at baseline in the men who later developed diabetes. This finding could be interpreted as reflecting greater sympathetic nervous system activity, a putative mediator of the metabolic dysfunction associated with both short sleep duration and obstructive sleep apnea.^20,21

Meisinger et al,²² in a study of more than 8,000 nondiabetic German men and women 25 to 74 years old, found a hazard ratio of developing diabetes of 1.60 (95% CI 1.05–2.45) in men and 1.98 (95% CI 1.20–3.29) in women who reported difficulty staying asleep, who thus would have shortened sleep duration. This effect was independent of other risk factors for diabetes.

Yaggi et al,²³ in a prospective study of 1,139 US men, also found a U-shaped relationship between sleep duration and the incidence of diabetes, with higher rates in people who slept less than 5 or more than 8 hours per night.

Cappuccio et al²⁴ performed a meta-analysis of all the prospective studies published to date. Their review included 10 prospective studies, with 107,756 participants followed for a median of 9.5 years. Sleep duration and sleep disturbances were self-reported in all the studies. They calculated that the risk of developing diabetes was 28% higher with short sleep duration (≤ 5 or < 6 hours in the different studies), 48% higher with long sleep duration (> 8 hours), 57% higher with difficulty going to sleep, and 84% higher with difficulty staying asleep.

Limitations of these studies. A consideration when trying to interpret the relationship between length of sleep and the incidence of diabetes is that sleep duration in these studies was self-reported, not measured. If a patient reports sleeping more than 8 hours per night, it could mean that he or she is not truly getting so much sleep, but rather is spending more time in bed trying to sleep.

Another possibility is that the higher incidence of type 2 diabetes in people who slept longer is due to undiagnosed obstructive sleep apnea, which is associated with daytime sleepiness and possibly longer sleep time to compensate for inefficient sleep.

Finally, depressive symptoms, unemployment, a low level of physical activity, and undiagnosed health conditions have all been associated with long sleep duration and could affect the relationship with diabetes risk.

In summary, epidemiologic studies from different geographic locations have consistently indicated that short sleep or poor sleep may increase the risk of developing type 2 diabetes mellitus and suggest that such an association spans different countries, cultures, and ethnic groups.

Therefore, there is a need for additional prospective epidemiologic studies that use objective measures of sleep. Furthermore, studies need to determine whether the cause of sleep restriction (eg, insomnia vs lifestyle choice) affects this relationship. Randomized, controlled, interventional studies would also be useful to determine whether lengthening sleep duration affects the development of impaired glucose tolerance or type 2 diabetes mellitus.

The effects of sleep loss on glucose metabolism are likely multifactorial, involving several interacting pathways.

Decreased brain glucose utilization has been shown on positron emission tomography in sleep-deprived subjects.²⁵

Hormonal dysregulation. Sleep deprivation is associated with disturbances in the secretion of the counterregulatory hormones growth hormone²⁶ and cortisol.¹¹

Young, healthy volunteers who were allowed to sleep only 4 hours per night for 6 nights showed a change in their patterns of growth hormone release, from a normal single pulse to a biphasic pattern.²⁶ They were exposed to a higher overall amount of growth hormone in the sleep-deprived condition, which could contribute to higher glucose levels.

Also, evening cortisol levels were significantly higher in young, healthy men who were allowed to sleep only 4 hours per night for 6 nights,¹¹ as well as in young, healthy women who were allowed to sleep only 3 hours for 1 night.²⁷ A cross-sectional analysis that included 2,751 men and women also demonstrated that short sleep duration and sleep disturbances are independently associated with more cortisol secretion in the evening.²⁸ Elevated evening cortisol levels can lead to morning insulin resistance.²⁹

Inflammation. Levels of inflammatory cytokines, inflammation, or both increase as sleep duration decreases, which in turn can also increase insulin resistance.^30,31

Sympathetic nervous system activity. Patients who have been sleep-deprived have been shown to have higher sympathetic nervous system activity, lower parasympathetic activity, or both.^11,32 The sympathetic nervous system inhibits insulin release while the parasympathetic system stimulates it, so these changes both increase glucose levels.³³ Moreover, overactivity of the sympathetic nervous system results in insulin resistance.³⁴

Excess weight is a well-established risk factor for type 2 diabetes mellitus, and several epidemiologic studies have suggested that sleep loss may increase the risk of becoming overweight or obese,^1,35 which would ultimately increase the risk of type 2 diabetes.

A primary mechanism linking sleep deprivation and weight gain is likely to be hyperactivity of the orexin system. Orexigenic neurons play a central role in wakefulness, but, as suggested by the name, they also promote feeding.³⁶ Studies in animals have indicated that the orexin system is overactive during sleep deprivation,^37–39 and this could be in part mediated by the increase in sympathetic activity.

Increased sympathetic activity also affects the levels of peripheral appetite hormones, inhibiting leptin release⁴⁰ and stimulating ghrelin release.⁴¹ Lower leptin levels and higher ghrelin levels act in concert to further activate orexin neurons,^42,43 resulting in increased food intake.

One could also argue that less time sleeping also allows more opportunity to eat.⁴⁴

Reduced energy expenditure. Sleep loss and its associated sleepiness and fatigue may result in reduced energy expenditure, partly due to less exercise but also due to less nonexercise activity thermogenesis. To date, reduced energy expenditure is an unexplored pathway that could link short sleep, the risk of obesity, and ultimately diabetes. In many overweight and obese people, this cascade of negative events is likely to be accelerated by sleep-disordered breathing, a reported independent risk factor for insulin resistance.^45,46

SLEEP QUALITY AND THE RISK OF DIABETES

Slow-wave sleep and diabetes

Slow-wave sleep, the most restorative sleep, is associated with metabolic, hormonal, and neurophysiologic changes that affect glucose homeostasis. Its disturbance may have deleterious effects on glucose tolerance.

Shallow slow-wave sleep occurs in elderly people⁴⁷ and in obese people, even in the absence of obstructive sleep apnea.^48,49 Both groups are also at higher risk of diabetes.⁵⁰ One wonders if the decreased slow-wave sleep could in part contribute to the risk of diabetes in these groups.

A few studies specifically tested the effect of experimental suppression of slow-wave sleep on glucose homeostasis.

Tasali et al⁵¹ evaluated nine young, lean, nondiabetic men and women after 2 consecutive nights of undisturbed sleep and after 3 consecutive nights of suppressed slow-wave sleep without a change in total sleep duration or in REM sleep duration. Slow-wave sleep was disturbed by “delivering acoustic stimuli of various frequencies and intensities” whenever the subjects started to go into stage 3 or stage 4 sleep. This decreased the amount of slow-wave sleep by nearly 90%, which is comparable to the degree of sleep fragmentation seen in moderate to severe obstructive sleep apnea. After 3 nights of slow-wave sleep suppression, insulin sensitivity decreased by 25%, without a compensatory increase in insulin release, which resulted in a reduction in glucose tolerance of 23%, a value seen in older adults with impaired glucose tolerance.⁵²

Stamatakis et al⁵³ confirmed these findings in a similar study of 11 healthy, normal volunteers whose sleep was fragmented for 2 nights across all stages of sleep using auditory and mechanical stimuli. Insulin sensitivity significantly decreased, as did glucose effectiveness (ability of glucose to dispose itself independently of an insulin response) after the 2 nights of disturbed sleep quality.

These results support the hypothesis that poor sleep quality with short durations of slow-wave sleep, as seen with aging and obesity, could contribute to the higher risk of type 2 diabetes in these populations. These data also suggest that more studies are needed to look at the relationship between amount and quality of slow-wave sleep and diabetes risk.

The most robust evidence that not only short sleep duration but also poor sleep quality affects diabetes risk comes from studies of metabolic function in patients with obstructive sleep apnea, an increasingly common condition.

Obstructive sleep apnea is characterized by recurrent episodes of partial or complete upper airway obstruction with intermittent hypoxia and microarousals, resulting in low amounts of slow-wave sleep and overall decreased sleep quality.⁵⁴

Obstructive sleep apnea is common in patients with type 2 diabetes, and several clinical and epidemiologic studies suggest that, untreated, it may worsen diabetes risk or control.^{21,45–46,55–59}

The Sleep AHEAD (Action for Health in Diabetes) study⁶⁰ revealed, in cross-sectional data, that more than 84% of obese patients with type 2 diabetes had obstructive sleep apnea (with an apnea-hypopnea index ≥ 5).

Aronsohn et al,⁵ in a study conducted in our laboratory in 60 patients with type 2 diabetes, found that 46 (77%) of them had obstructive sleep apnea. Furthermore, the worse the obstructive sleep apnea, the worse the glucose control. After controlling for age, sex, race, body mass index, number of diabetes medications, level of exercise, years of diabetes, and total sleep time, compared with patients without obstructive sleep apnea, the adjusted mean hemoglobin A_1c was increased in a linear trend by (in absolute percentage points):

• 1.49% in patients with mild obstructive sleep apnea (P = .0028)
• 1.93% in patients with moderate obstructive sleep apnea (P = .0033)
• 3.69% in patients with severe obstructive sleep apnea (P < .0001).

Other epidemiologic studies. A growing number of epidemiologic studies, in various geographic regions, have suggested an independent link between obstructive sleep apnea and risk of type 2 diabetes.⁶¹ Most of the studies have been cross-sectional, and while most had positive findings, a criticism is that the methodology varied among the studies, both in how obstructive sleep apnea was assessed (snoring vs polysomnography) and in the metabolic assessment (oral glucose tolerance test, homeostatic model assessment, hemoglobin A_1c, medical history, physician examination, or patient report).

A more recent prospective study of 544 nondiabetic patients⁶⁵ showed that the risk of developing type 2 diabetes over an average of 2.7 years of follow-up was a function of the severity of obstructive sleep apnea expressed in quartiles: for each increased quartile of severity there was a 43% increase in the incidence of diabetes. Additionally, in patients with moderate to severe sleep apnea, regular use of continuous positive airway pressure (CPAP) was associated with an attenuated risk.⁶⁵

Two prospective studies (not included in Table 1) used snoring as a marker of obstructive sleep apnea; at 10 years of follow-up, snoring was associated with a higher risk of developing diabetes in both men and women.^73,74

Does CPAP improve glucose metabolism? Other studies have specifically examined the effects of CPAP treatment on glucose metabolism, in both diabetic and nondiabetic populations. Accumulating evidence suggests that metabolic abnormalities can be partially corrected by CPAP treatment, which supports the concept of a causal link between obstructive sleep apnea and altered glucose control. This topic is beyond the scope of this review; please see previously published literature^61,75 for further information. Whether treating obstructive sleep apnea may delay the development or reduce the severity of type 2 diabetes is another important unanswered question.

Is obstructive sleep apnea a cause or consequence of diabetes? It may be a novel risk factor for type 2 diabetes, and its association with altered glucose metabolism is well supported by a large set of cross-sectional studies, but there are still insufficient longitudinal studies to indicate a direction of causality.

If obstructive sleep apnea is the cause, what is the mechanism? There are likely many. High levels of sympathetic nervous system activity, intermittent hypoxia, sleep fragmentation, and sleep loss in obstructive sleep apnea may all lead to dysregulation of the hypothalamic-pituitary axis, endothelial dysfunction, and alterations in cytokine and adipokine release and are all potential mechanisms of abnormal glucose metabolism in this population.

WHAT TO TELL PATIENTS

Taken together, the current evidence suggests that strategies to improve the duration and the quality of sleep should be considered as a potential intervention to prevent or delay the development of type 2 diabetes mellitus in at-risk populations. While further studies are needed to better elucidate the mechanisms of the relationship between sleep loss and diabetes risk and to determine if extending sleep and treating obstructive sleep apnea decreases the risk of diabetes, we urge clinicians to recommend at least 7 hours of uninterrupted sleep per night as a goal in maintaining a healthy lifestyle. Additionally, clinicians should systematically evaluate the risk of obstructive sleep apnea in their patients who have type 2 diabetes mellitus and the metabolic syndrome, and conversely, should assess for diabetes in patients with known obstructive sleep apnea.

Adults are sleeping less and less in our society. Yet sleep is no longer thought of as strictly a restorative process for the body. The importance of sleep for metabolic function and specifically glucose homeostasis is now widely accepted, as many studies have shown a correlation between sleep deprivation or poor sleep quality and an increased risk of diabetes.

Obesity and aging are both associated with worse sleep. As the prevalence of obesity and diabetes increases, and as the number of elderly people increases, it is imperative to target sleep in the overall treatment of our patients.

In the pages that follow, we examine the evidence of a link between sleep loss (both short sleep duration and poor-quality sleep) and the risk of diabetes. (For evidence linking short sleep duration and the related problem of obesity, we invite the reader to refer to previous publications on the topic.^1,2)

SLEEP LOSS, OBESITY, AND DIABETES ARE ALL ON THE RISE

The prevalence of obesity and, consequently, of type 2 diabetes mellitus has increased alarmingly worldwide and particularly in the United States in the past few decades. Such a rapid increase cannot be explained simply by an alteration in the genetic pool; it is more likely due to environmental, socioeconomic, behavioral, and demographic factors and the interaction between genetics and these factors. Besides traditional lifestyle factors such as high-calorie diets and sedentary habits, other, nontraditional behavioral and environmental factors could be contributing to the epidemic of obesity and diabetes.

