Jeffrey L. Greenwald, MD

Background

The last drug approved by the Food and Drug Administration (FDA) for heart failure (HF) was 10 years ago.[2] The new PARADIGM (Prospective Comparison of ARNI With ACEI to Determine Impact on Global Mortality and Morbidity in Heart Failure) heart failure study comparing a novel combination drug of a neprilysin inhibitor and angiotensin receptor blocker (ARB) to an angiotensin‐converting enzyme (ACE) inhibitor has cardiologists considering a possible change in the HF treatment algorithm. Neprilysin is a naturally occurring enzyme that breaks down the protective vasoactive peptides (brain natriuretic peptide, atrial natriuretic peptide, and bradykinin) made by the heart and the body in HF. These vasoactive peptides function to increase vasodilation and block sodium and water reabsorption. This novel neprilysin inhibitor extends the life of these vasoactive peptides, thus enhancing their effect. By inhibiting both neprilysin and the renin‐angiotensin system, there should be additional improvement in HF management. The neprilysin inhibitor was combined with an ARB instead of an ACE inhibitor because of significant angioedema seen in earlier phase trials when combined with an ACE inhibitor. This is believed related to increases in bradykinin due to both agents.

Findings

In this multicenter, blinded, randomized trial, over 10,000 patients with known HF (ejection fraction<35%, New York Heart Association class II or higher) went through 2 run‐in periods to ensure tolerance of both enalapril and the study drug, a combination of a neprilysin inhibitor and valsartan (neprilysin‐I/ARB). Eventually 8442 patients underwent randomization to either enalapril (10 mg twice a day) or neprilysin‐I/ARB (200 mg twice a day). The primary outcome was a combination of cardiovascular mortality and heart failure hospitalizations. The trial was stopped early at 27 months because of overwhelming benefit with neprilysin‐I/ARB (21.8% vs 26.5%; P<0.001). There was a 20% reduction specifically in cardiovascular mortality (13.3% vs 16.5%; hazard ratio [HR]: 0.80; P<0.001). The number needed to treat (NNT) was 32. There was also a 21% reduction in the risk of hospitalization (P<0.001). More patients with neprilysin‐I/ARB had symptomatic hypotension (14% vs 9.2%; P<0.001) but patients on the ACE inhibitor experienced more cough, hyperkalemia, and increases in their serum creatinine.

Cautions

There are 2 reasons clinicians may not see the same results in practice. First, the trial was stopped early, which can sometimes exaggerate benefits.[3] Second, the 2 run‐in periods eliminated patients who could not tolerate the medications at the trial doses. Additionally, although the study's authors were independent, the trial was funded by a pharmaceutical company.

Implications

This new combination drug of a neprilysin inhibitor and valsartan shows great promise at reducing cardiovascular mortality and hospitalizations for heart failure compared to enalapril alone. Given the high morbidity and mortality of heart failure, having a new agent in the treatment algorithm will be useful to patients and physicians. The drug was just approved by the FDA in July 2015 and will likely be offered as an alternative to ACE inhibitors.

Background

It is a common practice to withhold intravenous contrast material from computed tomography (CT) scans in patients with even moderately poor renal function out of concern for causing contrast‐induced nephropathy (CIN). Our understanding of CIN is based largely on observational studies and outcomes of cardiac catheterizations, where larger amounts of contrast are given intra‐arterially into an atherosclerotic aorta.[4] The exact mechanism of injury is not clear, possibly from direct tubule toxicity or renal vasoconstriction.[5] CIN is defined as a rise in creatinine >0.5 mg/dL or >25% rise in serum creatinine 24 to 48 hours after receiving intravenous contrast. Although it is usually self‐limited, there is concern that patients who develop CIN have an increase risk of dialysis and death.[6] In the last few years, radiologists have started to question whether the risk of CIN is overstated. A recent meta‐analysis of 13 studies demonstrated a similar likelihood of acute kidney injury in patients regardless of receiving intravenous contrast.[7] If the true incidence of CIN after venous contrast is actually lower, this raises the question of whether we are unnecessarily withholding contrast from CTs and thereby reducing their diagnostic accuracy. Two 2014 observational studies provide additional evidence that the concern for CIN may be overstated.

Findings

The 2 Mayo Clinic studies used the same database. They looked at all patients who underwent a contrast‐enhanced or unenhanced thoracic, abdominal, or pelvic CT between January 2000 and December 2010 at the Mayo Clinic. After limiting the data to patients with pre‐ and post‐CT creatinine measurements and excluding anyone on dialysis, with preexisting acute kidney injury, or who had received additional contrast within 14 days, they ended up with 41,229 patients, mostly inpatients. All of the patients were assigned propensity scores based on risk factors for the development of CIN and whether they would likely receive contrast. The patients were then subdivided into 4 renal function subgroups based on estimated glomerular filtration rate (eGFR). The patients who received contrast were matched based on their propensity scores to those who did not received contrast within their eGFR subgroups. Unmatched patients were eliminated, leaving a cohort of 12,508 matched patients. The outcome of the first article was acute kidney injury (AKI) defined as a rise in creatinine >0.5 mg/dL at 24 to 48 hours. Though AKI rose with worsening eGFR subgroups (eGFR > 90 [1.2%] vs eGFR < 30 [14%]), the rates of AKI were the same regardless of contrast exposure. There was no statistical difference in any of the eGFR subgroups. The second study looked at important clinical outcomesdeath and the need for dialysis. There was no statistical difference for emergent dialysis (odds ratio [OR]: 0.96, P=0.89) or 30‐day mortality (HR: 0.97; P=0.45) regardless of whether the patients received contrast or not.

Cautions

In propensity matching, unmeasured confounders can bias the results. However, the issue of whether venous contrast causes CIN will unlikely be settled in a randomized controlled trial. For patients with severe renal failure (eGFR < 30), there were far fewer patients in this subgroup, making it harder to draw conclusions. The amount of venous contrast given was not provided. Finally, this study evaluated intravenous contrast for CTs, not intra‐arterial contrast.

Implications

These 2 studies raise doubt as to whether the incidence of AKI after contrast‐enhanced CT can be attributed to the contrast itself. What exactly causes the rise in creatinine is probably multifactorial including lab variation, hydration, blood pressure changes, nephrotoxic drugs, and comorbid disease. In trying to decide whether to obtain a contrast‐enhanced CT for patients with chronic kidney dysfunction, these studies provide more evidence to consider in the decision‐making process. A conversation with the radiologist about the benefits gained from using contrast in an individual patient may be of value.

Background

Numerous medications and therapeutic approaches have been studied to prevent incident delirium in hospitalized medical and surgical patients with varying success. Many of the tested medications also have the potential for significant undesirable side effects. An earlier small trial of melatonin appeared to have impressive efficacy for this purpose and be well tolerated, but the substance is not regulated by the FDA.[8] Ramelteon, a melatonin receptor agonist, is approved by the FDA for insomnia, and authors hypothesized that it, too, may be effective in delirium prevention.

Findings

This study was a multicenter, single‐blinded, randomized controlled trial of the melatonin‐agonist ramelteon versus placebo in elderly patients admitted to the hospital ward or ICU with serious medical conditions. Researchers excluded intubated patients or those with Lewy body dementia, psychiatric disorders, and severe liver disease. Patients received either ramelteon or placebo nightly for up to a week, and the primary end point was incident delirium as determined by a blinded observer using a validated assessment tool. Sixty‐seven patients were enrolled. The baseline characteristics in the arms of the trial were similar. In the placebo arm, 11 of 34 patients (32%) developed delirium during the 7‐day observation period. In the ramelteon arm, 1 of 33 (3%) developed delirium (P=0.003). The rate of drug discontinuation was the same in each arm.

Cautions

This study is small, and the single‐blinded design (the physicians and patients knew which group they were in but the observers did not) limits the validity of these results, mandating a larger double‐blinded trial.

Implications

Ramelteon showed a dramatic impact on preventing incident delirium on elderly hospitalized patients with serious medical conditions admitted to the ward or intensive care unit (ICU) (nonintubated) in this small study. If larger trials concur with the impact of this well‐tolerated and inexpensive medication, the potential for delirium incidence reduction could have a dramatic impact on how care for delirium‐vulnerable patients is conducted as well as the systems‐level costs associated with delirium care. Further studies of this class of medications are needed to more definitively establish its value in delirium prevention.

Background

Delirium is common in hospitalized elderly patients, and numerous studies show that there are both short‐ and long‐term implications of developing delirium. Using well studied and validated tools has made identifying delirium fairly straightforward, yet its treatment remains difficult. Additionally, differentiating which patients will have a simpler clinical course from those at risk for a more morbid one has proved challenging. Using the Confusion Assessment Method (CAM), both in its short (4‐item) and long (10‐item) forms, as the basis for a prognostication tool, would allow for future research on treatment to have a scale against which to measure impact, and would allow clinicians to anticipate which patients were more likely to have difficult clinical courses.

Findings

The CAM Severity (CAM‐S) score was derived in 1219 subjects participating in 2 ongoing studies: 1 included high‐risk medical inpatients 70 years old or older, and the other included similarly aged patients undergoing major orthopedic, general, or vascular surgeries. Outcomes data were not available for the surgical patients. The CAM items were rated as either present/absent or absent/mild/severe, depending on the item, with an associated score attached to each item such that the 4‐item CAM had a score of 0 to 7 and the 10‐item CAM 0 to 19 (Table 2). Clinical outcomes from the medical patients cohort showed a dose response with increasing CAM‐S scores with respect to length of stay, adjusted cost, combined 90‐day end points of skilled nursing facility placement or death, and 90‐day mortality. Specifically, for patients with a CAM‐S (short form) score of 5 to 7, the 90‐day rate of death or nursing home residence was 62%, whereas the 90‐day postdischarge mortality rate was 36%.

Cautions

The CAM‐S, like the CAM, may work less well in patients with hypoactive delirium. This scale has been applied in a surgical cohort, but study outcomes were not presented in this article. This absence limits our ability to apply these results to a surgical population presently.

Implications

This study demonstrates that in medical inpatients, the CAM‐S is effective for prognostication. Moreover, the study points out that high‐scoring patients on the CAM‐S have quite poor prognoses, with more than one‐third dying by 3 months. This finding suggests that an important use of the CAM‐S is to identify patients about whom goals of care discussions should be held and end‐of‐life planning initiated if not previously done.

Background

Lactulose has been the principle treatment for acute hepatic encephalopathy (HE) since 1966.[9] It theoretically works by lowering the pH of the colon and trapping ammonia as ammonium, which is then expelled. Alternatively, it may simply decrease transit time through the colon. In fact, earlier treatments for HE were cathartics such as magnesium salts. Unfortunately 20% tp 30% of patients are poor responders to lactulose, and patients do not like it. This new study tests whether a modern‐day cathartic, polyethylene glycol, works as well as lactulose.

Findings

In this unblinded, randomized controlled trial, patients presenting to the emergency department with acute HE were assigned to either lactulose 20 to 30 g for a minimum of 3 doses over 24 hours or 4 L of polyethylene glycol (PEG) over 4 hours. The2 groups were similar in severity and etiology of liver disease. Patients were allowed to have received 1 dose of lactulose given in the emergency department prior to study enrollment. They were excluded if taking rifaximin. The primary outcome was improvement in the hepatic encephalopathy scoring algorithm (HESA) by 1 grade at 24 hours.[10] The algorithm scores HE from 0 (no clinical findings of HE) to 5 (comatose). Initial mean HESA scores in the 2 groups were identical (2.3).

In the lactulose group, 13/25 (52%) improved by at least 1 HESA score at 24 hours. Two patients (8%) completely cleared with a HESA score of 0. In comparison, 21/23 (91%) in the PEG group improved at 24 hours, and 10/23 (43%) had cleared with a HESA score of 0 (P<0.01). The median time to HE resolution was 2 days in the lactulose group compared with 1 day in the PEG group (P=0.01). There were no differences in serious adverse events. The majority (76%) of the PEG group received the full 4 L of PEG.

Cautions

The main limitations of the trial were the small sample size, that it was a single‐center study, and the fact it was unblinded. Additionally, 80% of the PEG group received 1 dose of lactulose prior to enrollment. Statistically, more patients in the PEG group developed hypokalemia, which can worsen HE. Therefore, if PEG is used for acute HE, potassium will need to be monitored.

Implications

The results are intriguing and may represent a new possible treatment for acute HE once larger studies are done. Interestingly, the ammonia level dropped further in the lactulose group than the PEG group, yet there was more cognitive improvement in the PEG group. This raises questions about the role of ammonia and catharsis in HE. Although lactulose and rifaximin continue to be the standard of care, cathartics may be returning as a viable alternative.

Background

Nonselective ‐blockers (NSBBs) are considered the aspirin of hepatologists, as they are used for primary and secondary prevention of variceal bleeds in patients with cirrhosis.[11] Since the 1980s, their benefit in reducing bleeding risk has been known, and more recently there has been evidence that they may reduce the risk of developing ascites in patients with compensated cirrhosis. Yet, there has been some contradictory evidence suggesting reduced survival in patients with decompensated cirrhosis and infections on NSBBs. This has led to the window hypothesis of NSBBs in cirrhosis, where NSBBs are beneficial only during a certain window period during the progression of cirrhosis.[12] Early on in cirrhosis, before the development of varices or ascites, NSBBs have no benefit. As cirrhosis progresses and portal hypertension develops, NSBBs play a major role in reducing bleeding from varices. However, in advanced cirrhosis, NSBBs may become harmful. In theory, they block the body's attempt to increase cardiac output during situations of increased physiologic stress, resulting in decreased mean arterial pressure and perfusion. This, in turn, causes end‐organ damage and increased risk of death. When exactly this NSBB window closes is unclear. A 2014 study suggests the window should close when patients develop spontaneous bacterial peritonitis (SBP).

Findings

This retrospective study followed 607 consecutive patients seen at a liver transplant center in Vienna, Austria, from 2006 to 2011. All of the patients were followed from the time of their first paracentesis. They were excluded if SBP was diagnosed during the first paracentesis. Patients were grouped based on whether they took an NSBB. As expected, more patients on an NSBB had varices (90% vs 62%; P<0.001) and a lower mean heart rate (77.5 vs 83.9 beats/minute; P<0.001). However, the 2 groups were similar in mean arterial pressure, systolic blood pressure, Model for End‐Stage Liver Disease score (17.5), Childs Pugh Score (CPS) (50% were C), and in the etiology of cirrhosis (55% were from alcoholic liver disease). They followed the patients for development of SBP. The primary outcome was transplant‐free survival. For the patients who never developed SBP, there was a 25% reduction in the risk of death for those on an NSBB adjusted for varices and CPS stage (HR=0.75, P=0.027). However, for the 182 patients who developed SBP, those on an NSBB had a 58% increase risk of death, again adjusted for varices and CPS stage (HR=1.58, P=0.014). Among the patients who developed SBP, there was a higher risk of hepatorenal syndrome (HRS) within 90 days for those on an NSBB (24% vs 11%, P=0.027). Although the mean arterial pressures (MAP) had been similar in the 2 groups before SBP, after the development of SBP, those on an NSBB had a significantly lower MAP (77.2 vs 82.6 mm Hg, P=0.005).

Cautions

This is a retrospective study, and although the authors controlled for varices and CPS, it is still possible the 2 groups were not similar. Whether patients were actually taking the NSBB is unknown, and doses of the NSBB were variable.

Implications

This study provides more evidence for the NSBB window hypothesis in the treatment of patients with cirrhosis. It suggests that the window on NSBB closes when patients develop SBP, as NSBBs appear to increase mortality and the risk of HRS. Thus, NSBB therapy should probably be discontinued in cirrhotic patients developing SBP. The question is for how long? The editorial accompanying the article says permanently.[13]

Background

Based on early research studies, many quality and regulatory organizations have stressed the importance of assessing hospitalized patients' venous thromboembolism (VTE) risk and prophylaxing those patients at increased risk either pharmacologically or mechanically. In 2011, a meta‐analysis of 40 studies of medical and stroke patients including approximately 52,000 patients failed to demonstrate a mortality benefit, showing that for every 3 pulmonary embolisms (PEs) prevented, it caused 4 major bleeding episodes per 1000 patients.[14] A second study in 2011, a multicenter, randomized controlled trial with medically complex patients deemed high risk for VTE, also failed to demonstrate a mortality benefit.[15] Despite these and other trials showing questionable benefit, guidelines continue to recommend that high‐risk medical patients should get pharmacologic prophylaxis against VTE.

Findings

This retrospective cohort trial retrospectively evaluated a cohort of 20,794 medical patients (non‐ICU) across 35 hospitals, excluding those with a Caprini score of <2 (ie, low risk for VTE). The authors divided the hospitals into tertiles based on adherence to VTE prophylaxis guidelines. Patients were followed to 90 days after hospitalization with telephone calls (reaching 56%) and chart reviews (100% reviewed) to identify clinically evident VTE events, excluding those that occurred within the first 3 days of index hospitalization. The study identified no statistically significant differences among the tertiles in terms of VTE rates, either in the hospital or at 90 days, though the overall VTE event rate was low. Interestingly, 85% of events took place postdischarge. Subgroup analyses also failed to identify a population of medical patients who benefited from prophylaxis.

Cautions

Debate about whether the Caprini risk score is the best available VTE risk scoring system exists. This study also excluded surgical and ICU patients.

Implications

This trial adds to the mounting literature suggesting that current guidelines‐based pharmacologic VTE prophylaxis for medical patients may offer no clear benefit in terms of incident VTE events or mortality. Although it is not yet time to abandon VTE prophylaxis completely, this study does raise the important question of whether it is time to revisit the quality guidelines and regulatory standards around VTE prophylaxis in medical inpatients. It also highlights the difficulty in assessing medical patients for their VTE risk. Though this study is provocative and important for its real‐world setting, further studies are required.

Background

There have been 6 large published randomized controlled trials of direct oral anticoagulants (DOACs) versus vitamin K antagonists (VKAs) in patients with acute VTE. Study sizes range from approximately 2500 to over 8000 subjects. All showed no significant difference between the arms with respect to efficacy (VTE or VTE‐related death) but had variable results with respect to major bleeding risk, a major concern given the nonreversibility of this group of medications. Additionally, subgroup analysis within these studies was challenging given sample size issues.

Findings

These 6 studies were combined in a meta‐analysis to address the DOACs' overall efficacy and safety profile, as well as looking in prespecified subgroups. The meta‐analysis included data from over 27,000 patients, evenly divided between DOACs (edoxaban, apixaban, rivaroxaban, and dabigatran) and VKAs, with the time in the therapeutic range (TTR) in the VKA arm being 64%. Overall, the primary efficacy endpoint (VTE and VTE‐related death) was similar (DOACs relative tisk [RR]=0.90; 95% confidence interval [CI]: 0.77‐1.06) but major bleeding (DOACs RR=0.61; 95% CI: 0.45‐0.83; NNT=150) and combined fatal and intracranial bleeding (DOACs RR=0.37; 95% CI: 0.27‐0.68; NNT=314) favored the DOACs. In subgroup analysis, there was no efficacy difference between the therapeutic groups in the subset specifically with DVT or with PE, or with patients weighing >100 kg, though safety data in these subsets were not evaluable. Patients with creatinine clearances of 30 to 49 mL/min demonstrated similar efficacy in both treatment arms, and the safety analysis in this subset with moderate renal impairment was better in the DOAC arm. Cancer patients achieved better efficacy with similar safety with the DOACs, whereas elderly patients achieved both better safety and efficacy with DOACs.

Cautions

As yet, there are inadequate data on patients with more advanced renal failure (creatinine clearance <30 mL/min) to advise using DOACs in that subset. Also, as there were no data comparing cancer patients with VTE that investigated DOACs versus low molecular weight heparins (the standard of care rather than warfarin since the CLOT [Comparison of Low‐molecular‐weight heparin versus Oral anticoagulant Therapy] trial[16]), the current meta‐analysis does not yet answer whether DOACs should be used in this population despite the efficacy benefit noted in the subgroup analysis.

Implications

This large meta‐analysis strongly suggests we can achieve comparable treatment efficacy from the DOACs as with VKAs, with better safety profiles in patients with acute VTE. In the subset of patients with moderate renal impairment (creatinine clearance 3049 mL/min), it appears safe and effective to choose DOACs.

Background

Adding to the previously published meta‐analyses of the original phase 3 randomized trials regarding the DOACs' impact on the atrial fibrillation (AF) treatment safety and efficacy literature relative to VKAs, a 2013 trial, ENGAGE AF‐TIMI 48 (Effective Anticoagulation with Factor Xa Next Generation in Atrial FibrillationThrombolysis in Myocardial Infarction 48), with edoxaban was published and warrants inclusion to have a better opportunity to glean important subgroup information.[17]

Findings

This meta‐analysis included data on 71,683 patients, 42,411 in the DOAC arm and 29,272 in the warfarin arm, as 2 of the trials were3‐arm studies, comparing warfarin to a high dose and a low dose of the DOAC. Meta‐analyses of the 4 trials were broken down into a high‐dose subsetthe 2 high‐dose arms and the standard doses used in the other 2 trialsand a low‐dose subsetthe 2 low‐dose arms and the standard doses used in the other 2 trials. With respect to the efficacy endpoint (incident stroke or systemic embolization), the high‐dose subset analyses of the DOACs yielded a 19% reduction (P<0.0001; NNT=142) relative to the VKAs. The safety endpoint of major bleeding in this analysis identified a 14% reduction in the DOAC group that was nonsignificant (P=0.06). Within the high‐dose subset, analyses favored DOACs with respect to hemorrhagic stroke (51% reduction; P<0.0001; NNT=220), intracranial hemorrhage (52% reduction; P<0.0001; NNT=132), and overall mortality (10% reduction; P=0.0003; NNT=129), whereas they increased the risk of gastrointestinal bleeding (25% increase; P=0.043; NNH=185). There was no significant difference between DOACs and warfarin with respect to ischemic stroke. The low‐dose subset had similar overall results with even fewer hemorrhage strokes balancing a higher incidence of ischemic strokes in the DOAC arm than in warfarin. Other important subgroup analyses suggest the safety and efficacy impact of DOACs is significant for VKA‐naive and experienced patients, though only statistically so for VKA‐naive patients. Additionally, the anticoagulation centers included in the study that had a TTR <66% seemed to gain a safety advantage from the DOACs, whereas both TTR groups (<66% and 66%) appeared to achieve an efficacy benefit from DOACs.

Cautions

There are not sufficient data to suggest routinely switching patients tolerating and well managed on VKAs to DOACs for AF.

Implications

DOACs reduce stroke and systemic emboli in patients with AF without increasing intracranial bleeding or hemorrhagic stroke, though at the cost of increased gastrointestinal bleeding in patients on the high‐dose regimens. Those patients on the low‐dose regimens have even a lower hemorrhagic stroke risk, the benefit of which is negated by a higher than VKA risk of ischemic strokes. Centers with lower TTRs (and perhaps by extrapolation, those patients with more difficulty staying in the therapeutic range) may gain more benefit by switching. New patients on treatment for AF should strongly be considered for DOAC therapy as the first line.

Background

The prevalence of AF rises with age, as does the prevalence of malignancy, limited mobility, and other comorbidities that increase the risk for VTEs. These factors may also increase the risk of bleeding with conventional therapy with heparins and VKAs. As such, understanding the implications of using DOACs in the elderly population is important.

Findings

This meta‐analysis included the elderly (age 75 years) subset of patients from existing AF treatment and VTE treatment and prophylaxis randomized trials comparing DOACs with VKAs, low‐molecular‐weight heparin (LMWH), aspirin, or placebo. The primary safety outcome was major bleeding. For AF trials, the efficacy endpoint was stroke or systemic embolization, whereas in VTE trials it was VTE or VTE‐related death. Authors were able to extract data on 25,031 patients across 10 trials that evaluated rivaroxaban, apixaban, and dabigatran (not edoxaban), with follow‐up data ranging from 35 days to 2 years. For safety outcomes, the 2 arms showed no statistical difference (DOAC: 6.4%; conventional therapy: 6.3%; OR: 1.02; 95% CI: 0.73‐1.43). For efficacy endpoints in VTE studies, DOACs were more effective (3.7% vs 7.0%; OR: 0.45; 95% CI: 0.27‐77; NNT=30). For AF, the efficacy analysis favored DOACs also (3.3% vs 4.7%; OR: 0.65; 95% CI: 0.48‐0.87; NNT=71). When analyzed by the efficacy of the individual DOAC, rivaroxaban and apixaban both appeared to outperform the VKA/LMWH arm for both VTE and AF treatment, whereas data on dabigatran were only available for AF, also showing an efficacy benefit. Individual DOAC analyses for safety endpoints showed all the 3 to be similar to VKA/LMWH.

Cautions

Authors note, however, that coexisting low body weight and renal insufficiency may influence dosing choices in this population. There are specific dosage recommendations in the elderly for some DOACs.

Implications

The use of DOACs in patients aged 75 years and older appears to confer a substantial efficacy advantage when used for treatment of VTE and AF patients. The safety data presented in this meta‐analysis suggest that this class is comparable to VKA/LMWH medications.

Background

There are over 870,000 hospital admissions yearly for skin infection, making it one of most common reasons for hospitalization in the United States.[18] Management often requires lengthy treatments with intravenous antibiotics, especially with the emergence of methicillin‐resistant Staphylococcus aureus. Results from 2 large randomized, double‐blinded, multicenter clinical trials were published looking at new once‐weekly intravenous antibiotics. Dalbavancin and oritavancin are both lipoglycopeptides in the same family as vancomycin. What is unique is that their serum drug concentrations exceed the minimum inhibitor concentrations for over a week. Both drugs were compared in noninferiority trials to vancomycin. The studies had similar outcomes. The dalbavancin results are presented below.

Findings

Researchers randomized 1312 patients with significant cellulitis, large abscess, or wound infection. Patients also had fever, leukocytosis, or bandemia, and the infection had to be deemed severe enough to require a minimum of 3 days of intravenous antibiotics. The patients could not have received any prior antibiotics. Over 80% of the patients had fevers, and more than half met the criteria for systemic inflammatory response syndrome. Patients were randomized to either dalbavancin (on day 1 and day 8) or vancomycin every 12 hours (1 gm or 15 mg/kg), with both groups receiving placebo dosing of the other drug. The blinded physicians could decide to switch to oral agent (placebo or linezolid in the vancomycin group) anytime after day 3, and the physicians could stop antibiotics anytime after day 10. Otherwise, all patients received 14 days of antibiotics.

The FDA‐approved outcome was cessation of spread of erythema at 48 to 72 hours and no fever at 3 independent readings. Results were similar in the dalbavancin group compared to the vancomycinlinezolid group (79.7% vs 79.8%). Dalbavancin was deemed noninferior to vancomycin. Blinded investigator's assessment of treatment success at 2 weeks was also similar (96% vs 96.7%, respectively). More treatment‐related adverse events occurred in the vancomycinlinezolid group (183 vs 139; P=0.02) and more deaths occurred in the vancomycin group (7 vs 1; P=0.03).

Cautions

These antibiotics have only been shown effective for complicated, acute bacterial skin infections. Their performance for other gram‐positive infections is unknown. In the future, it is possible that patients with severe skin infections will receive a dose of these antibiotics on hospital day 1 and be sent home with close follow‐up. However, that study has not been done yet to confirm efficacy and safety. Though the drugs appear safe, there needs to be more clinical use before they become standard of care, especially because of the long half‐life. Finally, these drugs are very expensive and provide broad spectrum gram‐positive coverage. They are not meant for a simple cellulitis.

Implications

These 2 new once‐weekly antibioticsdalbavancin and oritavancinare noninferior to vancomycin for acute bacterial skin infections. They provide alternative treatment choices for managing patients with significant infections requiring hospitalization. In the future, they may change the need for hospitalization of these patients or significantly reduce their length of stay. Though expensive, a significant reduction in hospitalization will offset costs.

Background

In 2013, a French study randomized adult family members of a patient undergoing cardiopulmonary resuscitation (CPR) occurring at home to either be invited to stay and watch the resuscitation or to have no specific invitation offered.[19] At 90 days, this study revealed that those who were invited to watch (and 79% did) had fewer symptoms of post‐traumatic stress disorder (PTSD) (27% vs 37%) and anxiety (15% vs 23%), though not depression, than did the group not offered the opportunity to watch (though 43% watched anyway). There were 570 subjects (family members) in the trial, of whom a greater number in the control arm declined to participate in a 90‐day follow‐up due to emotional distress. Notably, only 4% of the patients in this study undergoing CPR survived to day 28. Whether the apparent positive psychological impact of the offer to watch CPR for families was durable remained in question.

Findings

The study group followed the families up to 1 year. At that time, dropout rates were similar (with the assumption, as in the prior study, that those who dropped out of either arm had PTSD symptoms). At follow‐up, subjects were again assessed for PTSD, anxiety, and depression symptoms as well as for meeting criteria for having had a major depressive episode or complicated grief. Four hundred eight of the original 570 subjects were able to undergo reevaluation. The 1‐year results showed the group offered the chance to watch CPR had fewer PTSD symptoms (20% vs 32%) and depression symptoms (10% vs 16%), as well as fewer major depressive episodes (23% vs 31%) and less complicated grief (21% vs 36%) but without a durable impact on anxiety symptoms.

Cautions

The resuscitation efforts in question here occurred out of hospital (in the home). Part of the protocol for those family members observing CPR was that a clinician was assigned to stay with them and explain the resuscitation process as it occurred.

Implications

It is postulated that having the chance to observe CPR, if desired, may help the grieving process. This study clearly raises a question about the wisdom of routinely escorting patient's families out of the room during resuscitative efforts. It seems likely that the durable and important psychological effects observed in this study for family members would similarly persist in emergency department and inpatient settings, where staff can be with patients' families to talk them through the events they are witnessing. It is time to ask families if they prefer to stay and watch CPR and not automatically move them to a waiting room.