At the same time, people are sleeping less, and sleep disorders are on the rise. According to recent polls from the US Centers for Disease Control and Prevention, approximately 29% of US adults report sleeping less than 7 hours per night, and 50 to 70 million have chronic sleep and wakefulness disorders.³

The sleep curtailment of our times probably is partly self-imposed, as the pace and the opportunities of modern society place more demands on time for work and leisure activities and leave less time for sleep.

The quality of sleep has also declined as the population has aged and as the prevalence of obesity and its related sleep disorders has increased. Furthermore, patients with type 2 diabetes tend to sleep less, and to sleep poorly.^4,5 Poor sleep quality generally results in overall sleep loss.

GLUCOSE TOLERANCE HAS A CIRCADIAN RHYTHM

The human body regulates blood levels of glucose within a narrow range.

Glucose tolerance refers to the ability to maintain euglycemia by disposing of exogenous glucose via insulin-mediated and non–insulin-mediated mechanisms. Normal glucose tolerance depends on the ability of the pancreatic beta cells to produce insulin. As insulin sensitivity declines, insulin secretion increases to maintain normal glucose levels. Diabetes becomes manifest when the pancreatic beta cells fail to compensate for the decreased insulin sensitivity.

Glucose tolerance varies in a circadian rhythm, including during the different stages of sleep.

HOW SLEEP AFFECTS METABOLISM AND HORMONES

Sleep has often been thought of as a “restorative” process for the mind and the body; however, many studies have shown that it also directly affects many metabolic and hormonal processes.⁶

Sleep has five stages: rapid eye movement (REM) sleep and stages 1, 2, 3, and 4 of non-REM sleep. The deeper stages of non-REM sleep, ie, stages 3 and 4, are also known as slow-wave sleep and are thought to be the most restorative.

Additionally, the onset of slow-wave sleep is temporally associated with transient metabolic, hormonal, and neurophysiologic changes, all of which can affect glucose homeostasis. The brain uses less glucose,⁷ the pituitary gland releases more growth hormone and less corticotropin,⁸ the sympathetic nervous system is less active, and conversely, vagal tone is increased.⁹

As a result, in the first part of the night, when slow-wave sleep predominates, glucose metabolism is slower. These effects are reversed in the second part of the night, when REM sleep, stage 1, and awakening are more likely.

In view of these important changes in glucose metabolism during sleep, it is not surprising that getting less sleep or poorer sleep on a regular basis could affect overall glucose homeostasis.

SHORT SLEEP DURATION AND RISK OF DIABETES

Laboratory and epidemiologic evidence supports an association between short sleep duration (< 7 hours per night) and the risk of diabetes, and also between poor sleep quality and the risk of diabetes. We will explore putative mechanisms for these relationships.

Laboratory studies of short sleep duration and glucose metabolism

Studies in small numbers of healthy volunteers who underwent experimental sleep restriction or disruption have revealed mechanisms by which sleep loss might increase the risk of diabetes.

Kuhn et al¹⁰ performed the very first laboratory study of the effect of sleep deprivation on metabolism. Published in 1969, it showed that total sleep deprivation led to a marked increase in glucose levels.

A caution in extrapolating such results to real-life conditions is that total sleep deprivation is uncommon in humans and is inevitably followed by sleep recovery, with normalization of glucose metabolism. However, people in modern society are experiencing recurrent partial sleep deprivation, and its effect on glucose metabolism may be different.

Spiegel et al,¹¹ in landmark laboratory studies of partial sleep deprivation in healthy, lean adults, found that restricting sleep to 4 hours per night for 6 nights resulted in a 40% decrease in glucose tolerance, to levels similar to those seen in older adults with impaired glucose tolerance. This metabolic change was paralleled by an increase in the activity of the sympathetic nervous system, and both of these effects reversed with sleep recovery.

A criticism of these initial studies is that they restricted sleep to 4 hours, a restriction more severe than that seen in real life.

Nedeltcheva et al¹² more recently examined the effects of less-severe sleep curtailment (5.5 hours per night for 14 nights) in sedentary middle-aged men and women. This degree of bedtime restriction led to a decrease in glucose tolerance due to decreased insulin sensitivity in the absence of adequate beta cell compensation.

Such recurrent bedtime restriction is closer to the short sleep duration experienced by many people in everyday life, and in people at risk it may facilitate the development of insulin resistance, reduced glucose tolerance, and ultimately diabetes. Indeed, epidemiologic studies suggest that people who sleep less than 6 hours per night are at higher risk of type 2 diabetes.

Multiple cross-sectional epidemiologic studies have suggested an association between short sleep duration and diabetes, and several prospective epidemiologic studies have suggested that short sleep actually plays a causative role in diabetes.

The landmark observations of Spiegel et al¹¹ led to a number of epidemiologic studies examining the relationships between sleep duration and sleep disturbances and diabetes risk.¹³

The Sleep Heart Study¹⁴ was a large, cross-sectional, community-based study of the cardiovascular consequences of sleep-disordered breathing. The authors assessed the relationship between reported sleep duration and impaired glucose tolerance or type 2 diabetes in more than 1,400 men and women who had no history of insomnia. After adjustment for age, sex, race, body habitus, and apnea-hypopnea index, the prevalence of impaired glucose tolerance and type 2 diabetes was higher in those who reported sleeping 6 hours or less per night—or 9 hours or more per night (more below about the possible effect of too much sleep on the risk of diabetes).

The major limitations of the study were that it was cross-sectional in design, sleep duration was self-reported, the reasons for sleep curtailment were unknown, and possible confounding variables as physical activity, diet, and socioeconomic status were not measured.

Knutson et al,⁴ in our medical center, examined the association between self-reported sleep duration and sleep quality on the one hand and hemoglobin A_1c levels on the other in 161 black patients with type 2 diabetes. In patients without diabetic complications, glycemic control correlated with perceived sleep debt (calculated as the difference between self-reported actual and preferred weekday sleep duration); the authors calculated that a perceived sleep debt of 3 hours per night predicted a hemoglobin A_1c value 1.1 absolute percentage points higher than the median value. The analyses controlled for age, sex, body mass index, insulin use, and the presence of major complications; it excluded patients whose sleep was frequently disrupted by pain. The effect size was comparable to (but opposite) that of oral antidiabetic drugs. However, the direction of causality cannot be confirmed from this association, as it is possible that poor glycemic control in diabetic patients could impair their ability to achieve sufficient sleep.

To date, several major prospective studies have looked at the association between short sleep duration and sleep problems and the risk of developing type 2 diabetes in adults.

The Nurses Health Study¹⁵ followed 70,000 nondiabetic women for 10 years. Compared with nurses who slept 7 to 8 hours per 24 hours, those who slept 5 hours or less had a relative risk of diabetes of 1.34 even after controlling for many covariables, such as body mass index, shift work, hypertension, exercise, and depression.

The first National Health and Nutrition Examination Survey (NHANES I)¹⁶ examined the effect of sleep duration on the risk of incident diabetes in roughly 9,000 men and women over a period of 8 to 10 years. The statistical model included body mass index and hypertension and adjusted for physical activity, depression, alcohol consumption, ethnicity, education, marital status, and age. Findings: those who slept 5 hours or less per night were significantly more likely to develop type 2 diabetes than were those who slept 7 hours per night (odds ratio 1.57, 95% confidence interval [CI] 1.11–2.22), and so were those who slept 9 or more hours per night (odds ratio 1.57, 95% CI 1.10–2.24).

Kawakami et al¹⁷ followed 2,649 Japanese men for 8 years. Those who had difficulty going to sleep and staying asleep, which are both likely to result in shorter sleep duration, had higher age-adjusted risks of developing type 2 diabetes, with hazard ratios of 2.98 and 2.23, respectively.

Björkelund et al¹⁸ followed 6,599 nondiabetic Swedish men for an average of 15 years. Self-reported difficulty sleeping predicted the development of diabetes with an odds ratio of 1.52 even after controlling for age, body mass index at screening, changes in body mass index at follow-up, baseline glucose level, follow-up time, physical activity, family history of type 2 diabetes, smoking, social class, and alcohol intake.¹⁹

Interestingly, the authors found that the resting heart rate was higher at baseline in the men who later developed diabetes. This finding could be interpreted as reflecting greater sympathetic nervous system activity, a putative mediator of the metabolic dysfunction associated with both short sleep duration and obstructive sleep apnea.^20,21

Meisinger et al,²² in a study of more than 8,000 nondiabetic German men and women 25 to 74 years old, found a hazard ratio of developing diabetes of 1.60 (95% CI 1.05–2.45) in men and 1.98 (95% CI 1.20–3.29) in women who reported difficulty staying asleep, who thus would have shortened sleep duration. This effect was independent of other risk factors for diabetes.

Yaggi et al,²³ in a prospective study of 1,139 US men, also found a U-shaped relationship between sleep duration and the incidence of diabetes, with higher rates in people who slept less than 5 or more than 8 hours per night.

Cappuccio et al²⁴ performed a meta-analysis of all the prospective studies published to date. Their review included 10 prospective studies, with 107,756 participants followed for a median of 9.5 years. Sleep duration and sleep disturbances were self-reported in all the studies. They calculated that the risk of developing diabetes was 28% higher with short sleep duration (≤ 5 or < 6 hours in the different studies), 48% higher with long sleep duration (> 8 hours), 57% higher with difficulty going to sleep, and 84% higher with difficulty staying asleep.

Limitations of these studies. A consideration when trying to interpret the relationship between length of sleep and the incidence of diabetes is that sleep duration in these studies was self-reported, not measured. If a patient reports sleeping more than 8 hours per night, it could mean that he or she is not truly getting so much sleep, but rather is spending more time in bed trying to sleep.

Another possibility is that the higher incidence of type 2 diabetes in people who slept longer is due to undiagnosed obstructive sleep apnea, which is associated with daytime sleepiness and possibly longer sleep time to compensate for inefficient sleep.

Finally, depressive symptoms, unemployment, a low level of physical activity, and undiagnosed health conditions have all been associated with long sleep duration and could affect the relationship with diabetes risk.

In summary, epidemiologic studies from different geographic locations have consistently indicated that short sleep or poor sleep may increase the risk of developing type 2 diabetes mellitus and suggest that such an association spans different countries, cultures, and ethnic groups.

Therefore, there is a need for additional prospective epidemiologic studies that use objective measures of sleep. Furthermore, studies need to determine whether the cause of sleep restriction (eg, insomnia vs lifestyle choice) affects this relationship. Randomized, controlled, interventional studies would also be useful to determine whether lengthening sleep duration affects the development of impaired glucose tolerance or type 2 diabetes mellitus.

The effects of sleep loss on glucose metabolism are likely multifactorial, involving several interacting pathways.

Decreased brain glucose utilization has been shown on positron emission tomography in sleep-deprived subjects.²⁵

Hormonal dysregulation. Sleep deprivation is associated with disturbances in the secretion of the counterregulatory hormones growth hormone²⁶ and cortisol.¹¹

Young, healthy volunteers who were allowed to sleep only 4 hours per night for 6 nights showed a change in their patterns of growth hormone release, from a normal single pulse to a biphasic pattern.²⁶ They were exposed to a higher overall amount of growth hormone in the sleep-deprived condition, which could contribute to higher glucose levels.

Also, evening cortisol levels were significantly higher in young, healthy men who were allowed to sleep only 4 hours per night for 6 nights,¹¹ as well as in young, healthy women who were allowed to sleep only 3 hours for 1 night.²⁷ A cross-sectional analysis that included 2,751 men and women also demonstrated that short sleep duration and sleep disturbances are independently associated with more cortisol secretion in the evening.²⁸ Elevated evening cortisol levels can lead to morning insulin resistance.²⁹

Inflammation. Levels of inflammatory cytokines, inflammation, or both increase as sleep duration decreases, which in turn can also increase insulin resistance.^30,31

Sympathetic nervous system activity. Patients who have been sleep-deprived have been shown to have higher sympathetic nervous system activity, lower parasympathetic activity, or both.^11,32 The sympathetic nervous system inhibits insulin release while the parasympathetic system stimulates it, so these changes both increase glucose levels.³³ Moreover, overactivity of the sympathetic nervous system results in insulin resistance.³⁴

Excess weight is a well-established risk factor for type 2 diabetes mellitus, and several epidemiologic studies have suggested that sleep loss may increase the risk of becoming overweight or obese,^1,35 which would ultimately increase the risk of type 2 diabetes.

A primary mechanism linking sleep deprivation and weight gain is likely to be hyperactivity of the orexin system. Orexigenic neurons play a central role in wakefulness, but, as suggested by the name, they also promote feeding.³⁶ Studies in animals have indicated that the orexin system is overactive during sleep deprivation,^37–39 and this could be in part mediated by the increase in sympathetic activity.

Increased sympathetic activity also affects the levels of peripheral appetite hormones, inhibiting leptin release⁴⁰ and stimulating ghrelin release.⁴¹ Lower leptin levels and higher ghrelin levels act in concert to further activate orexin neurons,^42,43 resulting in increased food intake.