Background

The last drug approved by the Food and Drug Administration (FDA) for heart failure (HF) was 10 years ago.[2] The new PARADIGM (Prospective Comparison of ARNI With ACEI to Determine Impact on Global Mortality and Morbidity in Heart Failure) heart failure study comparing a novel combination drug of a neprilysin inhibitor and angiotensin receptor blocker (ARB) to an angiotensin‐converting enzyme (ACE) inhibitor has cardiologists considering a possible change in the HF treatment algorithm. Neprilysin is a naturally occurring enzyme that breaks down the protective vasoactive peptides (brain natriuretic peptide, atrial natriuretic peptide, and bradykinin) made by the heart and the body in HF. These vasoactive peptides function to increase vasodilation and block sodium and water reabsorption. This novel neprilysin inhibitor extends the life of these vasoactive peptides, thus enhancing their effect. By inhibiting both neprilysin and the renin‐angiotensin system, there should be additional improvement in HF management. The neprilysin inhibitor was combined with an ARB instead of an ACE inhibitor because of significant angioedema seen in earlier phase trials when combined with an ACE inhibitor. This is believed related to increases in bradykinin due to both agents.

Findings

In this multicenter, blinded, randomized trial, over 10,000 patients with known HF (ejection fraction<35%, New York Heart Association class II or higher) went through 2 run‐in periods to ensure tolerance of both enalapril and the study drug, a combination of a neprilysin inhibitor and valsartan (neprilysin‐I/ARB). Eventually 8442 patients underwent randomization to either enalapril (10 mg twice a day) or neprilysin‐I/ARB (200 mg twice a day). The primary outcome was a combination of cardiovascular mortality and heart failure hospitalizations. The trial was stopped early at 27 months because of overwhelming benefit with neprilysin‐I/ARB (21.8% vs 26.5%; P<0.001). There was a 20% reduction specifically in cardiovascular mortality (13.3% vs 16.5%; hazard ratio [HR]: 0.80; P<0.001). The number needed to treat (NNT) was 32. There was also a 21% reduction in the risk of hospitalization (P<0.001). More patients with neprilysin‐I/ARB had symptomatic hypotension (14% vs 9.2%; P<0.001) but patients on the ACE inhibitor experienced more cough, hyperkalemia, and increases in their serum creatinine.

Cautions

There are 2 reasons clinicians may not see the same results in practice. First, the trial was stopped early, which can sometimes exaggerate benefits.[3] Second, the 2 run‐in periods eliminated patients who could not tolerate the medications at the trial doses. Additionally, although the study's authors were independent, the trial was funded by a pharmaceutical company.

Implications

This new combination drug of a neprilysin inhibitor and valsartan shows great promise at reducing cardiovascular mortality and hospitalizations for heart failure compared to enalapril alone. Given the high morbidity and mortality of heart failure, having a new agent in the treatment algorithm will be useful to patients and physicians. The drug was just approved by the FDA in July 2015 and will likely be offered as an alternative to ACE inhibitors.

Background

It is a common practice to withhold intravenous contrast material from computed tomography (CT) scans in patients with even moderately poor renal function out of concern for causing contrast‐induced nephropathy (CIN). Our understanding of CIN is based largely on observational studies and outcomes of cardiac catheterizations, where larger amounts of contrast are given intra‐arterially into an atherosclerotic aorta.[4] The exact mechanism of injury is not clear, possibly from direct tubule toxicity or renal vasoconstriction.[5] CIN is defined as a rise in creatinine >0.5 mg/dL or >25% rise in serum creatinine 24 to 48 hours after receiving intravenous contrast. Although it is usually self‐limited, there is concern that patients who develop CIN have an increase risk of dialysis and death.[6] In the last few years, radiologists have started to question whether the risk of CIN is overstated. A recent meta‐analysis of 13 studies demonstrated a similar likelihood of acute kidney injury in patients regardless of receiving intravenous contrast.[7] If the true incidence of CIN after venous contrast is actually lower, this raises the question of whether we are unnecessarily withholding contrast from CTs and thereby reducing their diagnostic accuracy. Two 2014 observational studies provide additional evidence that the concern for CIN may be overstated.

Findings

The 2 Mayo Clinic studies used the same database. They looked at all patients who underwent a contrast‐enhanced or unenhanced thoracic, abdominal, or pelvic CT between January 2000 and December 2010 at the Mayo Clinic. After limiting the data to patients with pre‐ and post‐CT creatinine measurements and excluding anyone on dialysis, with preexisting acute kidney injury, or who had received additional contrast within 14 days, they ended up with 41,229 patients, mostly inpatients. All of the patients were assigned propensity scores based on risk factors for the development of CIN and whether they would likely receive contrast. The patients were then subdivided into 4 renal function subgroups based on estimated glomerular filtration rate (eGFR). The patients who received contrast were matched based on their propensity scores to those who did not received contrast within their eGFR subgroups. Unmatched patients were eliminated, leaving a cohort of 12,508 matched patients. The outcome of the first article was acute kidney injury (AKI) defined as a rise in creatinine >0.5 mg/dL at 24 to 48 hours. Though AKI rose with worsening eGFR subgroups (eGFR > 90 [1.2%] vs eGFR < 30 [14%]), the rates of AKI were the same regardless of contrast exposure. There was no statistical difference in any of the eGFR subgroups. The second study looked at important clinical outcomesdeath and the need for dialysis. There was no statistical difference for emergent dialysis (odds ratio [OR]: 0.96, P=0.89) or 30‐day mortality (HR: 0.97; P=0.45) regardless of whether the patients received contrast or not.

Cautions

In propensity matching, unmeasured confounders can bias the results. However, the issue of whether venous contrast causes CIN will unlikely be settled in a randomized controlled trial. For patients with severe renal failure (eGFR < 30), there were far fewer patients in this subgroup, making it harder to draw conclusions. The amount of venous contrast given was not provided. Finally, this study evaluated intravenous contrast for CTs, not intra‐arterial contrast.

Implications

These 2 studies raise doubt as to whether the incidence of AKI after contrast‐enhanced CT can be attributed to the contrast itself. What exactly causes the rise in creatinine is probably multifactorial including lab variation, hydration, blood pressure changes, nephrotoxic drugs, and comorbid disease. In trying to decide whether to obtain a contrast‐enhanced CT for patients with chronic kidney dysfunction, these studies provide more evidence to consider in the decision‐making process. A conversation with the radiologist about the benefits gained from using contrast in an individual patient may be of value.

Background

Numerous medications and therapeutic approaches have been studied to prevent incident delirium in hospitalized medical and surgical patients with varying success. Many of the tested medications also have the potential for significant undesirable side effects. An earlier small trial of melatonin appeared to have impressive efficacy for this purpose and be well tolerated, but the substance is not regulated by the FDA.[8] Ramelteon, a melatonin receptor agonist, is approved by the FDA for insomnia, and authors hypothesized that it, too, may be effective in delirium prevention.

Findings

This study was a multicenter, single‐blinded, randomized controlled trial of the melatonin‐agonist ramelteon versus placebo in elderly patients admitted to the hospital ward or ICU with serious medical conditions. Researchers excluded intubated patients or those with Lewy body dementia, psychiatric disorders, and severe liver disease. Patients received either ramelteon or placebo nightly for up to a week, and the primary end point was incident delirium as determined by a blinded observer using a validated assessment tool. Sixty‐seven patients were enrolled. The baseline characteristics in the arms of the trial were similar. In the placebo arm, 11 of 34 patients (32%) developed delirium during the 7‐day observation period. In the ramelteon arm, 1 of 33 (3%) developed delirium (P=0.003). The rate of drug discontinuation was the same in each arm.

Cautions

This study is small, and the single‐blinded design (the physicians and patients knew which group they were in but the observers did not) limits the validity of these results, mandating a larger double‐blinded trial.

Implications

Ramelteon showed a dramatic impact on preventing incident delirium on elderly hospitalized patients with serious medical conditions admitted to the ward or intensive care unit (ICU) (nonintubated) in this small study. If larger trials concur with the impact of this well‐tolerated and inexpensive medication, the potential for delirium incidence reduction could have a dramatic impact on how care for delirium‐vulnerable patients is conducted as well as the systems‐level costs associated with delirium care. Further studies of this class of medications are needed to more definitively establish its value in delirium prevention.

Background

Delirium is common in hospitalized elderly patients, and numerous studies show that there are both short‐ and long‐term implications of developing delirium. Using well studied and validated tools has made identifying delirium fairly straightforward, yet its treatment remains difficult. Additionally, differentiating which patients will have a simpler clinical course from those at risk for a more morbid one has proved challenging. Using the Confusion Assessment Method (CAM), both in its short (4‐item) and long (10‐item) forms, as the basis for a prognostication tool, would allow for future research on treatment to have a scale against which to measure impact, and would allow clinicians to anticipate which patients were more likely to have difficult clinical courses.

Findings

The CAM Severity (CAM‐S) score was derived in 1219 subjects participating in 2 ongoing studies: 1 included high‐risk medical inpatients 70 years old or older, and the other included similarly aged patients undergoing major orthopedic, general, or vascular surgeries. Outcomes data were not available for the surgical patients. The CAM items were rated as either present/absent or absent/mild/severe, depending on the item, with an associated score attached to each item such that the 4‐item CAM had a score of 0 to 7 and the 10‐item CAM 0 to 19 (Table 2). Clinical outcomes from the medical patients cohort showed a dose response with increasing CAM‐S scores with respect to length of stay, adjusted cost, combined 90‐day end points of skilled nursing facility placement or death, and 90‐day mortality. Specifically, for patients with a CAM‐S (short form) score of 5 to 7, the 90‐day rate of death or nursing home residence was 62%, whereas the 90‐day postdischarge mortality rate was 36%.

Cautions

The CAM‐S, like the CAM, may work less well in patients with hypoactive delirium. This scale has been applied in a surgical cohort, but study outcomes were not presented in this article. This absence limits our ability to apply these results to a surgical population presently.

Implications

This study demonstrates that in medical inpatients, the CAM‐S is effective for prognostication. Moreover, the study points out that high‐scoring patients on the CAM‐S have quite poor prognoses, with more than one‐third dying by 3 months. This finding suggests that an important use of the CAM‐S is to identify patients about whom goals of care discussions should be held and end‐of‐life planning initiated if not previously done.

Background

Lactulose has been the principle treatment for acute hepatic encephalopathy (HE) since 1966.[9] It theoretically works by lowering the pH of the colon and trapping ammonia as ammonium, which is then expelled. Alternatively, it may simply decrease transit time through the colon. In fact, earlier treatments for HE were cathartics such as magnesium salts. Unfortunately 20% tp 30% of patients are poor responders to lactulose, and patients do not like it. This new study tests whether a modern‐day cathartic, polyethylene glycol, works as well as lactulose.

Findings

In this unblinded, randomized controlled trial, patients presenting to the emergency department with acute HE were assigned to either lactulose 20 to 30 g for a minimum of 3 doses over 24 hours or 4 L of polyethylene glycol (PEG) over 4 hours. The2 groups were similar in severity and etiology of liver disease. Patients were allowed to have received 1 dose of lactulose given in the emergency department prior to study enrollment. They were excluded if taking rifaximin. The primary outcome was improvement in the hepatic encephalopathy scoring algorithm (HESA) by 1 grade at 24 hours.[10] The algorithm scores HE from 0 (no clinical findings of HE) to 5 (comatose). Initial mean HESA scores in the 2 groups were identical (2.3).

In the lactulose group, 13/25 (52%) improved by at least 1 HESA score at 24 hours. Two patients (8%) completely cleared with a HESA score of 0. In comparison, 21/23 (91%) in the PEG group improved at 24 hours, and 10/23 (43%) had cleared with a HESA score of 0 (P<0.01). The median time to HE resolution was 2 days in the lactulose group compared with 1 day in the PEG group (P=0.01). There were no differences in serious adverse events. The majority (76%) of the PEG group received the full 4 L of PEG.

Cautions

The main limitations of the trial were the small sample size, that it was a single‐center study, and the fact it was unblinded. Additionally, 80% of the PEG group received 1 dose of lactulose prior to enrollment. Statistically, more patients in the PEG group developed hypokalemia, which can worsen HE. Therefore, if PEG is used for acute HE, potassium will need to be monitored.

Implications

The results are intriguing and may represent a new possible treatment for acute HE once larger studies are done. Interestingly, the ammonia level dropped further in the lactulose group than the PEG group, yet there was more cognitive improvement in the PEG group. This raises questions about the role of ammonia and catharsis in HE. Although lactulose and rifaximin continue to be the standard of care, cathartics may be returning as a viable alternative.

Background

Nonselective ‐blockers (NSBBs) are considered the aspirin of hepatologists, as they are used for primary and secondary prevention of variceal bleeds in patients with cirrhosis.[11] Since the 1980s, their benefit in reducing bleeding risk has been known, and more recently there has been evidence that they may reduce the risk of developing ascites in patients with compensated cirrhosis. Yet, there has been some contradictory evidence suggesting reduced survival in patients with decompensated cirrhosis and infections on NSBBs. This has led to the window hypothesis of NSBBs in cirrhosis, where NSBBs are beneficial only during a certain window period during the progression of cirrhosis.[12] Early on in cirrhosis, before the development of varices or ascites, NSBBs have no benefit. As cirrhosis progresses and portal hypertension develops, NSBBs play a major role in reducing bleeding from varices. However, in advanced cirrhosis, NSBBs may become harmful. In theory, they block the body's attempt to increase cardiac output during situations of increased physiologic stress, resulting in decreased mean arterial pressure and perfusion. This, in turn, causes end‐organ damage and increased risk of death. When exactly this NSBB window closes is unclear. A 2014 study suggests the window should close when patients develop spontaneous bacterial peritonitis (SBP).

Findings

This retrospective study followed 607 consecutive patients seen at a liver transplant center in Vienna, Austria, from 2006 to 2011. All of the patients were followed from the time of their first paracentesis. They were excluded if SBP was diagnosed during the first paracentesis. Patients were grouped based on whether they took an NSBB. As expected, more patients on an NSBB had varices (90% vs 62%; P<0.001) and a lower mean heart rate (77.5 vs 83.9 beats/minute; P<0.001). However, the 2 groups were similar in mean arterial pressure, systolic blood pressure, Model for End‐Stage Liver Disease score (17.5), Childs Pugh Score (CPS) (50% were C), and in the etiology of cirrhosis (55% were from alcoholic liver disease). They followed the patients for development of SBP. The primary outcome was transplant‐free survival. For the patients who never developed SBP, there was a 25% reduction in the risk of death for those on an NSBB adjusted for varices and CPS stage (HR=0.75, P=0.027). However, for the 182 patients who developed SBP, those on an NSBB had a 58% increase risk of death, again adjusted for varices and CPS stage (HR=1.58, P=0.014). Among the patients who developed SBP, there was a higher risk of hepatorenal syndrome (HRS) within 90 days for those on an NSBB (24% vs 11%, P=0.027). Although the mean arterial pressures (MAP) had been similar in the 2 groups before SBP, after the development of SBP, those on an NSBB had a significantly lower MAP (77.2 vs 82.6 mm Hg, P=0.005).

Cautions

This is a retrospective study, and although the authors controlled for varices and CPS, it is still possible the 2 groups were not similar. Whether patients were actually taking the NSBB is unknown, and doses of the NSBB were variable.

Implications

This study provides more evidence for the NSBB window hypothesis in the treatment of patients with cirrhosis. It suggests that the window on NSBB closes when patients develop SBP, as NSBBs appear to increase mortality and the risk of HRS. Thus, NSBB therapy should probably be discontinued in cirrhotic patients developing SBP. The question is for how long? The editorial accompanying the article says permanently.[13]

Background

Based on early research studies, many quality and regulatory organizations have stressed the importance of assessing hospitalized patients' venous thromboembolism (VTE) risk and prophylaxing those patients at increased risk either pharmacologically or mechanically. In 2011, a meta‐analysis of 40 studies of medical and stroke patients including approximately 52,000 patients failed to demonstrate a mortality benefit, showing that for every 3 pulmonary embolisms (PEs) prevented, it caused 4 major bleeding episodes per 1000 patients.[14] A second study in 2011, a multicenter, randomized controlled trial with medically complex patients deemed high risk for VTE, also failed to demonstrate a mortality benefit.[15] Despite these and other trials showing questionable benefit, guidelines continue to recommend that high‐risk medical patients should get pharmacologic prophylaxis against VTE.

Findings

This retrospective cohort trial retrospectively evaluated a cohort of 20,794 medical patients (non‐ICU) across 35 hospitals, excluding those with a Caprini score of <2 (ie, low risk for VTE). The authors divided the hospitals into tertiles based on adherence to VTE prophylaxis guidelines. Patients were followed to 90 days after hospitalization with telephone calls (reaching 56%) and chart reviews (100% reviewed) to identify clinically evident VTE events, excluding those that occurred within the first 3 days of index hospitalization. The study identified no statistically significant differences among the tertiles in terms of VTE rates, either in the hospital or at 90 days, though the overall VTE event rate was low. Interestingly, 85% of events took place postdischarge. Subgroup analyses also failed to identify a population of medical patients who benefited from prophylaxis.

Cautions

Debate about whether the Caprini risk score is the best available VTE risk scoring system exists. This study also excluded surgical and ICU patients.

Implications

This trial adds to the mounting literature suggesting that current guidelines‐based pharmacologic VTE prophylaxis for medical patients may offer no clear benefit in terms of incident VTE events or mortality. Although it is not yet time to abandon VTE prophylaxis completely, this study does raise the important question of whether it is time to revisit the quality guidelines and regulatory standards around VTE prophylaxis in medical inpatients. It also highlights the difficulty in assessing medical patients for their VTE risk. Though this study is provocative and important for its real‐world setting, further studies are required.

Background

There have been 6 large published randomized controlled trials of direct oral anticoagulants (DOACs) versus vitamin K antagonists (VKAs) in patients with acute VTE. Study sizes range from approximately 2500 to over 8000 subjects. All showed no significant difference between the arms with respect to efficacy (VTE or VTE‐related death) but had variable results with respect to major bleeding risk, a major concern given the nonreversibility of this group of medications. Additionally, subgroup analysis within these studies was challenging given sample size issues.

Findings

These 6 studies were combined in a meta‐analysis to address the DOACs' overall efficacy and safety profile, as well as looking in prespecified subgroups. The meta‐analysis included data from over 27,000 patients, evenly divided between DOACs (edoxaban, apixaban, rivaroxaban, and dabigatran) and VKAs, with the time in the therapeutic range (TTR) in the VKA arm being 64%. Overall, the primary efficacy endpoint (VTE and VTE‐related death) was similar (DOACs relative tisk [RR]=0.90; 95% confidence interval [CI]: 0.77‐1.06) but major bleeding (DOACs RR=0.61; 95% CI: 0.45‐0.83; NNT=150) and combined fatal and intracranial bleeding (DOACs RR=0.37; 95% CI: 0.27‐0.68; NNT=314) favored the DOACs. In subgroup analysis, there was no efficacy difference between the therapeutic groups in the subset specifically with DVT or with PE, or with patients weighing >100 kg, though safety data in these subsets were not evaluable. Patients with creatinine clearances of 30 to 49 mL/min demonstrated similar efficacy in both treatment arms, and the safety analysis in this subset with moderate renal impairment was better in the DOAC arm. Cancer patients achieved better efficacy with similar safety with the DOACs, whereas elderly patients achieved both better safety and efficacy with DOACs.

Cautions

As yet, there are inadequate data on patients with more advanced renal failure (creatinine clearance <30 mL/min) to advise using DOACs in that subset. Also, as there were no data comparing cancer patients with VTE that investigated DOACs versus low molecular weight heparins (the standard of care rather than warfarin since the CLOT [Comparison of Low‐molecular‐weight heparin versus Oral anticoagulant Therapy] trial[16]), the current meta‐analysis does not yet answer whether DOACs should be used in this population despite the efficacy benefit noted in the subgroup analysis.

Implications

This large meta‐analysis strongly suggests we can achieve comparable treatment efficacy from the DOACs as with VKAs, with better safety profiles in patients with acute VTE. In the subset of patients with moderate renal impairment (creatinine clearance 3049 mL/min), it appears safe and effective to choose DOACs.

Background

Adding to the previously published meta‐analyses of the original phase 3 randomized trials regarding the DOACs' impact on the atrial fibrillation (AF) treatment safety and efficacy literature relative to VKAs, a 2013 trial, ENGAGE AF‐TIMI 48 (Effective Anticoagulation with Factor Xa Next Generation in Atrial FibrillationThrombolysis in Myocardial Infarction 48), with edoxaban was published and warrants inclusion to have a better opportunity to glean important subgroup information.[17]

Findings

This meta‐analysis included data on 71,683 patients, 42,411 in the DOAC arm and 29,272 in the warfarin arm, as 2 of the trials were3‐arm studies, comparing warfarin to a high dose and a low dose of the DOAC. Meta‐analyses of the 4 trials were broken down into a high‐dose subsetthe 2 high‐dose arms and the standard doses used in the other 2 trialsand a low‐dose subsetthe 2 low‐dose arms and the standard doses used in the other 2 trials. With respect to the efficacy endpoint (incident stroke or systemic embolization), the high‐dose subset analyses of the DOACs yielded a 19% reduction (P<0.0001; NNT=142) relative to the VKAs. The safety endpoint of major bleeding in this analysis identified a 14% reduction in the DOAC group that was nonsignificant (P=0.06). Within the high‐dose subset, analyses favored DOACs with respect to hemorrhagic stroke (51% reduction; P<0.0001; NNT=220), intracranial hemorrhage (52% reduction; P<0.0001; NNT=132), and overall mortality (10% reduction; P=0.0003; NNT=129), whereas they increased the risk of gastrointestinal bleeding (25% increase; P=0.043; NNH=185). There was no significant difference between DOACs and warfarin with respect to ischemic stroke. The low‐dose subset had similar overall results with even fewer hemorrhage strokes balancing a higher incidence of ischemic strokes in the DOAC arm than in warfarin. Other important subgroup analyses suggest the safety and efficacy impact of DOACs is significant for VKA‐naive and experienced patients, though only statistically so for VKA‐naive patients. Additionally, the anticoagulation centers included in the study that had a TTR <66% seemed to gain a safety advantage from the DOACs, whereas both TTR groups (<66% and 66%) appeared to achieve an efficacy benefit from DOACs.

Cautions

There are not sufficient data to suggest routinely switching patients tolerating and well managed on VKAs to DOACs for AF.

Implications

DOACs reduce stroke and systemic emboli in patients with AF without increasing intracranial bleeding or hemorrhagic stroke, though at the cost of increased gastrointestinal bleeding in patients on the high‐dose regimens. Those patients on the low‐dose regimens have even a lower hemorrhagic stroke risk, the benefit of which is negated by a higher than VKA risk of ischemic strokes. Centers with lower TTRs (and perhaps by extrapolation, those patients with more difficulty staying in the therapeutic range) may gain more benefit by switching. New patients on treatment for AF should strongly be considered for DOAC therapy as the first line.

Background

The prevalence of AF rises with age, as does the prevalence of malignancy, limited mobility, and other comorbidities that increase the risk for VTEs. These factors may also increase the risk of bleeding with conventional therapy with heparins and VKAs. As such, understanding the implications of using DOACs in the elderly population is important.

Findings

This meta‐analysis included the elderly (age 75 years) subset of patients from existing AF treatment and VTE treatment and prophylaxis randomized trials comparing DOACs with VKAs, low‐molecular‐weight heparin (LMWH), aspirin, or placebo. The primary safety outcome was major bleeding. For AF trials, the efficacy endpoint was stroke or systemic embolization, whereas in VTE trials it was VTE or VTE‐related death. Authors were able to extract data on 25,031 patients across 10 trials that evaluated rivaroxaban, apixaban, and dabigatran (not edoxaban), with follow‐up data ranging from 35 days to 2 years. For safety outcomes, the 2 arms showed no statistical difference (DOAC: 6.4%; conventional therapy: 6.3%; OR: 1.02; 95% CI: 0.73‐1.43). For efficacy endpoints in VTE studies, DOACs were more effective (3.7% vs 7.0%; OR: 0.45; 95% CI: 0.27‐77; NNT=30). For AF, the efficacy analysis favored DOACs also (3.3% vs 4.7%; OR: 0.65; 95% CI: 0.48‐0.87; NNT=71). When analyzed by the efficacy of the individual DOAC, rivaroxaban and apixaban both appeared to outperform the VKA/LMWH arm for both VTE and AF treatment, whereas data on dabigatran were only available for AF, also showing an efficacy benefit. Individual DOAC analyses for safety endpoints showed all the 3 to be similar to VKA/LMWH.

Cautions

Authors note, however, that coexisting low body weight and renal insufficiency may influence dosing choices in this population. There are specific dosage recommendations in the elderly for some DOACs.

Implications

The use of DOACs in patients aged 75 years and older appears to confer a substantial efficacy advantage when used for treatment of VTE and AF patients. The safety data presented in this meta‐analysis suggest that this class is comparable to VKA/LMWH medications.

Background

There are over 870,000 hospital admissions yearly for skin infection, making it one of most common reasons for hospitalization in the United States.[18] Management often requires lengthy treatments with intravenous antibiotics, especially with the emergence of methicillin‐resistant Staphylococcus aureus. Results from 2 large randomized, double‐blinded, multicenter clinical trials were published looking at new once‐weekly intravenous antibiotics. Dalbavancin and oritavancin are both lipoglycopeptides in the same family as vancomycin. What is unique is that their serum drug concentrations exceed the minimum inhibitor concentrations for over a week. Both drugs were compared in noninferiority trials to vancomycin. The studies had similar outcomes. The dalbavancin results are presented below.

Findings

Researchers randomized 1312 patients with significant cellulitis, large abscess, or wound infection. Patients also had fever, leukocytosis, or bandemia, and the infection had to be deemed severe enough to require a minimum of 3 days of intravenous antibiotics. The patients could not have received any prior antibiotics. Over 80% of the patients had fevers, and more than half met the criteria for systemic inflammatory response syndrome. Patients were randomized to either dalbavancin (on day 1 and day 8) or vancomycin every 12 hours (1 gm or 15 mg/kg), with both groups receiving placebo dosing of the other drug. The blinded physicians could decide to switch to oral agent (placebo or linezolid in the vancomycin group) anytime after day 3, and the physicians could stop antibiotics anytime after day 10. Otherwise, all patients received 14 days of antibiotics.

The FDA‐approved outcome was cessation of spread of erythema at 48 to 72 hours and no fever at 3 independent readings. Results were similar in the dalbavancin group compared to the vancomycinlinezolid group (79.7% vs 79.8%). Dalbavancin was deemed noninferior to vancomycin. Blinded investigator's assessment of treatment success at 2 weeks was also similar (96% vs 96.7%, respectively). More treatment‐related adverse events occurred in the vancomycinlinezolid group (183 vs 139; P=0.02) and more deaths occurred in the vancomycin group (7 vs 1; P=0.03).

Cautions

These antibiotics have only been shown effective for complicated, acute bacterial skin infections. Their performance for other gram‐positive infections is unknown. In the future, it is possible that patients with severe skin infections will receive a dose of these antibiotics on hospital day 1 and be sent home with close follow‐up. However, that study has not been done yet to confirm efficacy and safety. Though the drugs appear safe, there needs to be more clinical use before they become standard of care, especially because of the long half‐life. Finally, these drugs are very expensive and provide broad spectrum gram‐positive coverage. They are not meant for a simple cellulitis.

Implications

These 2 new once‐weekly antibioticsdalbavancin and oritavancinare noninferior to vancomycin for acute bacterial skin infections. They provide alternative treatment choices for managing patients with significant infections requiring hospitalization. In the future, they may change the need for hospitalization of these patients or significantly reduce their length of stay. Though expensive, a significant reduction in hospitalization will offset costs.

Background

In 2013, a French study randomized adult family members of a patient undergoing cardiopulmonary resuscitation (CPR) occurring at home to either be invited to stay and watch the resuscitation or to have no specific invitation offered.[19] At 90 days, this study revealed that those who were invited to watch (and 79% did) had fewer symptoms of post‐traumatic stress disorder (PTSD) (27% vs 37%) and anxiety (15% vs 23%), though not depression, than did the group not offered the opportunity to watch (though 43% watched anyway). There were 570 subjects (family members) in the trial, of whom a greater number in the control arm declined to participate in a 90‐day follow‐up due to emotional distress. Notably, only 4% of the patients in this study undergoing CPR survived to day 28. Whether the apparent positive psychological impact of the offer to watch CPR for families was durable remained in question.

Findings

The study group followed the families up to 1 year. At that time, dropout rates were similar (with the assumption, as in the prior study, that those who dropped out of either arm had PTSD symptoms). At follow‐up, subjects were again assessed for PTSD, anxiety, and depression symptoms as well as for meeting criteria for having had a major depressive episode or complicated grief. Four hundred eight of the original 570 subjects were able to undergo reevaluation. The 1‐year results showed the group offered the chance to watch CPR had fewer PTSD symptoms (20% vs 32%) and depression symptoms (10% vs 16%), as well as fewer major depressive episodes (23% vs 31%) and less complicated grief (21% vs 36%) but without a durable impact on anxiety symptoms.

Cautions

The resuscitation efforts in question here occurred out of hospital (in the home). Part of the protocol for those family members observing CPR was that a clinician was assigned to stay with them and explain the resuscitation process as it occurred.

Implications

It is postulated that having the chance to observe CPR, if desired, may help the grieving process. This study clearly raises a question about the wisdom of routinely escorting patient's families out of the room during resuscitative efforts. It seems likely that the durable and important psychological effects observed in this study for family members would similarly persist in emergency department and inpatient settings, where staff can be with patients' families to talk them through the events they are witnessing. It is time to ask families if they prefer to stay and watch CPR and not automatically move them to a waiting room.