One could also argue that less time sleeping also allows more opportunity to eat.⁴⁴

Reduced energy expenditure. Sleep loss and its associated sleepiness and fatigue may result in reduced energy expenditure, partly due to less exercise but also due to less nonexercise activity thermogenesis. To date, reduced energy expenditure is an unexplored pathway that could link short sleep, the risk of obesity, and ultimately diabetes. In many overweight and obese people, this cascade of negative events is likely to be accelerated by sleep-disordered breathing, a reported independent risk factor for insulin resistance.^45,46

SLEEP QUALITY AND THE RISK OF DIABETES

Slow-wave sleep and diabetes

Slow-wave sleep, the most restorative sleep, is associated with metabolic, hormonal, and neurophysiologic changes that affect glucose homeostasis. Its disturbance may have deleterious effects on glucose tolerance.

Shallow slow-wave sleep occurs in elderly people⁴⁷ and in obese people, even in the absence of obstructive sleep apnea.^48,49 Both groups are also at higher risk of diabetes.⁵⁰ One wonders if the decreased slow-wave sleep could in part contribute to the risk of diabetes in these groups.

A few studies specifically tested the effect of experimental suppression of slow-wave sleep on glucose homeostasis.

Tasali et al⁵¹ evaluated nine young, lean, nondiabetic men and women after 2 consecutive nights of undisturbed sleep and after 3 consecutive nights of suppressed slow-wave sleep without a change in total sleep duration or in REM sleep duration. Slow-wave sleep was disturbed by “delivering acoustic stimuli of various frequencies and intensities” whenever the subjects started to go into stage 3 or stage 4 sleep. This decreased the amount of slow-wave sleep by nearly 90%, which is comparable to the degree of sleep fragmentation seen in moderate to severe obstructive sleep apnea. After 3 nights of slow-wave sleep suppression, insulin sensitivity decreased by 25%, without a compensatory increase in insulin release, which resulted in a reduction in glucose tolerance of 23%, a value seen in older adults with impaired glucose tolerance.⁵²

Stamatakis et al⁵³ confirmed these findings in a similar study of 11 healthy, normal volunteers whose sleep was fragmented for 2 nights across all stages of sleep using auditory and mechanical stimuli. Insulin sensitivity significantly decreased, as did glucose effectiveness (ability of glucose to dispose itself independently of an insulin response) after the 2 nights of disturbed sleep quality.

These results support the hypothesis that poor sleep quality with short durations of slow-wave sleep, as seen with aging and obesity, could contribute to the higher risk of type 2 diabetes in these populations. These data also suggest that more studies are needed to look at the relationship between amount and quality of slow-wave sleep and diabetes risk.

The most robust evidence that not only short sleep duration but also poor sleep quality affects diabetes risk comes from studies of metabolic function in patients with obstructive sleep apnea, an increasingly common condition.

Obstructive sleep apnea is characterized by recurrent episodes of partial or complete upper airway obstruction with intermittent hypoxia and microarousals, resulting in low amounts of slow-wave sleep and overall decreased sleep quality.⁵⁴

Obstructive sleep apnea is common in patients with type 2 diabetes, and several clinical and epidemiologic studies suggest that, untreated, it may worsen diabetes risk or control.^{21,45–46,55–59}

The Sleep AHEAD (Action for Health in Diabetes) study⁶⁰ revealed, in cross-sectional data, that more than 84% of obese patients with type 2 diabetes had obstructive sleep apnea (with an apnea-hypopnea index ≥ 5).

Aronsohn et al,⁵ in a study conducted in our laboratory in 60 patients with type 2 diabetes, found that 46 (77%) of them had obstructive sleep apnea. Furthermore, the worse the obstructive sleep apnea, the worse the glucose control. After controlling for age, sex, race, body mass index, number of diabetes medications, level of exercise, years of diabetes, and total sleep time, compared with patients without obstructive sleep apnea, the adjusted mean hemoglobin A_1c was increased in a linear trend by (in absolute percentage points):

• 1.49% in patients with mild obstructive sleep apnea (P = .0028)
• 1.93% in patients with moderate obstructive sleep apnea (P = .0033)
• 3.69% in patients with severe obstructive sleep apnea (P < .0001).

Other epidemiologic studies. A growing number of epidemiologic studies, in various geographic regions, have suggested an independent link between obstructive sleep apnea and risk of type 2 diabetes.⁶¹ Most of the studies have been cross-sectional, and while most had positive findings, a criticism is that the methodology varied among the studies, both in how obstructive sleep apnea was assessed (snoring vs polysomnography) and in the metabolic assessment (oral glucose tolerance test, homeostatic model assessment, hemoglobin A_1c, medical history, physician examination, or patient report).

A more recent prospective study of 544 nondiabetic patients⁶⁵ showed that the risk of developing type 2 diabetes over an average of 2.7 years of follow-up was a function of the severity of obstructive sleep apnea expressed in quartiles: for each increased quartile of severity there was a 43% increase in the incidence of diabetes. Additionally, in patients with moderate to severe sleep apnea, regular use of continuous positive airway pressure (CPAP) was associated with an attenuated risk.⁶⁵

Two prospective studies (not included in Table 1) used snoring as a marker of obstructive sleep apnea; at 10 years of follow-up, snoring was associated with a higher risk of developing diabetes in both men and women.^73,74

Does CPAP improve glucose metabolism? Other studies have specifically examined the effects of CPAP treatment on glucose metabolism, in both diabetic and nondiabetic populations. Accumulating evidence suggests that metabolic abnormalities can be partially corrected by CPAP treatment, which supports the concept of a causal link between obstructive sleep apnea and altered glucose control. This topic is beyond the scope of this review; please see previously published literature^61,75 for further information. Whether treating obstructive sleep apnea may delay the development or reduce the severity of type 2 diabetes is another important unanswered question.

Is obstructive sleep apnea a cause or consequence of diabetes? It may be a novel risk factor for type 2 diabetes, and its association with altered glucose metabolism is well supported by a large set of cross-sectional studies, but there are still insufficient longitudinal studies to indicate a direction of causality.

If obstructive sleep apnea is the cause, what is the mechanism? There are likely many. High levels of sympathetic nervous system activity, intermittent hypoxia, sleep fragmentation, and sleep loss in obstructive sleep apnea may all lead to dysregulation of the hypothalamic-pituitary axis, endothelial dysfunction, and alterations in cytokine and adipokine release and are all potential mechanisms of abnormal glucose metabolism in this population.

WHAT TO TELL PATIENTS

Taken together, the current evidence suggests that strategies to improve the duration and the quality of sleep should be considered as a potential intervention to prevent or delay the development of type 2 diabetes mellitus in at-risk populations. While further studies are needed to better elucidate the mechanisms of the relationship between sleep loss and diabetes risk and to determine if extending sleep and treating obstructive sleep apnea decreases the risk of diabetes, we urge clinicians to recommend at least 7 hours of uninterrupted sleep per night as a goal in maintaining a healthy lifestyle. Additionally, clinicians should systematically evaluate the risk of obstructive sleep apnea in their patients who have type 2 diabetes mellitus and the metabolic syndrome, and conversely, should assess for diabetes in patients with known obstructive sleep apnea.

Adults are sleeping less and less in our society. Yet sleep is no longer thought of as strictly a restorative process for the body. The importance of sleep for metabolic function and specifically glucose homeostasis is now widely accepted, as many studies have shown a correlation between sleep deprivation or poor sleep quality and an increased risk of diabetes.

Obesity and aging are both associated with worse sleep. As the prevalence of obesity and diabetes increases, and as the number of elderly people increases, it is imperative to target sleep in the overall treatment of our patients.

In the pages that follow, we examine the evidence of a link between sleep loss (both short sleep duration and poor-quality sleep) and the risk of diabetes. (For evidence linking short sleep duration and the related problem of obesity, we invite the reader to refer to previous publications on the topic.^1,2)

SLEEP LOSS, OBESITY, AND DIABETES ARE ALL ON THE RISE

The prevalence of obesity and, consequently, of type 2 diabetes mellitus has increased alarmingly worldwide and particularly in the United States in the past few decades. Such a rapid increase cannot be explained simply by an alteration in the genetic pool; it is more likely due to environmental, socioeconomic, behavioral, and demographic factors and the interaction between genetics and these factors. Besides traditional lifestyle factors such as high-calorie diets and sedentary habits, other, nontraditional behavioral and environmental factors could be contributing to the epidemic of obesity and diabetes.

At the same time, people are sleeping less, and sleep disorders are on the rise. According to recent polls from the US Centers for Disease Control and Prevention, approximately 29% of US adults report sleeping less than 7 hours per night, and 50 to 70 million have chronic sleep and wakefulness disorders.³

The sleep curtailment of our times probably is partly self-imposed, as the pace and the opportunities of modern society place more demands on time for work and leisure activities and leave less time for sleep.

The quality of sleep has also declined as the population has aged and as the prevalence of obesity and its related sleep disorders has increased. Furthermore, patients with type 2 diabetes tend to sleep less, and to sleep poorly.^4,5 Poor sleep quality generally results in overall sleep loss.

GLUCOSE TOLERANCE HAS A CIRCADIAN RHYTHM

The human body regulates blood levels of glucose within a narrow range.

Glucose tolerance refers to the ability to maintain euglycemia by disposing of exogenous glucose via insulin-mediated and non–insulin-mediated mechanisms. Normal glucose tolerance depends on the ability of the pancreatic beta cells to produce insulin. As insulin sensitivity declines, insulin secretion increases to maintain normal glucose levels. Diabetes becomes manifest when the pancreatic beta cells fail to compensate for the decreased insulin sensitivity.

Glucose tolerance varies in a circadian rhythm, including during the different stages of sleep.

HOW SLEEP AFFECTS METABOLISM AND HORMONES

Sleep has often been thought of as a “restorative” process for the mind and the body; however, many studies have shown that it also directly affects many metabolic and hormonal processes.⁶

Sleep has five stages: rapid eye movement (REM) sleep and stages 1, 2, 3, and 4 of non-REM sleep. The deeper stages of non-REM sleep, ie, stages 3 and 4, are also known as slow-wave sleep and are thought to be the most restorative.

Additionally, the onset of slow-wave sleep is temporally associated with transient metabolic, hormonal, and neurophysiologic changes, all of which can affect glucose homeostasis. The brain uses less glucose,⁷ the pituitary gland releases more growth hormone and less corticotropin,⁸ the sympathetic nervous system is less active, and conversely, vagal tone is increased.⁹

As a result, in the first part of the night, when slow-wave sleep predominates, glucose metabolism is slower. These effects are reversed in the second part of the night, when REM sleep, stage 1, and awakening are more likely.

In view of these important changes in glucose metabolism during sleep, it is not surprising that getting less sleep or poorer sleep on a regular basis could affect overall glucose homeostasis.

SHORT SLEEP DURATION AND RISK OF DIABETES

Laboratory and epidemiologic evidence supports an association between short sleep duration (< 7 hours per night) and the risk of diabetes, and also between poor sleep quality and the risk of diabetes. We will explore putative mechanisms for these relationships.

Laboratory studies of short sleep duration and glucose metabolism

Studies in small numbers of healthy volunteers who underwent experimental sleep restriction or disruption have revealed mechanisms by which sleep loss might increase the risk of diabetes.

Kuhn et al¹⁰ performed the very first laboratory study of the effect of sleep deprivation on metabolism. Published in 1969, it showed that total sleep deprivation led to a marked increase in glucose levels.

A caution in extrapolating such results to real-life conditions is that total sleep deprivation is uncommon in humans and is inevitably followed by sleep recovery, with normalization of glucose metabolism. However, people in modern society are experiencing recurrent partial sleep deprivation, and its effect on glucose metabolism may be different.

Spiegel et al,¹¹ in landmark laboratory studies of partial sleep deprivation in healthy, lean adults, found that restricting sleep to 4 hours per night for 6 nights resulted in a 40% decrease in glucose tolerance, to levels similar to those seen in older adults with impaired glucose tolerance. This metabolic change was paralleled by an increase in the activity of the sympathetic nervous system, and both of these effects reversed with sleep recovery.

A criticism of these initial studies is that they restricted sleep to 4 hours, a restriction more severe than that seen in real life.

Nedeltcheva et al¹² more recently examined the effects of less-severe sleep curtailment (5.5 hours per night for 14 nights) in sedentary middle-aged men and women. This degree of bedtime restriction led to a decrease in glucose tolerance due to decreased insulin sensitivity in the absence of adequate beta cell compensation.

Such recurrent bedtime restriction is closer to the short sleep duration experienced by many people in everyday life, and in people at risk it may facilitate the development of insulin resistance, reduced glucose tolerance, and ultimately diabetes. Indeed, epidemiologic studies suggest that people who sleep less than 6 hours per night are at higher risk of type 2 diabetes.

Multiple cross-sectional epidemiologic studies have suggested an association between short sleep duration and diabetes, and several prospective epidemiologic studies have suggested that short sleep actually plays a causative role in diabetes.

The landmark observations of Spiegel et al¹¹ led to a number of epidemiologic studies examining the relationships between sleep duration and sleep disturbances and diabetes risk.¹³

The Sleep Heart Study¹⁴ was a large, cross-sectional, community-based study of the cardiovascular consequences of sleep-disordered breathing. The authors assessed the relationship between reported sleep duration and impaired glucose tolerance or type 2 diabetes in more than 1,400 men and women who had no history of insomnia. After adjustment for age, sex, race, body habitus, and apnea-hypopnea index, the prevalence of impaired glucose tolerance and type 2 diabetes was higher in those who reported sleeping 6 hours or less per night—or 9 hours or more per night (more below about the possible effect of too much sleep on the risk of diabetes).

The major limitations of the study were that it was cross-sectional in design, sleep duration was self-reported, the reasons for sleep curtailment were unknown, and possible confounding variables as physical activity, diet, and socioeconomic status were not measured.

Knutson et al,⁴ in our medical center, examined the association between self-reported sleep duration and sleep quality on the one hand and hemoglobin A_1c levels on the other in 161 black patients with type 2 diabetes. In patients without diabetic complications, glycemic control correlated with perceived sleep debt (calculated as the difference between self-reported actual and preferred weekday sleep duration); the authors calculated that a perceived sleep debt of 3 hours per night predicted a hemoglobin A_1c value 1.1 absolute percentage points higher than the median value. The analyses controlled for age, sex, body mass index, insulin use, and the presence of major complications; it excluded patients whose sleep was frequently disrupted by pain. The effect size was comparable to (but opposite) that of oral antidiabetic drugs. However, the direction of causality cannot be confirmed from this association, as it is possible that poor glycemic control in diabetic patients could impair their ability to achieve sufficient sleep.

To date, several major prospective studies have looked at the association between short sleep duration and sleep problems and the risk of developing type 2 diabetes in adults.

The Nurses Health Study¹⁵ followed 70,000 nondiabetic women for 10 years. Compared with nurses who slept 7 to 8 hours per 24 hours, those who slept 5 hours or less had a relative risk of diabetes of 1.34 even after controlling for many covariables, such as body mass index, shift work, hypertension, exercise, and depression.

The first National Health and Nutrition Examination Survey (NHANES I)¹⁶ examined the effect of sleep duration on the risk of incident diabetes in roughly 9,000 men and women over a period of 8 to 10 years. The statistical model included body mass index and hypertension and adjusted for physical activity, depression, alcohol consumption, ethnicity, education, marital status, and age. Findings: those who slept 5 hours or less per night were significantly more likely to develop type 2 diabetes than were those who slept 7 hours per night (odds ratio 1.57, 95% confidence interval [CI] 1.11–2.22), and so were those who slept 9 or more hours per night (odds ratio 1.57, 95% CI 1.10–2.24).

Kawakami et al¹⁷ followed 2,649 Japanese men for 8 years. Those who had difficulty going to sleep and staying asleep, which are both likely to result in shorter sleep duration, had higher age-adjusted risks of developing type 2 diabetes, with hazard ratios of 2.98 and 2.23, respectively.

Björkelund et al¹⁸ followed 6,599 nondiabetic Swedish men for an average of 15 years. Self-reported difficulty sleeping predicted the development of diabetes with an odds ratio of 1.52 even after controlling for age, body mass index at screening, changes in body mass index at follow-up, baseline glucose level, follow-up time, physical activity, family history of type 2 diabetes, smoking, social class, and alcohol intake.¹⁹

Interestingly, the authors found that the resting heart rate was higher at baseline in the men who later developed diabetes. This finding could be interpreted as reflecting greater sympathetic nervous system activity, a putative mediator of the metabolic dysfunction associated with both short sleep duration and obstructive sleep apnea.^20,21

Meisinger et al,²² in a study of more than 8,000 nondiabetic German men and women 25 to 74 years old, found a hazard ratio of developing diabetes of 1.60 (95% CI 1.05–2.45) in men and 1.98 (95% CI 1.20–3.29) in women who reported difficulty staying asleep, who thus would have shortened sleep duration. This effect was independent of other risk factors for diabetes.

Yaggi et al,²³ in a prospective study of 1,139 US men, also found a U-shaped relationship between sleep duration and the incidence of diabetes, with higher rates in people who slept less than 5 or more than 8 hours per night.

Cappuccio et al²⁴ performed a meta-analysis of all the prospective studies published to date. Their review included 10 prospective studies, with 107,756 participants followed for a median of 9.5 years. Sleep duration and sleep disturbances were self-reported in all the studies. They calculated that the risk of developing diabetes was 28% higher with short sleep duration (≤ 5 or < 6 hours in the different studies), 48% higher with long sleep duration (> 8 hours), 57% higher with difficulty going to sleep, and 84% higher with difficulty staying asleep.

Limitations of these studies. A consideration when trying to interpret the relationship between length of sleep and the incidence of diabetes is that sleep duration in these studies was self-reported, not measured. If a patient reports sleeping more than 8 hours per night, it could mean that he or she is not truly getting so much sleep, but rather is spending more time in bed trying to sleep.

Another possibility is that the higher incidence of type 2 diabetes in people who slept longer is due to undiagnosed obstructive sleep apnea, which is associated with daytime sleepiness and possibly longer sleep time to compensate for inefficient sleep.

Finally, depressive symptoms, unemployment, a low level of physical activity, and undiagnosed health conditions have all been associated with long sleep duration and could affect the relationship with diabetes risk.

In summary, epidemiologic studies from different geographic locations have consistently indicated that short sleep or poor sleep may increase the risk of developing type 2 diabetes mellitus and suggest that such an association spans different countries, cultures, and ethnic groups.

Therefore, there is a need for additional prospective epidemiologic studies that use objective measures of sleep. Furthermore, studies need to determine whether the cause of sleep restriction (eg, insomnia vs lifestyle choice) affects this relationship. Randomized, controlled, interventional studies would also be useful to determine whether lengthening sleep duration affects the development of impaired glucose tolerance or type 2 diabetes mellitus.

The effects of sleep loss on glucose metabolism are likely multifactorial, involving several interacting pathways.

Decreased brain glucose utilization has been shown on positron emission tomography in sleep-deprived subjects.²⁵

Hormonal dysregulation. Sleep deprivation is associated with disturbances in the secretion of the counterregulatory hormones growth hormone²⁶ and cortisol.¹¹

Young, healthy volunteers who were allowed to sleep only 4 hours per night for 6 nights showed a change in their patterns of growth hormone release, from a normal single pulse to a biphasic pattern.²⁶ They were exposed to a higher overall amount of growth hormone in the sleep-deprived condition, which could contribute to higher glucose levels.

Also, evening cortisol levels were significantly higher in young, healthy men who were allowed to sleep only 4 hours per night for 6 nights,¹¹ as well as in young, healthy women who were allowed to sleep only 3 hours for 1 night.²⁷ A cross-sectional analysis that included 2,751 men and women also demonstrated that short sleep duration and sleep disturbances are independently associated with more cortisol secretion in the evening.²⁸ Elevated evening cortisol levels can lead to morning insulin resistance.²⁹

Inflammation. Levels of inflammatory cytokines, inflammation, or both increase as sleep duration decreases, which in turn can also increase insulin resistance.^30,31

Sympathetic nervous system activity. Patients who have been sleep-deprived have been shown to have higher sympathetic nervous system activity, lower parasympathetic activity, or both.^11,32 The sympathetic nervous system inhibits insulin release while the parasympathetic system stimulates it, so these changes both increase glucose levels.³³ Moreover, overactivity of the sympathetic nervous system results in insulin resistance.³⁴

Excess weight is a well-established risk factor for type 2 diabetes mellitus, and several epidemiologic studies have suggested that sleep loss may increase the risk of becoming overweight or obese,^1,35 which would ultimately increase the risk of type 2 diabetes.

A primary mechanism linking sleep deprivation and weight gain is likely to be hyperactivity of the orexin system. Orexigenic neurons play a central role in wakefulness, but, as suggested by the name, they also promote feeding.³⁶ Studies in animals have indicated that the orexin system is overactive during sleep deprivation,^37–39 and this could be in part mediated by the increase in sympathetic activity.

Increased sympathetic activity also affects the levels of peripheral appetite hormones, inhibiting leptin release⁴⁰ and stimulating ghrelin release.⁴¹ Lower leptin levels and higher ghrelin levels act in concert to further activate orexin neurons,^42,43 resulting in increased food intake.

One could also argue that less time sleeping also allows more opportunity to eat.⁴⁴

Reduced energy expenditure. Sleep loss and its associated sleepiness and fatigue may result in reduced energy expenditure, partly due to less exercise but also due to less nonexercise activity thermogenesis. To date, reduced energy expenditure is an unexplored pathway that could link short sleep, the risk of obesity, and ultimately diabetes. In many overweight and obese people, this cascade of negative events is likely to be accelerated by sleep-disordered breathing, a reported independent risk factor for insulin resistance.^45,46

SLEEP QUALITY AND THE RISK OF DIABETES

Slow-wave sleep and diabetes

Slow-wave sleep, the most restorative sleep, is associated with metabolic, hormonal, and neurophysiologic changes that affect glucose homeostasis. Its disturbance may have deleterious effects on glucose tolerance.

Shallow slow-wave sleep occurs in elderly people⁴⁷ and in obese people, even in the absence of obstructive sleep apnea.^48,49 Both groups are also at higher risk of diabetes.⁵⁰ One wonders if the decreased slow-wave sleep could in part contribute to the risk of diabetes in these groups.

A few studies specifically tested the effect of experimental suppression of slow-wave sleep on glucose homeostasis.

Tasali et al⁵¹ evaluated nine young, lean, nondiabetic men and women after 2 consecutive nights of undisturbed sleep and after 3 consecutive nights of suppressed slow-wave sleep without a change in total sleep duration or in REM sleep duration. Slow-wave sleep was disturbed by “delivering acoustic stimuli of various frequencies and intensities” whenever the subjects started to go into stage 3 or stage 4 sleep. This decreased the amount of slow-wave sleep by nearly 90%, which is comparable to the degree of sleep fragmentation seen in moderate to severe obstructive sleep apnea. After 3 nights of slow-wave sleep suppression, insulin sensitivity decreased by 25%, without a compensatory increase in insulin release, which resulted in a reduction in glucose tolerance of 23%, a value seen in older adults with impaired glucose tolerance.⁵²

Stamatakis et al⁵³ confirmed these findings in a similar study of 11 healthy, normal volunteers whose sleep was fragmented for 2 nights across all stages of sleep using auditory and mechanical stimuli. Insulin sensitivity significantly decreased, as did glucose effectiveness (ability of glucose to dispose itself independently of an insulin response) after the 2 nights of disturbed sleep quality.

These results support the hypothesis that poor sleep quality with short durations of slow-wave sleep, as seen with aging and obesity, could contribute to the higher risk of type 2 diabetes in these populations. These data also suggest that more studies are needed to look at the relationship between amount and quality of slow-wave sleep and diabetes risk.

The most robust evidence that not only short sleep duration but also poor sleep quality affects diabetes risk comes from studies of metabolic function in patients with obstructive sleep apnea, an increasingly common condition.

Obstructive sleep apnea is characterized by recurrent episodes of partial or complete upper airway obstruction with intermittent hypoxia and microarousals, resulting in low amounts of slow-wave sleep and overall decreased sleep quality.⁵⁴

Obstructive sleep apnea is common in patients with type 2 diabetes, and several clinical and epidemiologic studies suggest that, untreated, it may worsen diabetes risk or control.^{21,45–46,55–59}

The Sleep AHEAD (Action for Health in Diabetes) study⁶⁰ revealed, in cross-sectional data, that more than 84% of obese patients with type 2 diabetes had obstructive sleep apnea (with an apnea-hypopnea index ≥ 5).

Aronsohn et al,⁵ in a study conducted in our laboratory in 60 patients with type 2 diabetes, found that 46 (77%) of them had obstructive sleep apnea. Furthermore, the worse the obstructive sleep apnea, the worse the glucose control. After controlling for age, sex, race, body mass index, number of diabetes medications, level of exercise, years of diabetes, and total sleep time, compared with patients without obstructive sleep apnea, the adjusted mean hemoglobin A_1c was increased in a linear trend by (in absolute percentage points):

• 1.49% in patients with mild obstructive sleep apnea (P = .0028)
• 1.93% in patients with moderate obstructive sleep apnea (P = .0033)
• 3.69% in patients with severe obstructive sleep apnea (P < .0001).

Other epidemiologic studies. A growing number of epidemiologic studies, in various geographic regions, have suggested an independent link between obstructive sleep apnea and risk of type 2 diabetes.⁶¹ Most of the studies have been cross-sectional, and while most had positive findings, a criticism is that the methodology varied among the studies, both in how obstructive sleep apnea was assessed (snoring vs polysomnography) and in the metabolic assessment (oral glucose tolerance test, homeostatic model assessment, hemoglobin A_1c, medical history, physician examination, or patient report).

A more recent prospective study of 544 nondiabetic patients⁶⁵ showed that the risk of developing type 2 diabetes over an average of 2.7 years of follow-up was a function of the severity of obstructive sleep apnea expressed in quartiles: for each increased quartile of severity there was a 43% increase in the incidence of diabetes. Additionally, in patients with moderate to severe sleep apnea, regular use of continuous positive airway pressure (CPAP) was associated with an attenuated risk.⁶⁵

Two prospective studies (not included in Table 1) used snoring as a marker of obstructive sleep apnea; at 10 years of follow-up, snoring was associated with a higher risk of developing diabetes in both men and women.^73,74

Does CPAP improve glucose metabolism? Other studies have specifically examined the effects of CPAP treatment on glucose metabolism, in both diabetic and nondiabetic populations. Accumulating evidence suggests that metabolic abnormalities can be partially corrected by CPAP treatment, which supports the concept of a causal link between obstructive sleep apnea and altered glucose control. This topic is beyond the scope of this review; please see previously published literature^61,75 for further information. Whether treating obstructive sleep apnea may delay the development or reduce the severity of type 2 diabetes is another important unanswered question.