Background

The last drug approved by the Food and Drug Administration (FDA) for heart failure (HF) was 10 years ago.[2] The new PARADIGM (Prospective Comparison of ARNI With ACEI to Determine Impact on Global Mortality and Morbidity in Heart Failure) heart failure study comparing a novel combination drug of a neprilysin inhibitor and angiotensin receptor blocker (ARB) to an angiotensin‐converting enzyme (ACE) inhibitor has cardiologists considering a possible change in the HF treatment algorithm. Neprilysin is a naturally occurring enzyme that breaks down the protective vasoactive peptides (brain natriuretic peptide, atrial natriuretic peptide, and bradykinin) made by the heart and the body in HF. These vasoactive peptides function to increase vasodilation and block sodium and water reabsorption. This novel neprilysin inhibitor extends the life of these vasoactive peptides, thus enhancing their effect. By inhibiting both neprilysin and the renin‐angiotensin system, there should be additional improvement in HF management. The neprilysin inhibitor was combined with an ARB instead of an ACE inhibitor because of significant angioedema seen in earlier phase trials when combined with an ACE inhibitor. This is believed related to increases in bradykinin due to both agents.

Findings

In this multicenter, blinded, randomized trial, over 10,000 patients with known HF (ejection fraction<35%, New York Heart Association class II or higher) went through 2 run‐in periods to ensure tolerance of both enalapril and the study drug, a combination of a neprilysin inhibitor and valsartan (neprilysin‐I/ARB). Eventually 8442 patients underwent randomization to either enalapril (10 mg twice a day) or neprilysin‐I/ARB (200 mg twice a day). The primary outcome was a combination of cardiovascular mortality and heart failure hospitalizations. The trial was stopped early at 27 months because of overwhelming benefit with neprilysin‐I/ARB (21.8% vs 26.5%; P<0.001). There was a 20% reduction specifically in cardiovascular mortality (13.3% vs 16.5%; hazard ratio [HR]: 0.80; P<0.001). The number needed to treat (NNT) was 32. There was also a 21% reduction in the risk of hospitalization (P<0.001). More patients with neprilysin‐I/ARB had symptomatic hypotension (14% vs 9.2%; P<0.001) but patients on the ACE inhibitor experienced more cough, hyperkalemia, and increases in their serum creatinine.

Cautions

There are 2 reasons clinicians may not see the same results in practice. First, the trial was stopped early, which can sometimes exaggerate benefits.[3] Second, the 2 run‐in periods eliminated patients who could not tolerate the medications at the trial doses. Additionally, although the study's authors were independent, the trial was funded by a pharmaceutical company.

Implications

This new combination drug of a neprilysin inhibitor and valsartan shows great promise at reducing cardiovascular mortality and hospitalizations for heart failure compared to enalapril alone. Given the high morbidity and mortality of heart failure, having a new agent in the treatment algorithm will be useful to patients and physicians. The drug was just approved by the FDA in July 2015 and will likely be offered as an alternative to ACE inhibitors.

Background

It is a common practice to withhold intravenous contrast material from computed tomography (CT) scans in patients with even moderately poor renal function out of concern for causing contrast‐induced nephropathy (CIN). Our understanding of CIN is based largely on observational studies and outcomes of cardiac catheterizations, where larger amounts of contrast are given intra‐arterially into an atherosclerotic aorta.[4] The exact mechanism of injury is not clear, possibly from direct tubule toxicity or renal vasoconstriction.[5] CIN is defined as a rise in creatinine >0.5 mg/dL or >25% rise in serum creatinine 24 to 48 hours after receiving intravenous contrast. Although it is usually self‐limited, there is concern that patients who develop CIN have an increase risk of dialysis and death.[6] In the last few years, radiologists have started to question whether the risk of CIN is overstated. A recent meta‐analysis of 13 studies demonstrated a similar likelihood of acute kidney injury in patients regardless of receiving intravenous contrast.[7] If the true incidence of CIN after venous contrast is actually lower, this raises the question of whether we are unnecessarily withholding contrast from CTs and thereby reducing their diagnostic accuracy. Two 2014 observational studies provide additional evidence that the concern for CIN may be overstated.

Findings

The 2 Mayo Clinic studies used the same database. They looked at all patients who underwent a contrast‐enhanced or unenhanced thoracic, abdominal, or pelvic CT between January 2000 and December 2010 at the Mayo Clinic. After limiting the data to patients with pre‐ and post‐CT creatinine measurements and excluding anyone on dialysis, with preexisting acute kidney injury, or who had received additional contrast within 14 days, they ended up with 41,229 patients, mostly inpatients. All of the patients were assigned propensity scores based on risk factors for the development of CIN and whether they would likely receive contrast. The patients were then subdivided into 4 renal function subgroups based on estimated glomerular filtration rate (eGFR). The patients who received contrast were matched based on their propensity scores to those who did not received contrast within their eGFR subgroups. Unmatched patients were eliminated, leaving a cohort of 12,508 matched patients. The outcome of the first article was acute kidney injury (AKI) defined as a rise in creatinine >0.5 mg/dL at 24 to 48 hours. Though AKI rose with worsening eGFR subgroups (eGFR > 90 [1.2%] vs eGFR < 30 [14%]), the rates of AKI were the same regardless of contrast exposure. There was no statistical difference in any of the eGFR subgroups. The second study looked at important clinical outcomesdeath and the need for dialysis. There was no statistical difference for emergent dialysis (odds ratio [OR]: 0.96, P=0.89) or 30‐day mortality (HR: 0.97; P=0.45) regardless of whether the patients received contrast or not.

Cautions

In propensity matching, unmeasured confounders can bias the results. However, the issue of whether venous contrast causes CIN will unlikely be settled in a randomized controlled trial. For patients with severe renal failure (eGFR < 30), there were far fewer patients in this subgroup, making it harder to draw conclusions. The amount of venous contrast given was not provided. Finally, this study evaluated intravenous contrast for CTs, not intra‐arterial contrast.

Implications

These 2 studies raise doubt as to whether the incidence of AKI after contrast‐enhanced CT can be attributed to the contrast itself. What exactly causes the rise in creatinine is probably multifactorial including lab variation, hydration, blood pressure changes, nephrotoxic drugs, and comorbid disease. In trying to decide whether to obtain a contrast‐enhanced CT for patients with chronic kidney dysfunction, these studies provide more evidence to consider in the decision‐making process. A conversation with the radiologist about the benefits gained from using contrast in an individual patient may be of value.

Background

Numerous medications and therapeutic approaches have been studied to prevent incident delirium in hospitalized medical and surgical patients with varying success. Many of the tested medications also have the potential for significant undesirable side effects. An earlier small trial of melatonin appeared to have impressive efficacy for this purpose and be well tolerated, but the substance is not regulated by the FDA.[8] Ramelteon, a melatonin receptor agonist, is approved by the FDA for insomnia, and authors hypothesized that it, too, may be effective in delirium prevention.

Findings

This study was a multicenter, single‐blinded, randomized controlled trial of the melatonin‐agonist ramelteon versus placebo in elderly patients admitted to the hospital ward or ICU with serious medical conditions. Researchers excluded intubated patients or those with Lewy body dementia, psychiatric disorders, and severe liver disease. Patients received either ramelteon or placebo nightly for up to a week, and the primary end point was incident delirium as determined by a blinded observer using a validated assessment tool. Sixty‐seven patients were enrolled. The baseline characteristics in the arms of the trial were similar. In the placebo arm, 11 of 34 patients (32%) developed delirium during the 7‐day observation period. In the ramelteon arm, 1 of 33 (3%) developed delirium (P=0.003). The rate of drug discontinuation was the same in each arm.

Cautions

This study is small, and the single‐blinded design (the physicians and patients knew which group they were in but the observers did not) limits the validity of these results, mandating a larger double‐blinded trial.

Implications

Ramelteon showed a dramatic impact on preventing incident delirium on elderly hospitalized patients with serious medical conditions admitted to the ward or intensive care unit (ICU) (nonintubated) in this small study. If larger trials concur with the impact of this well‐tolerated and inexpensive medication, the potential for delirium incidence reduction could have a dramatic impact on how care for delirium‐vulnerable patients is conducted as well as the systems‐level costs associated with delirium care. Further studies of this class of medications are needed to more definitively establish its value in delirium prevention.

Background

Delirium is common in hospitalized elderly patients, and numerous studies show that there are both short‐ and long‐term implications of developing delirium. Using well studied and validated tools has made identifying delirium fairly straightforward, yet its treatment remains difficult. Additionally, differentiating which patients will have a simpler clinical course from those at risk for a more morbid one has proved challenging. Using the Confusion Assessment Method (CAM), both in its short (4‐item) and long (10‐item) forms, as the basis for a prognostication tool, would allow for future research on treatment to have a scale against which to measure impact, and would allow clinicians to anticipate which patients were more likely to have difficult clinical courses.

Findings

The CAM Severity (CAM‐S) score was derived in 1219 subjects participating in 2 ongoing studies: 1 included high‐risk medical inpatients 70 years old or older, and the other included similarly aged patients undergoing major orthopedic, general, or vascular surgeries. Outcomes data were not available for the surgical patients. The CAM items were rated as either present/absent or absent/mild/severe, depending on the item, with an associated score attached to each item such that the 4‐item CAM had a score of 0 to 7 and the 10‐item CAM 0 to 19 (Table 2). Clinical outcomes from the medical patients cohort showed a dose response with increasing CAM‐S scores with respect to length of stay, adjusted cost, combined 90‐day end points of skilled nursing facility placement or death, and 90‐day mortality. Specifically, for patients with a CAM‐S (short form) score of 5 to 7, the 90‐day rate of death or nursing home residence was 62%, whereas the 90‐day postdischarge mortality rate was 36%.

Cautions

The CAM‐S, like the CAM, may work less well in patients with hypoactive delirium. This scale has been applied in a surgical cohort, but study outcomes were not presented in this article. This absence limits our ability to apply these results to a surgical population presently.

Implications

This study demonstrates that in medical inpatients, the CAM‐S is effective for prognostication. Moreover, the study points out that high‐scoring patients on the CAM‐S have quite poor prognoses, with more than one‐third dying by 3 months. This finding suggests that an important use of the CAM‐S is to identify patients about whom goals of care discussions should be held and end‐of‐life planning initiated if not previously done.

Background

Lactulose has been the principle treatment for acute hepatic encephalopathy (HE) since 1966.[9] It theoretically works by lowering the pH of the colon and trapping ammonia as ammonium, which is then expelled. Alternatively, it may simply decrease transit time through the colon. In fact, earlier treatments for HE were cathartics such as magnesium salts. Unfortunately 20% tp 30% of patients are poor responders to lactulose, and patients do not like it. This new study tests whether a modern‐day cathartic, polyethylene glycol, works as well as lactulose.

Findings

In this unblinded, randomized controlled trial, patients presenting to the emergency department with acute HE were assigned to either lactulose 20 to 30 g for a minimum of 3 doses over 24 hours or 4 L of polyethylene glycol (PEG) over 4 hours. The2 groups were similar in severity and etiology of liver disease. Patients were allowed to have received 1 dose of lactulose given in the emergency department prior to study enrollment. They were excluded if taking rifaximin. The primary outcome was improvement in the hepatic encephalopathy scoring algorithm (HESA) by 1 grade at 24 hours.[10] The algorithm scores HE from 0 (no clinical findings of HE) to 5 (comatose). Initial mean HESA scores in the 2 groups were identical (2.3).

In the lactulose group, 13/25 (52%) improved by at least 1 HESA score at 24 hours. Two patients (8%) completely cleared with a HESA score of 0. In comparison, 21/23 (91%) in the PEG group improved at 24 hours, and 10/23 (43%) had cleared with a HESA score of 0 (P<0.01). The median time to HE resolution was 2 days in the lactulose group compared with 1 day in the PEG group (P=0.01). There were no differences in serious adverse events. The majority (76%) of the PEG group received the full 4 L of PEG.

Cautions

The main limitations of the trial were the small sample size, that it was a single‐center study, and the fact it was unblinded. Additionally, 80% of the PEG group received 1 dose of lactulose prior to enrollment. Statistically, more patients in the PEG group developed hypokalemia, which can worsen HE. Therefore, if PEG is used for acute HE, potassium will need to be monitored.

Implications

The results are intriguing and may represent a new possible treatment for acute HE once larger studies are done. Interestingly, the ammonia level dropped further in the lactulose group than the PEG group, yet there was more cognitive improvement in the PEG group. This raises questions about the role of ammonia and catharsis in HE. Although lactulose and rifaximin continue to be the standard of care, cathartics may be returning as a viable alternative.

Background

Nonselective ‐blockers (NSBBs) are considered the aspirin of hepatologists, as they are used for primary and secondary prevention of variceal bleeds in patients with cirrhosis.[11] Since the 1980s, their benefit in reducing bleeding risk has been known, and more recently there has been evidence that they may reduce the risk of developing ascites in patients with compensated cirrhosis. Yet, there has been some contradictory evidence suggesting reduced survival in patients with decompensated cirrhosis and infections on NSBBs. This has led to the window hypothesis of NSBBs in cirrhosis, where NSBBs are beneficial only during a certain window period during the progression of cirrhosis.[12] Early on in cirrhosis, before the development of varices or ascites, NSBBs have no benefit. As cirrhosis progresses and portal hypertension develops, NSBBs play a major role in reducing bleeding from varices. However, in advanced cirrhosis, NSBBs may become harmful. In theory, they block the body's attempt to increase cardiac output during situations of increased physiologic stress, resulting in decreased mean arterial pressure and perfusion. This, in turn, causes end‐organ damage and increased risk of death. When exactly this NSBB window closes is unclear. A 2014 study suggests the window should close when patients develop spontaneous bacterial peritonitis (SBP).

Findings

This retrospective study followed 607 consecutive patients seen at a liver transplant center in Vienna, Austria, from 2006 to 2011. All of the patients were followed from the time of their first paracentesis. They were excluded if SBP was diagnosed during the first paracentesis. Patients were grouped based on whether they took an NSBB. As expected, more patients on an NSBB had varices (90% vs 62%; P<0.001) and a lower mean heart rate (77.5 vs 83.9 beats/minute; P<0.001). However, the 2 groups were similar in mean arterial pressure, systolic blood pressure, Model for End‐Stage Liver Disease score (17.5), Childs Pugh Score (CPS) (50% were C), and in the etiology of cirrhosis (55% were from alcoholic liver disease). They followed the patients for development of SBP. The primary outcome was transplant‐free survival. For the patients who never developed SBP, there was a 25% reduction in the risk of death for those on an NSBB adjusted for varices and CPS stage (HR=0.75, P=0.027). However, for the 182 patients who developed SBP, those on an NSBB had a 58% increase risk of death, again adjusted for varices and CPS stage (HR=1.58, P=0.014). Among the patients who developed SBP, there was a higher risk of hepatorenal syndrome (HRS) within 90 days for those on an NSBB (24% vs 11%, P=0.027). Although the mean arterial pressures (MAP) had been similar in the 2 groups before SBP, after the development of SBP, those on an NSBB had a significantly lower MAP (77.2 vs 82.6 mm Hg, P=0.005).

Cautions

This is a retrospective study, and although the authors controlled for varices and CPS, it is still possible the 2 groups were not similar. Whether patients were actually taking the NSBB is unknown, and doses of the NSBB were variable.