Is obstructive sleep apnea a cause or consequence of diabetes? It may be a novel risk factor for type 2 diabetes, and its association with altered glucose metabolism is well supported by a large set of cross-sectional studies, but there are still insufficient longitudinal studies to indicate a direction of causality.

If obstructive sleep apnea is the cause, what is the mechanism? There are likely many. High levels of sympathetic nervous system activity, intermittent hypoxia, sleep fragmentation, and sleep loss in obstructive sleep apnea may all lead to dysregulation of the hypothalamic-pituitary axis, endothelial dysfunction, and alterations in cytokine and adipokine release and are all potential mechanisms of abnormal glucose metabolism in this population.

WHAT TO TELL PATIENTS

Taken together, the current evidence suggests that strategies to improve the duration and the quality of sleep should be considered as a potential intervention to prevent or delay the development of type 2 diabetes mellitus in at-risk populations. While further studies are needed to better elucidate the mechanisms of the relationship between sleep loss and diabetes risk and to determine if extending sleep and treating obstructive sleep apnea decreases the risk of diabetes, we urge clinicians to recommend at least 7 hours of uninterrupted sleep per night as a goal in maintaining a healthy lifestyle. Additionally, clinicians should systematically evaluate the risk of obstructive sleep apnea in their patients who have type 2 diabetes mellitus and the metabolic syndrome, and conversely, should assess for diabetes in patients with known obstructive sleep apnea.

In Reply: We know from autopsy studies that most patients with giant cell arteritis, if not all, develop aortitis at some point during the course of their disease, but we don’t know (and no study yet has completely addressed) the following questions:

• What is the most clinically appropriate and cost-effective method of screening?
• How often should we be screening these patients?

Given the high cost of the most accurate and detailed available test, ie, magnetic resonance angiography of the aorta, annual chest radiography has been recommended by some experts in the field.

Although the high frequency of thoracic aneurysm justifies high clinical vigilance, we don’t know the most adequate and cost-effective test for screening for aortic aneurysm. Until we have an answer to these questions it is difficult to formulate specific guidelines, and different experts will continue to have different practices that are based on their own experience.

At this time, I carefully listen for bruits and murmurs on physical examination and check the blood pressure in all four extremities during patient follow-up visits. If I detect any abnormalities suggesting pathology of the aorta or major branches, I order magnetic resonance angiography of the entire aorta and its main branches.

In Reply: We know from autopsy studies that most patients with giant cell arteritis, if not all, develop aortitis at some point during the course of their disease, but we don’t know (and no study yet has completely addressed) the following questions:

• What is the most clinically appropriate and cost-effective method of screening?
• How often should we be screening these patients?

Given the high cost of the most accurate and detailed available test, ie, magnetic resonance angiography of the aorta, annual chest radiography has been recommended by some experts in the field.

Although the high frequency of thoracic aneurysm justifies high clinical vigilance, we don’t know the most adequate and cost-effective test for screening for aortic aneurysm. Until we have an answer to these questions it is difficult to formulate specific guidelines, and different experts will continue to have different practices that are based on their own experience.

At this time, I carefully listen for bruits and murmurs on physical examination and check the blood pressure in all four extremities during patient follow-up visits. If I detect any abnormalities suggesting pathology of the aorta or major branches, I order magnetic resonance angiography of the entire aorta and its main branches.

In Reply: We know from autopsy studies that most patients with giant cell arteritis, if not all, develop aortitis at some point during the course of their disease, but we don’t know (and no study yet has completely addressed) the following questions:

• What is the most clinically appropriate and cost-effective method of screening?
• How often should we be screening these patients?

Given the high cost of the most accurate and detailed available test, ie, magnetic resonance angiography of the aorta, annual chest radiography has been recommended by some experts in the field.

Although the high frequency of thoracic aneurysm justifies high clinical vigilance, we don’t know the most adequate and cost-effective test for screening for aortic aneurysm. Until we have an answer to these questions it is difficult to formulate specific guidelines, and different experts will continue to have different practices that are based on their own experience.

At this time, I carefully listen for bruits and murmurs on physical examination and check the blood pressure in all four extremities during patient follow-up visits. If I detect any abnormalities suggesting pathology of the aorta or major branches, I order magnetic resonance angiography of the entire aorta and its main branches.

To the Editor: As a practicing internist, I found Dr. Alexandra Villa-Forte’s review of giant-cell arteritis (Cleve Clin J Med 2011; 78:265–270) both interesting and useful, as usual for the Cleveland Clinic Journal of Medicine. However, she did not mention the recommendation by some experts that patients who have had temporal arteritis should receive annual chest x-rays, for a decade or longer, to screen for the development of thoracic aortic aneurysm. Does she agree with this precaution? Is it advisable, in addition, to screen for abdominal aortic aneurysm by means of abdominal ultrasonography? If so, at what time intervals should this be done?

To the Editor: As a practicing internist, I found Dr. Alexandra Villa-Forte’s review of giant-cell arteritis (Cleve Clin J Med 2011; 78:265–270) both interesting and useful, as usual for the Cleveland Clinic Journal of Medicine. However, she did not mention the recommendation by some experts that patients who have had temporal arteritis should receive annual chest x-rays, for a decade or longer, to screen for the development of thoracic aortic aneurysm. Does she agree with this precaution? Is it advisable, in addition, to screen for abdominal aortic aneurysm by means of abdominal ultrasonography? If so, at what time intervals should this be done?

To the Editor: As a practicing internist, I found Dr. Alexandra Villa-Forte’s review of giant-cell arteritis (Cleve Clin J Med 2011; 78:265–270) both interesting and useful, as usual for the Cleveland Clinic Journal of Medicine. However, she did not mention the recommendation by some experts that patients who have had temporal arteritis should receive annual chest x-rays, for a decade or longer, to screen for the development of thoracic aortic aneurysm. Does she agree with this precaution? Is it advisable, in addition, to screen for abdominal aortic aneurysm by means of abdominal ultrasonography? If so, at what time intervals should this be done?

A 72-year-old man with a 15-year history of a heart murmur presents to his cardiologist with shortness of breath on exertion over the past 12 months. He otherwise feels well and reports no chest discomfort, palpitations, or swelling of his legs or feet. He is not taking any cardiac drugs, and his health has previously been excellent.

Q: Which of the following findings on 12-lead ECG is not commonly reported in chronic severe aortic regurgitation?

• Left ventricular hypertrophy
• QRS complex left-axis deviation
• A negative U wave
• Atrial fibrillation

A: The correct answer is a negative U wave.

In long-standing left ventricular volume overload, such as in chronic aortic regurgitation, characteristic findings on ECG include lateral precordial narrow Q waves and left ventricular hypertrophy. The ST segment and T wave are often normal or nearly normal. The QRS complex vector may demonstrate left-axis deviation, but this is not absolute. In contrast, pressure overload conditions such as aortic stenosis and systemic hypertension commonly manifest as left ventricular hypertrophy with strain pattern of ST depression in lateral precordial leads and asymmetric T-wave inversion.

A negative U wave, best identified in leads V₄ to V₆, is a common finding in left ventricular volume overload. A negative U wave represents a negative deflection of small amplitude (normally < 0.1 to 3 mV) immediately following the T wave. Although not routinely reported, the negative U wave is an indicator of underlying structural heart disease.¹

Q: A negative U wave has been associated with which of the following conditions?

• Aortic or mitral regurgitation
• Myocardial ischemia
• Hypertension
• All of the above

A: The correct answer is all of the above.

Negative U waves have been identified in regurgitant valvular heart disease with left ventricular volume overload, in myocardial ischemia, ^2,3 and in hypertension.⁴ During exercise stress testing, the transient appearance of negative U waves strongly suggests flow-limiting coronary artery disease. Moreover, changes in the U wave during exercise stress testing may be a sign of well-developed coronary collaterals.⁵ Therefore, it is prudent to note their presence on resting ECG and to investigate further with cardiac stress testing and imaging.

The pathogenesis of the negative U wave remains unclear. Of the various hypotheses put forth, a mechano-electric phenomenon may best explain its diverse pathology.

A 72-year-old man with a 15-year history of a heart murmur presents to his cardiologist with shortness of breath on exertion over the past 12 months. He otherwise feels well and reports no chest discomfort, palpitations, or swelling of his legs or feet. He is not taking any cardiac drugs, and his health has previously been excellent.

Q: Which of the following findings on 12-lead ECG is not commonly reported in chronic severe aortic regurgitation?

• Left ventricular hypertrophy
• QRS complex left-axis deviation
• A negative U wave
• Atrial fibrillation

A: The correct answer is a negative U wave.

In long-standing left ventricular volume overload, such as in chronic aortic regurgitation, characteristic findings on ECG include lateral precordial narrow Q waves and left ventricular hypertrophy. The ST segment and T wave are often normal or nearly normal. The QRS complex vector may demonstrate left-axis deviation, but this is not absolute. In contrast, pressure overload conditions such as aortic stenosis and systemic hypertension commonly manifest as left ventricular hypertrophy with strain pattern of ST depression in lateral precordial leads and asymmetric T-wave inversion.

A negative U wave, best identified in leads V₄ to V₆, is a common finding in left ventricular volume overload. A negative U wave represents a negative deflection of small amplitude (normally < 0.1 to 3 mV) immediately following the T wave. Although not routinely reported, the negative U wave is an indicator of underlying structural heart disease.¹

Q: A negative U wave has been associated with which of the following conditions?

• Aortic or mitral regurgitation
• Myocardial ischemia
• Hypertension
• All of the above

A: The correct answer is all of the above.

Negative U waves have been identified in regurgitant valvular heart disease with left ventricular volume overload, in myocardial ischemia, ^2,3 and in hypertension.⁴ During exercise stress testing, the transient appearance of negative U waves strongly suggests flow-limiting coronary artery disease. Moreover, changes in the U wave during exercise stress testing may be a sign of well-developed coronary collaterals.⁵ Therefore, it is prudent to note their presence on resting ECG and to investigate further with cardiac stress testing and imaging.

The pathogenesis of the negative U wave remains unclear. Of the various hypotheses put forth, a mechano-electric phenomenon may best explain its diverse pathology.

A 72-year-old man with a 15-year history of a heart murmur presents to his cardiologist with shortness of breath on exertion over the past 12 months. He otherwise feels well and reports no chest discomfort, palpitations, or swelling of his legs or feet. He is not taking any cardiac drugs, and his health has previously been excellent.

Q: Which of the following findings on 12-lead ECG is not commonly reported in chronic severe aortic regurgitation?

• Left ventricular hypertrophy
• QRS complex left-axis deviation
• A negative U wave
• Atrial fibrillation

A: The correct answer is a negative U wave.

In long-standing left ventricular volume overload, such as in chronic aortic regurgitation, characteristic findings on ECG include lateral precordial narrow Q waves and left ventricular hypertrophy. The ST segment and T wave are often normal or nearly normal. The QRS complex vector may demonstrate left-axis deviation, but this is not absolute. In contrast, pressure overload conditions such as aortic stenosis and systemic hypertension commonly manifest as left ventricular hypertrophy with strain pattern of ST depression in lateral precordial leads and asymmetric T-wave inversion.

A negative U wave, best identified in leads V₄ to V₆, is a common finding in left ventricular volume overload. A negative U wave represents a negative deflection of small amplitude (normally < 0.1 to 3 mV) immediately following the T wave. Although not routinely reported, the negative U wave is an indicator of underlying structural heart disease.¹

Q: A negative U wave has been associated with which of the following conditions?

• Aortic or mitral regurgitation
• Myocardial ischemia
• Hypertension
• All of the above

A: The correct answer is all of the above.

Negative U waves have been identified in regurgitant valvular heart disease with left ventricular volume overload, in myocardial ischemia, ^2,3 and in hypertension.⁴ During exercise stress testing, the transient appearance of negative U waves strongly suggests flow-limiting coronary artery disease. Moreover, changes in the U wave during exercise stress testing may be a sign of well-developed coronary collaterals.⁵ Therefore, it is prudent to note their presence on resting ECG and to investigate further with cardiac stress testing and imaging.

The pathogenesis of the negative U wave remains unclear. Of the various hypotheses put forth, a mechano-electric phenomenon may best explain its diverse pathology.

A study of twins finds that shared environmental factors influence the risk of autism more than previously thought and challenges previous findings about the significance of genetics.