Implications

This study provides more evidence for the NSBB window hypothesis in the treatment of patients with cirrhosis. It suggests that the window on NSBB closes when patients develop SBP, as NSBBs appear to increase mortality and the risk of HRS. Thus, NSBB therapy should probably be discontinued in cirrhotic patients developing SBP. The question is for how long? The editorial accompanying the article says permanently.[13]

Background

Based on early research studies, many quality and regulatory organizations have stressed the importance of assessing hospitalized patients' venous thromboembolism (VTE) risk and prophylaxing those patients at increased risk either pharmacologically or mechanically. In 2011, a meta‐analysis of 40 studies of medical and stroke patients including approximately 52,000 patients failed to demonstrate a mortality benefit, showing that for every 3 pulmonary embolisms (PEs) prevented, it caused 4 major bleeding episodes per 1000 patients.[14] A second study in 2011, a multicenter, randomized controlled trial with medically complex patients deemed high risk for VTE, also failed to demonstrate a mortality benefit.[15] Despite these and other trials showing questionable benefit, guidelines continue to recommend that high‐risk medical patients should get pharmacologic prophylaxis against VTE.

Findings

This retrospective cohort trial retrospectively evaluated a cohort of 20,794 medical patients (non‐ICU) across 35 hospitals, excluding those with a Caprini score of <2 (ie, low risk for VTE). The authors divided the hospitals into tertiles based on adherence to VTE prophylaxis guidelines. Patients were followed to 90 days after hospitalization with telephone calls (reaching 56%) and chart reviews (100% reviewed) to identify clinically evident VTE events, excluding those that occurred within the first 3 days of index hospitalization. The study identified no statistically significant differences among the tertiles in terms of VTE rates, either in the hospital or at 90 days, though the overall VTE event rate was low. Interestingly, 85% of events took place postdischarge. Subgroup analyses also failed to identify a population of medical patients who benefited from prophylaxis.

Cautions

Debate about whether the Caprini risk score is the best available VTE risk scoring system exists. This study also excluded surgical and ICU patients.

Implications

This trial adds to the mounting literature suggesting that current guidelines‐based pharmacologic VTE prophylaxis for medical patients may offer no clear benefit in terms of incident VTE events or mortality. Although it is not yet time to abandon VTE prophylaxis completely, this study does raise the important question of whether it is time to revisit the quality guidelines and regulatory standards around VTE prophylaxis in medical inpatients. It also highlights the difficulty in assessing medical patients for their VTE risk. Though this study is provocative and important for its real‐world setting, further studies are required.

Background

There have been 6 large published randomized controlled trials of direct oral anticoagulants (DOACs) versus vitamin K antagonists (VKAs) in patients with acute VTE. Study sizes range from approximately 2500 to over 8000 subjects. All showed no significant difference between the arms with respect to efficacy (VTE or VTE‐related death) but had variable results with respect to major bleeding risk, a major concern given the nonreversibility of this group of medications. Additionally, subgroup analysis within these studies was challenging given sample size issues.

Findings

These 6 studies were combined in a meta‐analysis to address the DOACs' overall efficacy and safety profile, as well as looking in prespecified subgroups. The meta‐analysis included data from over 27,000 patients, evenly divided between DOACs (edoxaban, apixaban, rivaroxaban, and dabigatran) and VKAs, with the time in the therapeutic range (TTR) in the VKA arm being 64%. Overall, the primary efficacy endpoint (VTE and VTE‐related death) was similar (DOACs relative tisk [RR]=0.90; 95% confidence interval [CI]: 0.77‐1.06) but major bleeding (DOACs RR=0.61; 95% CI: 0.45‐0.83; NNT=150) and combined fatal and intracranial bleeding (DOACs RR=0.37; 95% CI: 0.27‐0.68; NNT=314) favored the DOACs. In subgroup analysis, there was no efficacy difference between the therapeutic groups in the subset specifically with DVT or with PE, or with patients weighing >100 kg, though safety data in these subsets were not evaluable. Patients with creatinine clearances of 30 to 49 mL/min demonstrated similar efficacy in both treatment arms, and the safety analysis in this subset with moderate renal impairment was better in the DOAC arm. Cancer patients achieved better efficacy with similar safety with the DOACs, whereas elderly patients achieved both better safety and efficacy with DOACs.

Cautions

As yet, there are inadequate data on patients with more advanced renal failure (creatinine clearance <30 mL/min) to advise using DOACs in that subset. Also, as there were no data comparing cancer patients with VTE that investigated DOACs versus low molecular weight heparins (the standard of care rather than warfarin since the CLOT [Comparison of Low‐molecular‐weight heparin versus Oral anticoagulant Therapy] trial[16]), the current meta‐analysis does not yet answer whether DOACs should be used in this population despite the efficacy benefit noted in the subgroup analysis.

Implications

This large meta‐analysis strongly suggests we can achieve comparable treatment efficacy from the DOACs as with VKAs, with better safety profiles in patients with acute VTE. In the subset of patients with moderate renal impairment (creatinine clearance 3049 mL/min), it appears safe and effective to choose DOACs.

Background

Adding to the previously published meta‐analyses of the original phase 3 randomized trials regarding the DOACs' impact on the atrial fibrillation (AF) treatment safety and efficacy literature relative to VKAs, a 2013 trial, ENGAGE AF‐TIMI 48 (Effective Anticoagulation with Factor Xa Next Generation in Atrial FibrillationThrombolysis in Myocardial Infarction 48), with edoxaban was published and warrants inclusion to have a better opportunity to glean important subgroup information.[17]

Findings

This meta‐analysis included data on 71,683 patients, 42,411 in the DOAC arm and 29,272 in the warfarin arm, as 2 of the trials were3‐arm studies, comparing warfarin to a high dose and a low dose of the DOAC. Meta‐analyses of the 4 trials were broken down into a high‐dose subsetthe 2 high‐dose arms and the standard doses used in the other 2 trialsand a low‐dose subsetthe 2 low‐dose arms and the standard doses used in the other 2 trials. With respect to the efficacy endpoint (incident stroke or systemic embolization), the high‐dose subset analyses of the DOACs yielded a 19% reduction (P<0.0001; NNT=142) relative to the VKAs. The safety endpoint of major bleeding in this analysis identified a 14% reduction in the DOAC group that was nonsignificant (P=0.06). Within the high‐dose subset, analyses favored DOACs with respect to hemorrhagic stroke (51% reduction; P<0.0001; NNT=220), intracranial hemorrhage (52% reduction; P<0.0001; NNT=132), and overall mortality (10% reduction; P=0.0003; NNT=129), whereas they increased the risk of gastrointestinal bleeding (25% increase; P=0.043; NNH=185). There was no significant difference between DOACs and warfarin with respect to ischemic stroke. The low‐dose subset had similar overall results with even fewer hemorrhage strokes balancing a higher incidence of ischemic strokes in the DOAC arm than in warfarin. Other important subgroup analyses suggest the safety and efficacy impact of DOACs is significant for VKA‐naive and experienced patients, though only statistically so for VKA‐naive patients. Additionally, the anticoagulation centers included in the study that had a TTR <66% seemed to gain a safety advantage from the DOACs, whereas both TTR groups (<66% and 66%) appeared to achieve an efficacy benefit from DOACs.

Cautions

There are not sufficient data to suggest routinely switching patients tolerating and well managed on VKAs to DOACs for AF.

Implications

DOACs reduce stroke and systemic emboli in patients with AF without increasing intracranial bleeding or hemorrhagic stroke, though at the cost of increased gastrointestinal bleeding in patients on the high‐dose regimens. Those patients on the low‐dose regimens have even a lower hemorrhagic stroke risk, the benefit of which is negated by a higher than VKA risk of ischemic strokes. Centers with lower TTRs (and perhaps by extrapolation, those patients with more difficulty staying in the therapeutic range) may gain more benefit by switching. New patients on treatment for AF should strongly be considered for DOAC therapy as the first line.

Background

The prevalence of AF rises with age, as does the prevalence of malignancy, limited mobility, and other comorbidities that increase the risk for VTEs. These factors may also increase the risk of bleeding with conventional therapy with heparins and VKAs. As such, understanding the implications of using DOACs in the elderly population is important.

Findings

This meta‐analysis included the elderly (age 75 years) subset of patients from existing AF treatment and VTE treatment and prophylaxis randomized trials comparing DOACs with VKAs, low‐molecular‐weight heparin (LMWH), aspirin, or placebo. The primary safety outcome was major bleeding. For AF trials, the efficacy endpoint was stroke or systemic embolization, whereas in VTE trials it was VTE or VTE‐related death. Authors were able to extract data on 25,031 patients across 10 trials that evaluated rivaroxaban, apixaban, and dabigatran (not edoxaban), with follow‐up data ranging from 35 days to 2 years. For safety outcomes, the 2 arms showed no statistical difference (DOAC: 6.4%; conventional therapy: 6.3%; OR: 1.02; 95% CI: 0.73‐1.43). For efficacy endpoints in VTE studies, DOACs were more effective (3.7% vs 7.0%; OR: 0.45; 95% CI: 0.27‐77; NNT=30). For AF, the efficacy analysis favored DOACs also (3.3% vs 4.7%; OR: 0.65; 95% CI: 0.48‐0.87; NNT=71). When analyzed by the efficacy of the individual DOAC, rivaroxaban and apixaban both appeared to outperform the VKA/LMWH arm for both VTE and AF treatment, whereas data on dabigatran were only available for AF, also showing an efficacy benefit. Individual DOAC analyses for safety endpoints showed all the 3 to be similar to VKA/LMWH.

Cautions

Authors note, however, that coexisting low body weight and renal insufficiency may influence dosing choices in this population. There are specific dosage recommendations in the elderly for some DOACs.

Implications

The use of DOACs in patients aged 75 years and older appears to confer a substantial efficacy advantage when used for treatment of VTE and AF patients. The safety data presented in this meta‐analysis suggest that this class is comparable to VKA/LMWH medications.

Background

There are over 870,000 hospital admissions yearly for skin infection, making it one of most common reasons for hospitalization in the United States.[18] Management often requires lengthy treatments with intravenous antibiotics, especially with the emergence of methicillin‐resistant Staphylococcus aureus. Results from 2 large randomized, double‐blinded, multicenter clinical trials were published looking at new once‐weekly intravenous antibiotics. Dalbavancin and oritavancin are both lipoglycopeptides in the same family as vancomycin. What is unique is that their serum drug concentrations exceed the minimum inhibitor concentrations for over a week. Both drugs were compared in noninferiority trials to vancomycin. The studies had similar outcomes. The dalbavancin results are presented below.

Findings

Researchers randomized 1312 patients with significant cellulitis, large abscess, or wound infection. Patients also had fever, leukocytosis, or bandemia, and the infection had to be deemed severe enough to require a minimum of 3 days of intravenous antibiotics. The patients could not have received any prior antibiotics. Over 80% of the patients had fevers, and more than half met the criteria for systemic inflammatory response syndrome. Patients were randomized to either dalbavancin (on day 1 and day 8) or vancomycin every 12 hours (1 gm or 15 mg/kg), with both groups receiving placebo dosing of the other drug. The blinded physicians could decide to switch to oral agent (placebo or linezolid in the vancomycin group) anytime after day 3, and the physicians could stop antibiotics anytime after day 10. Otherwise, all patients received 14 days of antibiotics.

The FDA‐approved outcome was cessation of spread of erythema at 48 to 72 hours and no fever at 3 independent readings. Results were similar in the dalbavancin group compared to the vancomycinlinezolid group (79.7% vs 79.8%). Dalbavancin was deemed noninferior to vancomycin. Blinded investigator's assessment of treatment success at 2 weeks was also similar (96% vs 96.7%, respectively). More treatment‐related adverse events occurred in the vancomycinlinezolid group (183 vs 139; P=0.02) and more deaths occurred in the vancomycin group (7 vs 1; P=0.03).

Cautions

These antibiotics have only been shown effective for complicated, acute bacterial skin infections. Their performance for other gram‐positive infections is unknown. In the future, it is possible that patients with severe skin infections will receive a dose of these antibiotics on hospital day 1 and be sent home with close follow‐up. However, that study has not been done yet to confirm efficacy and safety. Though the drugs appear safe, there needs to be more clinical use before they become standard of care, especially because of the long half‐life. Finally, these drugs are very expensive and provide broad spectrum gram‐positive coverage. They are not meant for a simple cellulitis.

Implications

These 2 new once‐weekly antibioticsdalbavancin and oritavancinare noninferior to vancomycin for acute bacterial skin infections. They provide alternative treatment choices for managing patients with significant infections requiring hospitalization. In the future, they may change the need for hospitalization of these patients or significantly reduce their length of stay. Though expensive, a significant reduction in hospitalization will offset costs.

Background

In 2013, a French study randomized adult family members of a patient undergoing cardiopulmonary resuscitation (CPR) occurring at home to either be invited to stay and watch the resuscitation or to have no specific invitation offered.[19] At 90 days, this study revealed that those who were invited to watch (and 79% did) had fewer symptoms of post‐traumatic stress disorder (PTSD) (27% vs 37%) and anxiety (15% vs 23%), though not depression, than did the group not offered the opportunity to watch (though 43% watched anyway). There were 570 subjects (family members) in the trial, of whom a greater number in the control arm declined to participate in a 90‐day follow‐up due to emotional distress. Notably, only 4% of the patients in this study undergoing CPR survived to day 28. Whether the apparent positive psychological impact of the offer to watch CPR for families was durable remained in question.

Findings

The study group followed the families up to 1 year. At that time, dropout rates were similar (with the assumption, as in the prior study, that those who dropped out of either arm had PTSD symptoms). At follow‐up, subjects were again assessed for PTSD, anxiety, and depression symptoms as well as for meeting criteria for having had a major depressive episode or complicated grief. Four hundred eight of the original 570 subjects were able to undergo reevaluation. The 1‐year results showed the group offered the chance to watch CPR had fewer PTSD symptoms (20% vs 32%) and depression symptoms (10% vs 16%), as well as fewer major depressive episodes (23% vs 31%) and less complicated grief (21% vs 36%) but without a durable impact on anxiety symptoms.

Cautions

The resuscitation efforts in question here occurred out of hospital (in the home). Part of the protocol for those family members observing CPR was that a clinician was assigned to stay with them and explain the resuscitation process as it occurred.

Implications

It is postulated that having the chance to observe CPR, if desired, may help the grieving process. This study clearly raises a question about the wisdom of routinely escorting patient's families out of the room during resuscitative efforts. It seems likely that the durable and important psychological effects observed in this study for family members would similarly persist in emergency department and inpatient settings, where staff can be with patients' families to talk them through the events they are witnessing. It is time to ask families if they prefer to stay and watch CPR and not automatically move them to a waiting room.

Participants

The unit of observation for the prospective cohort study was the clinical acute‐care unit within hospitals. Sites were instructed to designate a pilot unit for the intervention that cared for medical or mixed medicalsurgical patient populations. Sites were also asked to provide outcome data for a clinically and organizationally similar non‐BOOST unit to provide a site‐matched control. Control units were matched by local site leadership based on comparable patient demographics, clinical mix, and extent of housestaff presence. An initial cohort of 6 hospitals in 2008 was followed by a second cohort of 24 hospitals initiated in 2009. All hospitals were invited to participate in the national effectiveness analysis, which required submission of readmission and length of stay data for both a BOOST intervention unit and a clinically matched control unit.

Description of the Intervention

The BOOST intervention consisted of 2 major sequential processes, planning and implementation, both facilitated by external site mentorsphysicians expert in QI and care transitionsfor a period of 12 months. Extensive background on the planning and implementation components is available at www.hospitalmedicine.org/BOOST. The planning process consisted of institutional self‐assessment, team development, enlistment of stakeholder support, and process mapping. This approach was intended to prioritize the list of evidence‐based tools in BOOST that would best address individual institutional contexts. Mentors encouraged sites to implement tools sequentially according to this local context analysis with the goal of complete implementation of the BOOST toolkit.

Mentor engagement with sites consisted of a 2‐day kickoff training on the BOOST tools, where site teams met their mentor and initiated development of structured action plans, followed by 5 to 6 scheduled phone calls in the subsequent 12 months. During these conference calls, mentors gauged progress and sought to help troubleshoot barriers to implementation. Some mentors also conducted a site visit with participant sites. Project BOOST provided sites with several collaborative activities including online webinars and an online listserv. Sites also received a quarterly newsletter.

Outcome Measures

The primary outcome was 30‐day rehospitalization defined as same hospital, all‐cause rehospitalization. Home discharges as well as discharges or transfers to other healthcare facilities were included in the discharge calculation. Elective or scheduled rehospitalizations as well as multiple rehospitalizations in the same 30‐day window were considered individual rehospitalization events. Rehospitalization was reported as a ratio of 30‐day rehospitalizations divided by live discharges in a calendar month. Length of stay was reported as the mean length of stay among live discharges in a calendar month. Outcomes were calculated at the participant site and then uploaded as overall monthly unit outcomes to a Web‐based research database.

To account for seasonal trends as well as marked variation in month‐to‐month rehospitalization rates identified in longitudinal data, we elected to compare 3‐month year‐over‐year averages to determine relative changes in readmission rates from the period prior to BOOST implementation to the period after BOOST implementation. We calculated averages for rehospitalization and length of stay in the 3‐month period preceding the sites' first reported month of front‐line implementation and in the corresponding 3‐month period in the subsequent calendar year. For example, if a site reported implementing its first tool in April 2010, the average readmission rate in the unit for January 2011 through March 2011 was subtracted from the average readmission rate for January 2010 through March 2010.

Sites were surveyed regarding tool implementation rates 6 months and 24 months after the 2009 kickoff training session. Surveys were electronically completed by site leaders in consultation with site team members. The survey identified new tool implementation as well as modification of existing care processes using the BOOST tools (admission risk assessment, discharge readiness checklist, teach back use, mandate regarding discharge summary completion, follow‐up phone calls to >80% of discharges). Use of a sixth tool, creation of individualized written discharge instructions, was not measured. We credited sites with tool implementation if they reported either de novo tool use or alteration of previous care processes influenced by BOOST tools.

Clinical outcome reporting was voluntary, and sites did not receive compensation and were not subject to penalty for the degree of implementation or outcome reporting. No patient‐level information was collected for the analysis, which was approved by the Northwestern University institutional review board.

Data Sources and Methods

Readmission and length of stay data, including the unit level readmission rate, as collected from administrative sources at each hospital, were collected using templated spreadsheet software between December 2008 and June 2010, after which data were loaded directly to a Web‐based data‐tracking platform. Sites were asked to load data as they became available. Sites were asked to report the number of study unit discharges as well as the number of those discharges readmitted within 30 days; however, reporting of the number of patient discharges was inconsistent across sites. Serial outreach consisting of monthly phone calls or email messaging to site leaders was conducted throughout 2011 to increase site participation in the project analysis.

Implementation date information was collected from 2 sources. The first was through online surveys distributed in November 2009 and April 2011. The second was through fields in the Web‐based data tracking platform to which sites uploaded data. In cases where disagreement was found between these 2 sources, the site leader was contacted for clarification.

Practice setting (community teaching, community nonteaching, academic medical center) was determined by site‐leader report within the Web‐based data tracking platform. Data for hospital characteristics (number of licensed beds and geographic region) were obtained from the American Hospital Association's Annual Survey of Hospitals.[18] Hospital region was characterized as West, South, Midwest, or Northeast.

Analysis

The null hypothesis was that no prepost difference existed in readmission rates within BOOST units, and no difference existed in the prepost change in readmission rates in BOOST units when compared to site‐matched control units. The Wilcoxon rank sum test was used to test whether observed changes described above were significantly different from 0, supporting rejection of the null hypotheses. We performed similar tests to determine the significance of observed changes in length of stay. We performed our analysis using SAS 9.3 (SAS Institute Inc., Cary, NC).

Participants

The unit of observation for the prospective cohort study was the clinical acute‐care unit within hospitals. Sites were instructed to designate a pilot unit for the intervention that cared for medical or mixed medicalsurgical patient populations. Sites were also asked to provide outcome data for a clinically and organizationally similar non‐BOOST unit to provide a site‐matched control. Control units were matched by local site leadership based on comparable patient demographics, clinical mix, and extent of housestaff presence. An initial cohort of 6 hospitals in 2008 was followed by a second cohort of 24 hospitals initiated in 2009. All hospitals were invited to participate in the national effectiveness analysis, which required submission of readmission and length of stay data for both a BOOST intervention unit and a clinically matched control unit.

Description of the Intervention

The BOOST intervention consisted of 2 major sequential processes, planning and implementation, both facilitated by external site mentorsphysicians expert in QI and care transitionsfor a period of 12 months. Extensive background on the planning and implementation components is available at www.hospitalmedicine.org/BOOST. The planning process consisted of institutional self‐assessment, team development, enlistment of stakeholder support, and process mapping. This approach was intended to prioritize the list of evidence‐based tools in BOOST that would best address individual institutional contexts. Mentors encouraged sites to implement tools sequentially according to this local context analysis with the goal of complete implementation of the BOOST toolkit.

Mentor engagement with sites consisted of a 2‐day kickoff training on the BOOST tools, where site teams met their mentor and initiated development of structured action plans, followed by 5 to 6 scheduled phone calls in the subsequent 12 months. During these conference calls, mentors gauged progress and sought to help troubleshoot barriers to implementation. Some mentors also conducted a site visit with participant sites. Project BOOST provided sites with several collaborative activities including online webinars and an online listserv. Sites also received a quarterly newsletter.

Outcome Measures

The primary outcome was 30‐day rehospitalization defined as same hospital, all‐cause rehospitalization. Home discharges as well as discharges or transfers to other healthcare facilities were included in the discharge calculation. Elective or scheduled rehospitalizations as well as multiple rehospitalizations in the same 30‐day window were considered individual rehospitalization events. Rehospitalization was reported as a ratio of 30‐day rehospitalizations divided by live discharges in a calendar month. Length of stay was reported as the mean length of stay among live discharges in a calendar month. Outcomes were calculated at the participant site and then uploaded as overall monthly unit outcomes to a Web‐based research database.

To account for seasonal trends as well as marked variation in month‐to‐month rehospitalization rates identified in longitudinal data, we elected to compare 3‐month year‐over‐year averages to determine relative changes in readmission rates from the period prior to BOOST implementation to the period after BOOST implementation. We calculated averages for rehospitalization and length of stay in the 3‐month period preceding the sites' first reported month of front‐line implementation and in the corresponding 3‐month period in the subsequent calendar year. For example, if a site reported implementing its first tool in April 2010, the average readmission rate in the unit for January 2011 through March 2011 was subtracted from the average readmission rate for January 2010 through March 2010.

Sites were surveyed regarding tool implementation rates 6 months and 24 months after the 2009 kickoff training session. Surveys were electronically completed by site leaders in consultation with site team members. The survey identified new tool implementation as well as modification of existing care processes using the BOOST tools (admission risk assessment, discharge readiness checklist, teach back use, mandate regarding discharge summary completion, follow‐up phone calls to >80% of discharges). Use of a sixth tool, creation of individualized written discharge instructions, was not measured. We credited sites with tool implementation if they reported either de novo tool use or alteration of previous care processes influenced by BOOST tools.

Clinical outcome reporting was voluntary, and sites did not receive compensation and were not subject to penalty for the degree of implementation or outcome reporting. No patient‐level information was collected for the analysis, which was approved by the Northwestern University institutional review board.

Data Sources and Methods

Readmission and length of stay data, including the unit level readmission rate, as collected from administrative sources at each hospital, were collected using templated spreadsheet software between December 2008 and June 2010, after which data were loaded directly to a Web‐based data‐tracking platform. Sites were asked to load data as they became available. Sites were asked to report the number of study unit discharges as well as the number of those discharges readmitted within 30 days; however, reporting of the number of patient discharges was inconsistent across sites. Serial outreach consisting of monthly phone calls or email messaging to site leaders was conducted throughout 2011 to increase site participation in the project analysis.

Implementation date information was collected from 2 sources. The first was through online surveys distributed in November 2009 and April 2011. The second was through fields in the Web‐based data tracking platform to which sites uploaded data. In cases where disagreement was found between these 2 sources, the site leader was contacted for clarification.

Practice setting (community teaching, community nonteaching, academic medical center) was determined by site‐leader report within the Web‐based data tracking platform. Data for hospital characteristics (number of licensed beds and geographic region) were obtained from the American Hospital Association's Annual Survey of Hospitals.[18] Hospital region was characterized as West, South, Midwest, or Northeast.

Analysis

The null hypothesis was that no prepost difference existed in readmission rates within BOOST units, and no difference existed in the prepost change in readmission rates in BOOST units when compared to site‐matched control units. The Wilcoxon rank sum test was used to test whether observed changes described above were significantly different from 0, supporting rejection of the null hypotheses. We performed similar tests to determine the significance of observed changes in length of stay. We performed our analysis using SAS 9.3 (SAS Institute Inc., Cary, NC).

Participants

The unit of observation for the prospective cohort study was the clinical acute‐care unit within hospitals. Sites were instructed to designate a pilot unit for the intervention that cared for medical or mixed medicalsurgical patient populations. Sites were also asked to provide outcome data for a clinically and organizationally similar non‐BOOST unit to provide a site‐matched control. Control units were matched by local site leadership based on comparable patient demographics, clinical mix, and extent of housestaff presence. An initial cohort of 6 hospitals in 2008 was followed by a second cohort of 24 hospitals initiated in 2009. All hospitals were invited to participate in the national effectiveness analysis, which required submission of readmission and length of stay data for both a BOOST intervention unit and a clinically matched control unit.

Description of the Intervention

The BOOST intervention consisted of 2 major sequential processes, planning and implementation, both facilitated by external site mentorsphysicians expert in QI and care transitionsfor a period of 12 months. Extensive background on the planning and implementation components is available at www.hospitalmedicine.org/BOOST. The planning process consisted of institutional self‐assessment, team development, enlistment of stakeholder support, and process mapping. This approach was intended to prioritize the list of evidence‐based tools in BOOST that would best address individual institutional contexts. Mentors encouraged sites to implement tools sequentially according to this local context analysis with the goal of complete implementation of the BOOST toolkit.

Mentor engagement with sites consisted of a 2‐day kickoff training on the BOOST tools, where site teams met their mentor and initiated development of structured action plans, followed by 5 to 6 scheduled phone calls in the subsequent 12 months. During these conference calls, mentors gauged progress and sought to help troubleshoot barriers to implementation. Some mentors also conducted a site visit with participant sites. Project BOOST provided sites with several collaborative activities including online webinars and an online listserv. Sites also received a quarterly newsletter.

Outcome Measures

The primary outcome was 30‐day rehospitalization defined as same hospital, all‐cause rehospitalization. Home discharges as well as discharges or transfers to other healthcare facilities were included in the discharge calculation. Elective or scheduled rehospitalizations as well as multiple rehospitalizations in the same 30‐day window were considered individual rehospitalization events. Rehospitalization was reported as a ratio of 30‐day rehospitalizations divided by live discharges in a calendar month. Length of stay was reported as the mean length of stay among live discharges in a calendar month. Outcomes were calculated at the participant site and then uploaded as overall monthly unit outcomes to a Web‐based research database.

To account for seasonal trends as well as marked variation in month‐to‐month rehospitalization rates identified in longitudinal data, we elected to compare 3‐month year‐over‐year averages to determine relative changes in readmission rates from the period prior to BOOST implementation to the period after BOOST implementation. We calculated averages for rehospitalization and length of stay in the 3‐month period preceding the sites' first reported month of front‐line implementation and in the corresponding 3‐month period in the subsequent calendar year. For example, if a site reported implementing its first tool in April 2010, the average readmission rate in the unit for January 2011 through March 2011 was subtracted from the average readmission rate for January 2010 through March 2010.

Sites were surveyed regarding tool implementation rates 6 months and 24 months after the 2009 kickoff training session. Surveys were electronically completed by site leaders in consultation with site team members. The survey identified new tool implementation as well as modification of existing care processes using the BOOST tools (admission risk assessment, discharge readiness checklist, teach back use, mandate regarding discharge summary completion, follow‐up phone calls to >80% of discharges). Use of a sixth tool, creation of individualized written discharge instructions, was not measured. We credited sites with tool implementation if they reported either de novo tool use or alteration of previous care processes influenced by BOOST tools.

Clinical outcome reporting was voluntary, and sites did not receive compensation and were not subject to penalty for the degree of implementation or outcome reporting. No patient‐level information was collected for the analysis, which was approved by the Northwestern University institutional review board.

Data Sources and Methods

Readmission and length of stay data, including the unit level readmission rate, as collected from administrative sources at each hospital, were collected using templated spreadsheet software between December 2008 and June 2010, after which data were loaded directly to a Web‐based data‐tracking platform. Sites were asked to load data as they became available. Sites were asked to report the number of study unit discharges as well as the number of those discharges readmitted within 30 days; however, reporting of the number of patient discharges was inconsistent across sites. Serial outreach consisting of monthly phone calls or email messaging to site leaders was conducted throughout 2011 to increase site participation in the project analysis.

Implementation date information was collected from 2 sources. The first was through online surveys distributed in November 2009 and April 2011. The second was through fields in the Web‐based data tracking platform to which sites uploaded data. In cases where disagreement was found between these 2 sources, the site leader was contacted for clarification.

Practice setting (community teaching, community nonteaching, academic medical center) was determined by site‐leader report within the Web‐based data tracking platform. Data for hospital characteristics (number of licensed beds and geographic region) were obtained from the American Hospital Association's Annual Survey of Hospitals.[18] Hospital region was characterized as West, South, Midwest, or Northeast.

Analysis

The null hypothesis was that no prepost difference existed in readmission rates within BOOST units, and no difference existed in the prepost change in readmission rates in BOOST units when compared to site‐matched control units. The Wilcoxon rank sum test was used to test whether observed changes described above were significantly different from 0, supporting rejection of the null hypotheses. We performed similar tests to determine the significance of observed changes in length of stay. We performed our analysis using SAS 9.3 (SAS Institute Inc., Cary, NC).