Among identical and fraternal twins in whom at least one child has autism or autism spectrum disorder (ASD), shared environmental factors have a more substantial impact regarding development of the condition than do genetics, according to a study in the July 4 online Archives of General Psychiatry.
“A large proportion of the variance in liability can be explained by shared environmental factors (55% for autism and 58% for ASD) in addition to moderate genetic heritability (37% for autism and 38% for ASD),” reported Joachim Hallmayer, MD, Associate Professor of Psychiatry and Behavioral Sciences, Stanford University School of Medicine in Palo Alto, California, and colleagues. “Our study provides evidence that the rate of concordance in dizygotic twins may have been seriously underestimated in previous studies and the influence of genetic factors on the susceptibility to develop autism, overestimated.”
A Shift in the Environment Versus Genetics Debate?
The study included data from monozygotic twin pairs (45 male, nine female) and dizygotic twin pairs (45 male, 13 female, and 80 sex-discordant) who were born between 1987 and 2004. The monozygotic twins were slightly older and had shorter gestation periods. The mothers of the dizygotic twins were also older than the mothers of the monozygotic twins, “consistent with the known increase in dizygotic twinning with maternal age, and more likely to be white and non-Hispanic,” noted the investigators.
For twins with strict autism, the researchers found that the probandwise concordance for male twins was 0.58 for 40 monozygotic pairs and 0.21 for 31 dizygotic pairs; for female twins, concordance was 0.60 for seven monozygotic twin pairs and 0.27 for 10 dizygotic pairs. For children with ASD, the probandwise concordance for male twins was 0.77 for 45 monozygotic pairs and 0.31 for 45 dizygotic pairs; for female twins, concordance was 0.50 for nine monozygotic pairs and 0.36 for 13 dizygotic pairs.
“Because of the reported high heritability of autism, a major focus of research in autism has been on finding the underlying genetic causes, with less emphasis on potential environmental triggers or causes,” Dr. Hallmayer and colleagues wrote. “The finding of significant influence of the shared environment, experiences that are common to both twin individuals, may be important for future research paradigms.”
Increasing evidence has shown that overt symptoms of autism emerge toward the end of the first year of life, the authors noted. “Because the prenatal environment and early postnatal environment are shared between twin individuals, we hypothesize that at least some of the environmental factors impacting susceptibility to autism exert their effect during this critical period of life,” Dr. Hallmayer’s group commented. “Nongenetic risk factors that may index environmental influences included parental age, low birth weight, multiple births, and maternal infections during pregnancy. Future studies that seek to elucidate such factors and their role in enhancing or suppressing genetic susceptibility are likely to enhance our understanding of autism.”
A Disorder of Fetal Programming?
In an accompanying editorial, Peter Szatmari, MD, of the Offord Centre for Child Studies, McMaster University in Hamilton, Ontario, Canada, stated, “Perhaps ASD can be considered, at least in part, a disorder of fetal programming. There is in fact evidence that certain risk factors that affect the maternal fetal environment may place the fetus at increased risk for ASD. Clearly a renewed effort needs to be undertaken through the use of well-designed community-based epidemiologic studies.
“Whatever happens in the future, the finding by Hallmayer and colleagues is an extraordinarily important one and has the potential to shift autism research into a new field of study in much the same way that the original twin study by Folstein and Rutter accomplished back in 1977,” Dr. Szatmari concluded.

A Link Between Maternal Antidepressant Use and Autism Risk in Offspring?
Exposure to selective serotonin reuptake inhibitors (SSRIs) among pregnant women, especially during the first trimester, may modestly increase the risk of autism spectrum disorder (ASD) in their children, according to a study in the July 4 online Archives of General Psychiatry.
The findings were based on 298 children with ASD and 1,507 randomly selected control children and their mothers enrolled in the Kaiser Permanente Medical Care Program in Northern California. Data regarding prenatal exposure to antidepressants were available for 20 children and 50 controls. After adjusted logistic regression, the researchers found a twofold increased risk of ASD associated with SSRI treatment in mothers in the year before delivery (adjusted odds ratio, 2.2). The strongest effect was linked with treatment during the first trimester (adjusted odds ratio, 3.8). No increased risk was observed among mothers with a history of mental health treatment in the absence of prenatal exposure to SSRIs.
“The fraction of cases of ASD that may be attributed to use of antidepressants by the mother during pregnancy is less than 3% in our population, and it is reasonable to conclude that prenatal SSRI exposure is very unlikely to be a major risk factor for ASD,” stated Lisa A. Croen, PhD, of the Division of Research, Kaiser Permanente Northern California in Oakland, and colleagues. “Although these findings indicate that maternal treatment with SSRIs during pregnancy may confer some risk to the fetus with regard to neurodevelopment, this potential risk must be balanced with the risk to the mother or fetus of untreated mental health disorders.”
“Perhaps it is a coincidence that the odds ratio for ASD risk in the study by Croen and colleagues increases when first-trimester exposure to SSRIs is the sole factor,” stated Pat Levitt, PhD, of the Keck School of Medicine, University of Southern California, Los Angeles, in a related commentary. “However, it is exactly that time of human brain development during which cortical and subcortical neuronal populations are being produced, migrating to their final destinations and beginning the long process of wiring. While much occurs later, the establishment of a strong foundation developmentally may be an essential component of healthy brain development.”

A study of twins finds that shared environmental factors influence the risk of autism more than previously thought and challenges previous findings about the significance of genetics.

Among identical and fraternal twins in whom at least one child has autism or autism spectrum disorder (ASD), shared environmental factors have a more substantial impact regarding development of the condition than do genetics, according to a study in the July 4 online Archives of General Psychiatry.
“A large proportion of the variance in liability can be explained by shared environmental factors (55% for autism and 58% for ASD) in addition to moderate genetic heritability (37% for autism and 38% for ASD),” reported Joachim Hallmayer, MD, Associate Professor of Psychiatry and Behavioral Sciences, Stanford University School of Medicine in Palo Alto, California, and colleagues. “Our study provides evidence that the rate of concordance in dizygotic twins may have been seriously underestimated in previous studies and the influence of genetic factors on the susceptibility to develop autism, overestimated.”
A Shift in the Environment Versus Genetics Debate?
The study included data from monozygotic twin pairs (45 male, nine female) and dizygotic twin pairs (45 male, 13 female, and 80 sex-discordant) who were born between 1987 and 2004. The monozygotic twins were slightly older and had shorter gestation periods. The mothers of the dizygotic twins were also older than the mothers of the monozygotic twins, “consistent with the known increase in dizygotic twinning with maternal age, and more likely to be white and non-Hispanic,” noted the investigators.
For twins with strict autism, the researchers found that the probandwise concordance for male twins was 0.58 for 40 monozygotic pairs and 0.21 for 31 dizygotic pairs; for female twins, concordance was 0.60 for seven monozygotic twin pairs and 0.27 for 10 dizygotic pairs. For children with ASD, the probandwise concordance for male twins was 0.77 for 45 monozygotic pairs and 0.31 for 45 dizygotic pairs; for female twins, concordance was 0.50 for nine monozygotic pairs and 0.36 for 13 dizygotic pairs.
“Because of the reported high heritability of autism, a major focus of research in autism has been on finding the underlying genetic causes, with less emphasis on potential environmental triggers or causes,” Dr. Hallmayer and colleagues wrote. “The finding of significant influence of the shared environment, experiences that are common to both twin individuals, may be important for future research paradigms.”
Increasing evidence has shown that overt symptoms of autism emerge toward the end of the first year of life, the authors noted. “Because the prenatal environment and early postnatal environment are shared between twin individuals, we hypothesize that at least some of the environmental factors impacting susceptibility to autism exert their effect during this critical period of life,” Dr. Hallmayer’s group commented. “Nongenetic risk factors that may index environmental influences included parental age, low birth weight, multiple births, and maternal infections during pregnancy. Future studies that seek to elucidate such factors and their role in enhancing or suppressing genetic susceptibility are likely to enhance our understanding of autism.”
A Disorder of Fetal Programming?
In an accompanying editorial, Peter Szatmari, MD, of the Offord Centre for Child Studies, McMaster University in Hamilton, Ontario, Canada, stated, “Perhaps ASD can be considered, at least in part, a disorder of fetal programming. There is in fact evidence that certain risk factors that affect the maternal fetal environment may place the fetus at increased risk for ASD. Clearly a renewed effort needs to be undertaken through the use of well-designed community-based epidemiologic studies.
“Whatever happens in the future, the finding by Hallmayer and colleagues is an extraordinarily important one and has the potential to shift autism research into a new field of study in much the same way that the original twin study by Folstein and Rutter accomplished back in 1977,” Dr. Szatmari concluded.

A Link Between Maternal Antidepressant Use and Autism Risk in Offspring?
Exposure to selective serotonin reuptake inhibitors (SSRIs) among pregnant women, especially during the first trimester, may modestly increase the risk of autism spectrum disorder (ASD) in their children, according to a study in the July 4 online Archives of General Psychiatry.
The findings were based on 298 children with ASD and 1,507 randomly selected control children and their mothers enrolled in the Kaiser Permanente Medical Care Program in Northern California. Data regarding prenatal exposure to antidepressants were available for 20 children and 50 controls. After adjusted logistic regression, the researchers found a twofold increased risk of ASD associated with SSRI treatment in mothers in the year before delivery (adjusted odds ratio, 2.2). The strongest effect was linked with treatment during the first trimester (adjusted odds ratio, 3.8). No increased risk was observed among mothers with a history of mental health treatment in the absence of prenatal exposure to SSRIs.
“The fraction of cases of ASD that may be attributed to use of antidepressants by the mother during pregnancy is less than 3% in our population, and it is reasonable to conclude that prenatal SSRI exposure is very unlikely to be a major risk factor for ASD,” stated Lisa A. Croen, PhD, of the Division of Research, Kaiser Permanente Northern California in Oakland, and colleagues. “Although these findings indicate that maternal treatment with SSRIs during pregnancy may confer some risk to the fetus with regard to neurodevelopment, this potential risk must be balanced with the risk to the mother or fetus of untreated mental health disorders.”
“Perhaps it is a coincidence that the odds ratio for ASD risk in the study by Croen and colleagues increases when first-trimester exposure to SSRIs is the sole factor,” stated Pat Levitt, PhD, of the Keck School of Medicine, University of Southern California, Los Angeles, in a related commentary. “However, it is exactly that time of human brain development during which cortical and subcortical neuronal populations are being produced, migrating to their final destinations and beginning the long process of wiring. While much occurs later, the establishment of a strong foundation developmentally may be an essential component of healthy brain development.”

A study of twins finds that shared environmental factors influence the risk of autism more than previously thought and challenges previous findings about the significance of genetics.

Among identical and fraternal twins in whom at least one child has autism or autism spectrum disorder (ASD), shared environmental factors have a more substantial impact regarding development of the condition than do genetics, according to a study in the July 4 online Archives of General Psychiatry.
“A large proportion of the variance in liability can be explained by shared environmental factors (55% for autism and 58% for ASD) in addition to moderate genetic heritability (37% for autism and 38% for ASD),” reported Joachim Hallmayer, MD, Associate Professor of Psychiatry and Behavioral Sciences, Stanford University School of Medicine in Palo Alto, California, and colleagues. “Our study provides evidence that the rate of concordance in dizygotic twins may have been seriously underestimated in previous studies and the influence of genetic factors on the susceptibility to develop autism, overestimated.”
A Shift in the Environment Versus Genetics Debate?
The study included data from monozygotic twin pairs (45 male, nine female) and dizygotic twin pairs (45 male, 13 female, and 80 sex-discordant) who were born between 1987 and 2004. The monozygotic twins were slightly older and had shorter gestation periods. The mothers of the dizygotic twins were also older than the mothers of the monozygotic twins, “consistent with the known increase in dizygotic twinning with maternal age, and more likely to be white and non-Hispanic,” noted the investigators.
For twins with strict autism, the researchers found that the probandwise concordance for male twins was 0.58 for 40 monozygotic pairs and 0.21 for 31 dizygotic pairs; for female twins, concordance was 0.60 for seven monozygotic twin pairs and 0.27 for 10 dizygotic pairs. For children with ASD, the probandwise concordance for male twins was 0.77 for 45 monozygotic pairs and 0.31 for 45 dizygotic pairs; for female twins, concordance was 0.50 for nine monozygotic pairs and 0.36 for 13 dizygotic pairs.
“Because of the reported high heritability of autism, a major focus of research in autism has been on finding the underlying genetic causes, with less emphasis on potential environmental triggers or causes,” Dr. Hallmayer and colleagues wrote. “The finding of significant influence of the shared environment, experiences that are common to both twin individuals, may be important for future research paradigms.”
Increasing evidence has shown that overt symptoms of autism emerge toward the end of the first year of life, the authors noted. “Because the prenatal environment and early postnatal environment are shared between twin individuals, we hypothesize that at least some of the environmental factors impacting susceptibility to autism exert their effect during this critical period of life,” Dr. Hallmayer’s group commented. “Nongenetic risk factors that may index environmental influences included parental age, low birth weight, multiple births, and maternal infections during pregnancy. Future studies that seek to elucidate such factors and their role in enhancing or suppressing genetic susceptibility are likely to enhance our understanding of autism.”
A Disorder of Fetal Programming?
In an accompanying editorial, Peter Szatmari, MD, of the Offord Centre for Child Studies, McMaster University in Hamilton, Ontario, Canada, stated, “Perhaps ASD can be considered, at least in part, a disorder of fetal programming. There is in fact evidence that certain risk factors that affect the maternal fetal environment may place the fetus at increased risk for ASD. Clearly a renewed effort needs to be undertaken through the use of well-designed community-based epidemiologic studies.
“Whatever happens in the future, the finding by Hallmayer and colleagues is an extraordinarily important one and has the potential to shift autism research into a new field of study in much the same way that the original twin study by Folstein and Rutter accomplished back in 1977,” Dr. Szatmari concluded.

A Link Between Maternal Antidepressant Use and Autism Risk in Offspring?
Exposure to selective serotonin reuptake inhibitors (SSRIs) among pregnant women, especially during the first trimester, may modestly increase the risk of autism spectrum disorder (ASD) in their children, according to a study in the July 4 online Archives of General Psychiatry.
The findings were based on 298 children with ASD and 1,507 randomly selected control children and their mothers enrolled in the Kaiser Permanente Medical Care Program in Northern California. Data regarding prenatal exposure to antidepressants were available for 20 children and 50 controls. After adjusted logistic regression, the researchers found a twofold increased risk of ASD associated with SSRI treatment in mothers in the year before delivery (adjusted odds ratio, 2.2). The strongest effect was linked with treatment during the first trimester (adjusted odds ratio, 3.8). No increased risk was observed among mothers with a history of mental health treatment in the absence of prenatal exposure to SSRIs.
“The fraction of cases of ASD that may be attributed to use of antidepressants by the mother during pregnancy is less than 3% in our population, and it is reasonable to conclude that prenatal SSRI exposure is very unlikely to be a major risk factor for ASD,” stated Lisa A. Croen, PhD, of the Division of Research, Kaiser Permanente Northern California in Oakland, and colleagues. “Although these findings indicate that maternal treatment with SSRIs during pregnancy may confer some risk to the fetus with regard to neurodevelopment, this potential risk must be balanced with the risk to the mother or fetus of untreated mental health disorders.”
“Perhaps it is a coincidence that the odds ratio for ASD risk in the study by Croen and colleagues increases when first-trimester exposure to SSRIs is the sole factor,” stated Pat Levitt, PhD, of the Keck School of Medicine, University of Southern California, Los Angeles, in a related commentary. “However, it is exactly that time of human brain development during which cortical and subcortical neuronal populations are being produced, migrating to their final destinations and beginning the long process of wiring. While much occurs later, the establishment of a strong foundation developmentally may be an essential component of healthy brain development.”

A 46-year-old man presented to a hospital emergency department (ED) with a four-day history of right ear pain. He described the pain as a constant, dull, burning pain radiating to the neck and face, associated with a feeling of congestion. The patient also stated that the right side of his face had felt numb for about one day.

Three days earlier, the man had been seen by his primary health care provider, who told him that his ear looked normal and free of infection. The day before his current presentation to the ED, however, he noticed what he described as an “acne-like” rash on his ear lobe. Shortly before coming to the ED, the patient also developed numbness over his right upper lip, which he likened to the effects of procaine during a dental visit. He reported drooling from the right side of his mouth while drinking water and difficulty blinking his right eye. 

He denied any tinnitus, fever, headache, or change in hearing. A review of symptoms was positive only for mild dizziness during the previous two to three days.

The patient was a well-appearing white man. He was alert and oriented to identity, time, and place. His skin was warm, dry, and intact. The examiner noticed a small area of erythematous rash with vesicles on the man’s right ear lobe. The external auditory canals appeared within normal limits, with no erythema or edema, and were nontender bilaterally. The tympanic membranes were normal bilaterally, without bulging or discernible fluid levels.

The ocular exam was normal with no visual acuity changes and no fluorescein uptake; external ocular movements were intact. A slight droop was noted in the right eyelid, but there was no droop on the contralateral side of his face. When asked to puff up his cheeks, the patient found it difficult to do so on the right side of his mouth without releasing air from his lips.

The remainder of the cranial nerves were intact. Muscle strength was 5/5 in all extremities and equal bilaterally. The man’s gait was within normal limits, and the remaining findings in the physical exam were normal.

The initial diagnosis considered in the differential was otitis externa, because it is a common explanation for ear pain in patients who present to the ED.^1,2 Also, in otitis, pain is characteristically present in the affected ear, and erythema is often found in the external auditory canal.³ However, this diagnosis was deemed unlikely because otitis externa would not explain the neurologic findings or the vesicular rash.¹

Bell’s palsy was next in the differential, as it was considered consistent with the patient’s unilateral neurologic deficits.⁴ In addition to weakness or palsy of the facial nerve, many patients with Bell’s palsy complain of mastoid pain, which can be confused with a complaint of ear pain.⁵ However, patients with Bell’s palsy have no rash, and this diagnosis was considered unlikely.

The painful, burning rash on the patient’s face was characteristic of herpes zoster (shingles), which was next in the differential. Infrequently, shingles can also cause weakness in the nerve it affects. In the case patient, weakness that was evident in the affected nerve resembled that seen in Bell’s palsy. This combination of symptoms is referred to as Ramsay Hunt syndrome—which in this case was decided to be the correct diagnosis.

DISCUSSION
Ramsay Hunt syndrome (RHS, also known as geniculate herpes^5,6) is caused by the varicella-zoster virus, most commonly known as the cause of chickenpox. In the United States, RHS is believed to affect only about one in 1,500 persons, although 20% to 30% of persons experience herpes zoster infection at some time.⁷

Soon after a chickenpox infection subsides, the virus spreads along the sensory nerve fibers of the peripheral and cranial nerves. The virus then becomes dormant in the dorsal root ganglion, where in some patients it later reactivates in the form of shingles.⁸

In RHS, the ganglia of cranial nerve VII (CN VII, the facial nerve, which innervates the facial muscles) are infected; for this reason, the condition is also referred to as zoster oticus.⁹ Because of the involvement and weakening of the facial nerve, the presentation of RHS often resembles that of Bell’s palsy or facial nerve palsy.

While most cases of Bell’s palsy are idiopathic,^10,11 RHS can usually be attributed to viral infection—most commonly, infection with herpes simplex virus type 1 (HSV-1).¹² RHS can be differentiated from Bell’s palsy by the presence of a rash on the ipsilateral side. The rash appears in the form of inflamed vesicles on an erythematous base and may be present around the ear (see figure), the eardrum, the hard and soft palate, or the tongue.⁶ When the rash is painful, it is often described as a burning pain. Loss of taste may occur in the anterior portion of the tongue.^9,12

Unlike shingles, which usually manifests as a sensory neuropathy, RHS is distinguished by motor neuropathy.⁷ The patient usually reports weakness in the facial muscles on one side, leading to difficulty drinking water or puffing out the cheek and to drooling on one side of the face. A complaint of dryness in the ipsilateral eye may result from weakness or an inability to close the eyelid.

It is important to note that as in Bell’s palsy, RHS can be differentiated from stroke by the patient’s inability to wrinkle the forehead. The motor muscles of the forehead are innervated by both sides of the brain; in the case of stroke, only one side of the brain is affected, and movement of the forehead remains possible on the contralateral side. In facial nerve palsy, the nerve itself is affected; thus, no movement of the forehead is possible.¹³ Other common complaints in patients with facial nerve palsy include vertigo, hearing loss, and changes in facial sensation.

RHS was first described in 1907 as herpes zoster associated with Bell’s palsy by the neurologist J. Ramsay Hunt, for whom the condition is named.^9,14 RHS is more common in men than women. It occurs most commonly in adults and is rare in children younger than 6.^13,15

Diagnosis
In most cases, a diagnosis of RHS is made on a clinical basis.¹ However, a polymerase chain reaction (PCR) assay can be performed on samples of tear fluid or submandibular saliva to detect the zoster virus.^16,17 PCR can also be performed using exudates from the geniculate zone of the ear (a small area in the center of the auricle^6,14), which is more sensitive than tears or blood.^18,19 Findings from a complete blood count and the erythrocyte sedimentation rate can be used to differentiate between infectious and inflammatory causes.¹³

Head CT or MRI can be obtained to rule out any structural lesions. In one study, Kim et al²⁰ examined MRI changes in patients with either Bell’s palsy or RHS. In both conditions, researchers were able to identify swelling of the labyrinthine segment of the facial nerve on temporal MRI scans.²⁰ Although CT has not been shown to have any prognostic or diagnostic application, it can occasionally be used if decompression of the facial nerve is warranted.¹¹

Treatment
Data used to support the use of corticosteroids for treatment of Bell’s palsy^10,21,22 have been extrapolated to justify their use for treatment of RHS,²³ and prednisolone is the most common choice.¹⁰ Steroids reduce the associated inflammation, resulting in decreased pain and neurologic symptoms. A daily dose for one to two weeks, followed by a slow taper, is the preferred prescribing method.¹⁰

The addition of acyclovir has been recommended to inhibit viral DNA replication^9,23 (valacyclovir and famciclovir have also been mentioned^12,18). If started within three days of symptom onset, acyclovir can help reduce pain and hasten resolution of symptoms.

In a large retrospective study, it was demonstrated that patients treated with prednisone at 1.0 mg/kg/d for five days, followed by a 10-day taper, combined with acyclovir, showed long-term improvement that was statistically significant.²³ Complete facial recovery was reported in only 52% of patients, however. Risk factors for a poor prognosis include hypertension, diabetes mellitus, and advancing age.⁷

Artificial tears are also prescribed to keep the affected eye from becoming irritated and dry. The patient can be instructed to tape the eye closed at night.¹⁰

Early diagnosis and treatment (ie, within three days of symptom onset, and preferably with a combination of acyclovir and steroids²³) is an important factor in a good prognosis.^7,23 Because RHS-affected patients have only about a 50% chance of full recovery,²³ proper follow-up care is extremely important. Follow-up visits are recommended at two weeks, six weeks, and three months.¹³ For optimal outcomes in patients with this neurologic diagnosis, referral to a neurologist is recommended for ongoing management. This practitioner is likely to detect subtle changes in patient presentation and can perform follow-up testing as needed.

THE CASE PATIENT
One week after the patient’s visit to the ED, he was contacted by hospital staff for a standard satisfaction and quality control survey. The patient (who had been treated with steroids and acyclovir, ibuprofen, and artificial tears) reported almost complete resolution of his pain; any mild pain, he said, was easily tolerated or could be resolved with OTC medication. He reported only minimal persistent facial weakness, stating that he was able to eat, drink, and speak normally.

The patient had not been seen by any health care provider for follow-up, but he agreed to make an appointment as soon as possible.     

REFERENCES
1. Kim D, Bhimani M. Ramsay Hunt syndrome presenting as simple otitis externa. CJEM. 2008;10(3):247-250.

2. Agius AM, Pickles JM, Burch KL. A prospective study of otitis externa. Clin Otolaryngol. 1992;17(2):150-154.

3. Rosenfeld RM, Brown L, Cannon CR, et al; American Academy of Otolaryngology—Head and Neck Surgery Foundation. Clinical practice guideline: acute otitis externa. Otolaryngol Head Neck Surg. 2006;134(4 suppl):S4-S23.

4. Holland J, Bernstein J. Bell’s palsy. Clin Evid (Online). 2011 Mar 7;2011.pii:1204.

5. Jacewicz M. Bell’s palsy (2007). www.merckmanuals.com/professional/sec16/ch219/ch219i.html. Accessed May 26, 2011.

6. Harrison K. Discussion: the Ramsay Hunt Syndrome. Proc Royal Soc Med. 1953;47(371):11-24.

7. Bhupal HK. Ramsay Hunt syndrome presenting in primary care. Practitioner. 2010;254(1727):33-35.

8. Aizawa H, Ohtani F, Furuta Y, et al. Variable patterns of varicella-zoster virus reactivation in Ramsay Hunt syndrome. J Med Virol. 2004;74(2):355-360.

9. Gondivkar S, Parikh V, Parikh R. Herpes zoster oticus: a rare clinical entity. Contemp Clin Dent. 2010;1(2):127-129.

10. Sullivan FM, Swan IRC, Donnan PT, et al. Early treatment with prednisolone or acyclovir in Bell’s palsy. N Engl J Med. 2007;357(16):1598-1607.

11. Gilden DH. Bell’s palsy. N Engl J Med. 2004;351(13):1323-1331.

12. Diaz GA, Rakita RM, Koelle DM. A case of Ramsay Hunt–like syndrome caused by herpes simplex virus type 2. Clin Infect Dis. 2005;40(10):1545-1547.

13. Miravalle AA. Ramsay Hunt syndrome. http://emedicine.medscape.com/article/1166804-over iew. Accessed July 22, 2011.

14. Hunt JR. On herpetic inflammation of the geniculate ganglion: a new syndrome and its complications. J Nerv Ment Dis. 1907;34:73-96.

15. Sandoval CC, Núñez FA, Lizama CM, et al. Ramsay Hunt syndrome in children: four cases and review [in Spanish]. Rev Chilena Infectol. 2008; 25(6):458-464.

16. Murakami S, Nakashiro Y, Mizobuchi M, et al. Varicella-zoster virus distribution in Ramsay Hunt syndrome revealed by polymerase chain reaction. Acta Otolaryngol. 1998;118(2):145-149.

17. Hiroshige K, Ikeda M, Hondo R. Detection of varicella zoster virus DNA in tear fluid and saliva of patients with Ramsay Hunt syndrome. Otol Neurol. 2002;23(4):602-607.

18. Sweeney CJ, Gilden DH. Ramsay Hunt syndrome. J Neurol Neurosurg Psychiatr. 2001;71(2):148-154.

19. Murakami S, Honda N, Mizobuchi M, et al. Rapid diagnosis of varicella zoster virus infection in acute facial palsy. Neurology. 1998;51(4):1202-1205.

20. Kim IS, Shin SH, Kim J, et al. Correlation between MRI and operative findings in Bell’s palsy and Ramsay Hunt syndrome. Yonsei Med J. 2007;48(6):963-968.

21. Engström M, Berg T, Stjernquist-Desatnik A, et al. Prednisolone and valaciclovir in Bell’s palsy: a randomised, double-blind, placebo-controlled, multicentre trial. Lancet Neurol. 2008;7(11):993-1000.

22. Hato N, Yamada H, Kohno H, et al. Valacyclovir and prednisolone treatment for Bell’s palsy: a multicenter, randomized, placebo-controlled study. Otol Neurotol. 2007;28(3):408-413.

23. Murakami S, Hato N, Horiuchi J, et al. Treatment of Ramsay Hunt syndrome with acyclovir-prednisone: significance of early diagnosis and treatment. Ann Neurol. 1997;41(3):353-357.

A 46-year-old man presented to a hospital emergency department (ED) with a four-day history of right ear pain. He described the pain as a constant, dull, burning pain radiating to the neck and face, associated with a feeling of congestion. The patient also stated that the right side of his face had felt numb for about one day.

Three days earlier, the man had been seen by his primary health care provider, who told him that his ear looked normal and free of infection. The day before his current presentation to the ED, however, he noticed what he described as an “acne-like” rash on his ear lobe. Shortly before coming to the ED, the patient also developed numbness over his right upper lip, which he likened to the effects of procaine during a dental visit. He reported drooling from the right side of his mouth while drinking water and difficulty blinking his right eye. 

He denied any tinnitus, fever, headache, or change in hearing. A review of symptoms was positive only for mild dizziness during the previous two to three days.

The patient was a well-appearing white man. He was alert and oriented to identity, time, and place. His skin was warm, dry, and intact. The examiner noticed a small area of erythematous rash with vesicles on the man’s right ear lobe. The external auditory canals appeared within normal limits, with no erythema or edema, and were nontender bilaterally. The tympanic membranes were normal bilaterally, without bulging or discernible fluid levels.

The ocular exam was normal with no visual acuity changes and no fluorescein uptake; external ocular movements were intact. A slight droop was noted in the right eyelid, but there was no droop on the contralateral side of his face. When asked to puff up his cheeks, the patient found it difficult to do so on the right side of his mouth without releasing air from his lips.

The remainder of the cranial nerves were intact. Muscle strength was 5/5 in all extremities and equal bilaterally. The man’s gait was within normal limits, and the remaining findings in the physical exam were normal.

The initial diagnosis considered in the differential was otitis externa, because it is a common explanation for ear pain in patients who present to the ED.^1,2 Also, in otitis, pain is characteristically present in the affected ear, and erythema is often found in the external auditory canal.³ However, this diagnosis was deemed unlikely because otitis externa would not explain the neurologic findings or the vesicular rash.¹

Bell’s palsy was next in the differential, as it was considered consistent with the patient’s unilateral neurologic deficits.⁴ In addition to weakness or palsy of the facial nerve, many patients with Bell’s palsy complain of mastoid pain, which can be confused with a complaint of ear pain.⁵ However, patients with Bell’s palsy have no rash, and this diagnosis was considered unlikely.

The painful, burning rash on the patient’s face was characteristic of herpes zoster (shingles), which was next in the differential. Infrequently, shingles can also cause weakness in the nerve it affects. In the case patient, weakness that was evident in the affected nerve resembled that seen in Bell’s palsy. This combination of symptoms is referred to as Ramsay Hunt syndrome—which in this case was decided to be the correct diagnosis.

DISCUSSION
Ramsay Hunt syndrome (RHS, also known as geniculate herpes^5,6) is caused by the varicella-zoster virus, most commonly known as the cause of chickenpox. In the United States, RHS is believed to affect only about one in 1,500 persons, although 20% to 30% of persons experience herpes zoster infection at some time.⁷

Soon after a chickenpox infection subsides, the virus spreads along the sensory nerve fibers of the peripheral and cranial nerves. The virus then becomes dormant in the dorsal root ganglion, where in some patients it later reactivates in the form of shingles.⁸

In RHS, the ganglia of cranial nerve VII (CN VII, the facial nerve, which innervates the facial muscles) are infected; for this reason, the condition is also referred to as zoster oticus.⁹ Because of the involvement and weakening of the facial nerve, the presentation of RHS often resembles that of Bell’s palsy or facial nerve palsy.

While most cases of Bell’s palsy are idiopathic,^10,11 RHS can usually be attributed to viral infection—most commonly, infection with herpes simplex virus type 1 (HSV-1).¹² RHS can be differentiated from Bell’s palsy by the presence of a rash on the ipsilateral side. The rash appears in the form of inflamed vesicles on an erythematous base and may be present around the ear (see figure), the eardrum, the hard and soft palate, or the tongue.⁶ When the rash is painful, it is often described as a burning pain. Loss of taste may occur in the anterior portion of the tongue.^9,12

Unlike shingles, which usually manifests as a sensory neuropathy, RHS is distinguished by motor neuropathy.⁷ The patient usually reports weakness in the facial muscles on one side, leading to difficulty drinking water or puffing out the cheek and to drooling on one side of the face. A complaint of dryness in the ipsilateral eye may result from weakness or an inability to close the eyelid.

It is important to note that as in Bell’s palsy, RHS can be differentiated from stroke by the patient’s inability to wrinkle the forehead. The motor muscles of the forehead are innervated by both sides of the brain; in the case of stroke, only one side of the brain is affected, and movement of the forehead remains possible on the contralateral side. In facial nerve palsy, the nerve itself is affected; thus, no movement of the forehead is possible.¹³ Other common complaints in patients with facial nerve palsy include vertigo, hearing loss, and changes in facial sensation.

RHS was first described in 1907 as herpes zoster associated with Bell’s palsy by the neurologist J. Ramsay Hunt, for whom the condition is named.^9,14 RHS is more common in men than women. It occurs most commonly in adults and is rare in children younger than 6.^13,15

Diagnosis
In most cases, a diagnosis of RHS is made on a clinical basis.¹ However, a polymerase chain reaction (PCR) assay can be performed on samples of tear fluid or submandibular saliva to detect the zoster virus.^16,17 PCR can also be performed using exudates from the geniculate zone of the ear (a small area in the center of the auricle^6,14), which is more sensitive than tears or blood.^18,19 Findings from a complete blood count and the erythrocyte sedimentation rate can be used to differentiate between infectious and inflammatory causes.¹³

Head CT or MRI can be obtained to rule out any structural lesions. In one study, Kim et al²⁰ examined MRI changes in patients with either Bell’s palsy or RHS. In both conditions, researchers were able to identify swelling of the labyrinthine segment of the facial nerve on temporal MRI scans.²⁰ Although CT has not been shown to have any prognostic or diagnostic application, it can occasionally be used if decompression of the facial nerve is warranted.¹¹

Treatment
Data used to support the use of corticosteroids for treatment of Bell’s palsy^10,21,22 have been extrapolated to justify their use for treatment of RHS,²³ and prednisolone is the most common choice.¹⁰ Steroids reduce the associated inflammation, resulting in decreased pain and neurologic symptoms. A daily dose for one to two weeks, followed by a slow taper, is the preferred prescribing method.¹⁰

The addition of acyclovir has been recommended to inhibit viral DNA replication^9,23 (valacyclovir and famciclovir have also been mentioned^12,18). If started within three days of symptom onset, acyclovir can help reduce pain and hasten resolution of symptoms.

In a large retrospective study, it was demonstrated that patients treated with prednisone at 1.0 mg/kg/d for five days, followed by a 10-day taper, combined with acyclovir, showed long-term improvement that was statistically significant.²³ Complete facial recovery was reported in only 52% of patients, however. Risk factors for a poor prognosis include hypertension, diabetes mellitus, and advancing age.⁷

Artificial tears are also prescribed to keep the affected eye from becoming irritated and dry. The patient can be instructed to tape the eye closed at night.¹⁰

Early diagnosis and treatment (ie, within three days of symptom onset, and preferably with a combination of acyclovir and steroids²³) is an important factor in a good prognosis.^7,23 Because RHS-affected patients have only about a 50% chance of full recovery,²³ proper follow-up care is extremely important. Follow-up visits are recommended at two weeks, six weeks, and three months.¹³ For optimal outcomes in patients with this neurologic diagnosis, referral to a neurologist is recommended for ongoing management. This practitioner is likely to detect subtle changes in patient presentation and can perform follow-up testing as needed.

THE CASE PATIENT
One week after the patient’s visit to the ED, he was contacted by hospital staff for a standard satisfaction and quality control survey. The patient (who had been treated with steroids and acyclovir, ibuprofen, and artificial tears) reported almost complete resolution of his pain; any mild pain, he said, was easily tolerated or could be resolved with OTC medication. He reported only minimal persistent facial weakness, stating that he was able to eat, drink, and speak normally.

The patient had not been seen by any health care provider for follow-up, but he agreed to make an appointment as soon as possible.     

REFERENCES
1. Kim D, Bhimani M. Ramsay Hunt syndrome presenting as simple otitis externa. CJEM. 2008;10(3):247-250.

2. Agius AM, Pickles JM, Burch KL. A prospective study of otitis externa. Clin Otolaryngol. 1992;17(2):150-154.

3. Rosenfeld RM, Brown L, Cannon CR, et al; American Academy of Otolaryngology—Head and Neck Surgery Foundation. Clinical practice guideline: acute otitis externa. Otolaryngol Head Neck Surg. 2006;134(4 suppl):S4-S23.

4. Holland J, Bernstein J. Bell’s palsy. Clin Evid (Online). 2011 Mar 7;2011.pii:1204.

5. Jacewicz M. Bell’s palsy (2007). www.merckmanuals.com/professional/sec16/ch219/ch219i.html. Accessed May 26, 2011.

6. Harrison K. Discussion: the Ramsay Hunt Syndrome. Proc Royal Soc Med. 1953;47(371):11-24.

7. Bhupal HK. Ramsay Hunt syndrome presenting in primary care. Practitioner. 2010;254(1727):33-35.

8. Aizawa H, Ohtani F, Furuta Y, et al. Variable patterns of varicella-zoster virus reactivation in Ramsay Hunt syndrome. J Med Virol. 2004;74(2):355-360.

9. Gondivkar S, Parikh V, Parikh R. Herpes zoster oticus: a rare clinical entity. Contemp Clin Dent. 2010;1(2):127-129.

10. Sullivan FM, Swan IRC, Donnan PT, et al. Early treatment with prednisolone or acyclovir in Bell’s palsy. N Engl J Med. 2007;357(16):1598-1607.

11. Gilden DH. Bell’s palsy. N Engl J Med. 2004;351(13):1323-1331.

12. Diaz GA, Rakita RM, Koelle DM. A case of Ramsay Hunt–like syndrome caused by herpes simplex virus type 2. Clin Infect Dis. 2005;40(10):1545-1547.

13. Miravalle AA. Ramsay Hunt syndrome. http://emedicine.medscape.com/article/1166804-over iew. Accessed July 22, 2011.

14. Hunt JR. On herpetic inflammation of the geniculate ganglion: a new syndrome and its complications. J Nerv Ment Dis. 1907;34:73-96.

15. Sandoval CC, Núñez FA, Lizama CM, et al. Ramsay Hunt syndrome in children: four cases and review [in Spanish]. Rev Chilena Infectol. 2008; 25(6):458-464.

16. Murakami S, Nakashiro Y, Mizobuchi M, et al. Varicella-zoster virus distribution in Ramsay Hunt syndrome revealed by polymerase chain reaction. Acta Otolaryngol. 1998;118(2):145-149.

17. Hiroshige K, Ikeda M, Hondo R. Detection of varicella zoster virus DNA in tear fluid and saliva of patients with Ramsay Hunt syndrome. Otol Neurol. 2002;23(4):602-607.

18. Sweeney CJ, Gilden DH. Ramsay Hunt syndrome. J Neurol Neurosurg Psychiatr. 2001;71(2):148-154.

19. Murakami S, Honda N, Mizobuchi M, et al. Rapid diagnosis of varicella zoster virus infection in acute facial palsy. Neurology. 1998;51(4):1202-1205.

20. Kim IS, Shin SH, Kim J, et al. Correlation between MRI and operative findings in Bell’s palsy and Ramsay Hunt syndrome. Yonsei Med J. 2007;48(6):963-968.

21. Engström M, Berg T, Stjernquist-Desatnik A, et al. Prednisolone and valaciclovir in Bell’s palsy: a randomised, double-blind, placebo-controlled, multicentre trial. Lancet Neurol. 2008;7(11):993-1000.

22. Hato N, Yamada H, Kohno H, et al. Valacyclovir and prednisolone treatment for Bell’s palsy: a multicenter, randomized, placebo-controlled study. Otol Neurotol. 2007;28(3):408-413.

23. Murakami S, Hato N, Horiuchi J, et al. Treatment of Ramsay Hunt syndrome with acyclovir-prednisone: significance of early diagnosis and treatment. Ann Neurol. 1997;41(3):353-357.