Addressing Barriers to Medication Reconciliation

In order to implement successful medication reconciliation processes, one must build the steps with the patient and family/caregiver as the focus and demonstrate an understanding of the intent of these processes. At its roots, medication reconciliation was developed to ensure that clinicians do not inadvertently add, change, or omit medications and that changes made are communicated to all relevant caregivers.

A number of key issues with respect to successful medication reconciliation processes surfaced in discussions with stakeholders. We believe addressing these issues is necessary before meaningful and standardized implementation can be achieved. After each discussion below, we provide suggested first steps to address these issues.

1. Achieve Consensus on the Definition of Medication and Reconciliation

Despite proposed definitions of these terms by various organizations, there was little agreement about them in the healthcare community. This ambiguity contributed to general confusion about what actually constitutes medication reconciliation. There needs to be a single, clear, and broadly accepted definition of what constitutes a medication. For the purposes of medication reconciliation, the term medication should be broadly inclusive of substances that may have an impact on the patient's care and treatments as well as those substances that may interact with other therapies potentially used during the medical care episode. Illicit or recreational substances may also have impact on therapies considered and therefore may influence this definition.12 Concretely, this definition should encompass prescription and over‐the‐counter medications as well as herbal and dietary supplements.

The term reconciliation in its simplest form implies the process of verifying that a patient's current list of medications (including dose, route, and frequency) are correct and that the medications are currently medically necessary and safe. Reconciliation suggests a process which, by necessity, will vary based on clinical context and setting. Further defining this termand the process of reconciliation itselfshould be carried out using patient safety principles with a focus on patient‐ and family‐centeredness.

Designing hospital‐based medication reconciliation processes should:

• Employ a multidisciplinary approach that involves nurses, pharmacists, and other appropriate personnel from the inpatient setting as well as ambulatory and community/retail areas, both ambulatory and inpatient physicians, and a patient/family representative;

• Involve hospital leaders who support, provide guidance, and remove barriers for the multidisciplinary team working to implement the processes;

• Clearly define the roles of each participant in the processes developed;

• Include methods to assess and address any special needs due to the developmental stage, age, dependency, language or literacy levels of patients and their family/caregiver;

• Use clinically relevant process measures (e.g., adherence to procedural steps) and outcome measures (e.g., change in the number of ADEs, unnecessary hospitalizations, or emergency department visits) where appropriate to assess the impact of the process;

• Include feedback systems to allow for clinically significant process improvement.

Once a common understanding of the terms and intent of medication reconciliation is achieved, it will be important for accrediting organizations, medical societies, quality improvement organizations, and other interested parties to adopt the same language.

2. Clarify Roles and Responsibilities

Given the differences in organizational and practice structures in hospitals and the varying numbers of health professionals involved in a patient's care, no one process design will meet the needs of all sites. As it is clear that interdisciplinary teams are best suited to develop, implement, and carry out complex patient‐centered processes like medication reconciliation, it is crucial that all involved parties have clearly defined roles and responsibilities, including patients and their families/caregivers. It is also important to recognize that these responsibilities may change depending on the dependency or vulnerability of the patient (e.g., children or geriatric patients) or the transition of care being undertaken by the patient (i.e., admission, transfer, or discharge), thus requiring sites to develop clear policies about these roles and responsibilities and how they may change in various situations.

3. Develop Measurement Tools

Ensuring that medication reconciliation processes result in clinically meaningful outcomes requires the development and standardization of a limited number of metrics that may be used by organizations and reported centrally for benchmarking. This core set of measures should be developed by clinical, quality, accreditation, and regulatory organizations (see #10 below) through a consensus building process utilizing multi‐stakeholder input. The set should be supplemented by additional site‐specific measures determined locally that focus on steps in the process itself and allow sites to perform continuous quality improvement. Sites should be encouraged to develop tools locally to support and facilitate organizational and professional adherence to medication reconciliation processes.

4. Phased Implementation

Ultimately, comprehensive medication reconciliation processes need to be implemented in hospitals. However, to succeed in integrating complex processes like medication reconciliation into routine hospital practices, implementation may be facilitated by using a phased approach to allow for participants to adapt new processes and procedures to the local environment iteratively. While the most appropriate phased approach to implementation will vary by site and setting, options for phasing might include:

• Starting with one clinical area or service.

• Starting with either the admission or discharge reconciliation process.

• Starting with a patient population at high risk for adverse events.

• Starting with a focus on high‐risk medications.13, 14

Irrespective of the phasing strategy employed, development of a clear and pragmatic schedule for the entire implementation process should be established. Phasing decisions should be made based on organizational resources and the clinical needs of the patient population within each clinical setting. As noted, the ultimate goal is to develop comprehensive reconciliation processes occurring during all significant care transitions (i.e., admission, service or site‐of‐care transfers, and discharge) for all hospitalized patients and involving all of their medications. Flexibility in design should be encouraged to ensure the processes can work within local workflow as long as progress toward this primary goal is made.

5. Develop Risk Stratification Systems

Medication‐related adverse events related to inadequate reconciliation are more likely to occur in hospitalized patients with certain identifiable risk factors. For example, the MATCH study documented that polypharmacy and age over 65 years were independently associated with increased risk for errors at the time of hospital admission.7 Other factors that may increase the likelihood of medication‐related adverse events at care transitions in the hospital might include: patients with multiple providers, developmental/cognitive impairment, dependency/vulnerability, multiple or high‐risk medications, or poor health literacy or limited English proficiency. Research is needed to elucidate these risk factors further.

An alert system for key risk factors for complications related to incompletely, inappropriately, or inaccurately completed medication reconciliation due to patient, clinician, or system factors should be developed, tested, and broadly implemented. Additionally, an alert system would help maintain vigilance toward this patient safety issue and, potentially, help focus additional resources on high‐risk patients. Such a tool has been tested in ambulatory settings.15

6. Study Interventions and Processes

Despite having been an NPSG since 2005, there is still a relative paucity of literature about broadly applicable and effective implementation strategies and demonstrated interventions that improve medication safety related to medication reconciliation. Some strategies that have shown to reduce medication errors at transitions include the involvement of pharmacist medication review on discharge16, 17 and the usefulness of planning by multidisciplinary groups.18 Other studies have outlined the continuing barriers to successful implementation of reconciliation, including the difficulty patients have in accurately recalling their current medications19 and the high cost in nurse and pharmacist time of tracking down a patient's ongoing prescriptions.20, 21 Studies evaluating potential solutions to overcome these and other common barriers are still needed.

Future research should focus on a comprehensive review of implementation strategies, (specifically including the role of health information technology‐based innovations) clinically relevant outcomes, and best practices, while being sensitive to the different needs of varying care settings (e.g., pediatric vs. adult centers, emergency departments vs. inpatient units, community hospital vs. academic medical center, etc.) as well as the resource requirements engendered in the interventions.

7. Disseminate Success

Best practices and lessons learned, especially those rigorously tested and driven by data, stratified by patient type, care setting (emergency department, intensive care, surgical ward, etc.) and institutional type (community, teaching, safety net, critical access, etc.) need to be disseminated so others can adopt and adapt them effectively. High‐quality case studies with clear explanations of successes, failures, and lessons learned may prove valuable sources of information. This knowledge should foster a learning community approach and accelerate implementation at new sites.

8. Promote the Personal Health Record

A fully integrated and transferable personal health record should be accepted as the standard for health information storage and interoperability, giving both the patient (or family/caregiver) and clinical providers access and ownership. Both the HL7 Continuity of Care Document (CCD) and the Continuity of Care Record (CCR) meet these criteria. The CCR was endorsed by the American Society for Testing and Materials22 and a coalition of other medical societies.23 Notably, CCR and CCD were recently adopted as standards for structured electronic health record (EHR) exchange through the July 2010 publication of the Final Rule of the Health Information Technology for Economic and Clinical Health Act provision of the American Recovery and Reinvestment Act of 2009 (ARRA/HITECH) and is now part of the formal US Department of Health and Human Services certification criteria for EHR technologies.24

Mandating a content exchange standard such as the CCR or the CCD should also have the desired effect of ensuring that patients (and their caregivers) become increasingly involved in maintaining an accurate list of the medications they take. Additionally, systems must be sufficiently flexible to address the unique medication management needs of children and geriatric patients. An electronic version of a personal health record is a promising method for improving consistency across care platforms, but to be implemented effectively the record must be compatible across all settings, including, where possible, the patient's home. All health care organizations, pharmacy systems, and insurers, must make medication reconciliation‐related interoperability and accessibility a priority as they pursue information technology strategies.

9. Promote Partnerships

At a broader health care system level, leveraging existing partnerships and creating new ones among health care, public/private sector‐affiliated organizations (e.g., community and mail order pharmacies, pharmaceutical organizations and manufacturers, and insurers), and public health organizations are extremely important mechanisms for broader scale impact. This view recognizes the numerous opportunities to educate and influence patients about medication safety outside the dyadic relationship of the clinician and patient in traditional clinical settings. Partnerships between health care and public entities may capitalize on these opportunities to foster adoption of healthy medication practices (e.g., maintaining an accurate and updated medication list), thereby supporting medication reconciliation efforts when individuals encounter health care settings. Partnership and information sharing could be enhanced through the use of a central coordinating body or coalition. This body could generate a shared common vision and contribute expertise to the myriad issues in medication reconciliation.

Partnerships should utilize the following:

• Social marketing techniques to engage the community. Included within this strategy must be a clear and compelling message that transmits the importance of safe medication practices. Current messages such as keep a list while important, do not offer enough of a sense of urgency or importance. A more powerful message could involve highly publicized medication errors or close calls that would resonate with a broad audience.

• Local and national champions. Such individuals should be trusted for their health knowledge (e.g., television health care reporters) or be prominent, influential, and trusted figures in other circles (e.g., clergy, politicians, movie celebrities). Indeed, taking advantage of popular media by weaving a theme into a movie or television program about medication safety may prove effective.

Relevant partnerships would include:

• Quality organizations partnering with other stakeholders to establish unambiguous and unified medication reconciliation standards across the care continuum.

• Health systems partnering with community pharmacy providers to ensure an uninterrupted communication link in both the inpatient and outpatient settings.

• Manufacturers and distributors of medications partnering with health care and public health organizations, the media, insurers and other constituents to promote the importance of maintaining and sharing an accurate list of medications.

• Public health systems partnering with community‐based organizations to encourage and promote the established standards for medication safety through messaging and educational campaigns.

All partnerships must consider issues of patient language and literacy as well as the needs of vulnerable populations in the scope of their activities.

10. Align Financial Incentives With Newly Developed Regulatory and Accreditation Requirements

Implementing and performing medication reconciliation takes time, particularly at the outset of a new program. Time requirements and associated costs are major barriers to undertaking comprehensive medication reconciliation, despite its recognized importance for reducing avoidable injury to patients. At present, systems that impede efficiency and slow hospital throughput may be discouraged due to their potential for having an adverse impact on access, finances, and other aspects of care delivery. Moreover, the changed economic climate with reduced hospital fiscal margins limits resources for new initiatives. Currently, failed medication reconciliationand the related avoidable adverse events, culminating in readmission to the hospital or emergency departmentyields additional revenue for hospitals and other providers in some reimbursement models.

Alignment of financial incentives that ensured adequate time and resources for appropriate medication reconciliation processes would facilitate implementation. Additionally, start‐up funding to create and implement these processes needs to be made available.

One example illustrating efforts to align payment policy with medication safety efforts occurred when the Office of the National Coordinator (ONC), in publishing its Final Rule under the 2009 HITECH Act,24 endorsed the importance of financially supporting proper medication reconciliation, particularly at first encounter and transitions in care, by requiring EHR systems seeking certification under the rule to support the care team in the task of reconciliation. For example, vendors will have to support the ability to compare 2 or more medication lists electronically, create medication lists, drug allergy lists, perform drug formulary look‐ups, drug‐drug and drug‐allergy checks, and support creating patient summaries after each visit or post discharge that include medication lists. The ONC, in defining Meaningful Use for eligible health care organizations, included in that definition the goal of exchanging meaningful clinical information among the professional health care teams. This goal is demonstrated through organizations reporting that they performed medication reconciliation for at least 50% of transitions of care in which the patient is transitioned into the care of the eligible professional or admitted to the eligible hospital's or Critical Access Hospital's inpatient or emergency department. Organizations able to demonstrate this level of compliance, along with other Meaningful Use requirements, will be eligible to receive stimulus funds through 2015 and avoid financial penalties that begin after that period.

Addressing Barriers to Medication Reconciliation

In order to implement successful medication reconciliation processes, one must build the steps with the patient and family/caregiver as the focus and demonstrate an understanding of the intent of these processes. At its roots, medication reconciliation was developed to ensure that clinicians do not inadvertently add, change, or omit medications and that changes made are communicated to all relevant caregivers.

A number of key issues with respect to successful medication reconciliation processes surfaced in discussions with stakeholders. We believe addressing these issues is necessary before meaningful and standardized implementation can be achieved. After each discussion below, we provide suggested first steps to address these issues.

1. Achieve Consensus on the Definition of Medication and Reconciliation

Despite proposed definitions of these terms by various organizations, there was little agreement about them in the healthcare community. This ambiguity contributed to general confusion about what actually constitutes medication reconciliation. There needs to be a single, clear, and broadly accepted definition of what constitutes a medication. For the purposes of medication reconciliation, the term medication should be broadly inclusive of substances that may have an impact on the patient's care and treatments as well as those substances that may interact with other therapies potentially used during the medical care episode. Illicit or recreational substances may also have impact on therapies considered and therefore may influence this definition.12 Concretely, this definition should encompass prescription and over‐the‐counter medications as well as herbal and dietary supplements.

The term reconciliation in its simplest form implies the process of verifying that a patient's current list of medications (including dose, route, and frequency) are correct and that the medications are currently medically necessary and safe. Reconciliation suggests a process which, by necessity, will vary based on clinical context and setting. Further defining this termand the process of reconciliation itselfshould be carried out using patient safety principles with a focus on patient‐ and family‐centeredness.

Designing hospital‐based medication reconciliation processes should:

• Employ a multidisciplinary approach that involves nurses, pharmacists, and other appropriate personnel from the inpatient setting as well as ambulatory and community/retail areas, both ambulatory and inpatient physicians, and a patient/family representative;

• Involve hospital leaders who support, provide guidance, and remove barriers for the multidisciplinary team working to implement the processes;

• Clearly define the roles of each participant in the processes developed;

• Include methods to assess and address any special needs due to the developmental stage, age, dependency, language or literacy levels of patients and their family/caregiver;

• Use clinically relevant process measures (e.g., adherence to procedural steps) and outcome measures (e.g., change in the number of ADEs, unnecessary hospitalizations, or emergency department visits) where appropriate to assess the impact of the process;

• Include feedback systems to allow for clinically significant process improvement.

Once a common understanding of the terms and intent of medication reconciliation is achieved, it will be important for accrediting organizations, medical societies, quality improvement organizations, and other interested parties to adopt the same language.

2. Clarify Roles and Responsibilities

Given the differences in organizational and practice structures in hospitals and the varying numbers of health professionals involved in a patient's care, no one process design will meet the needs of all sites. As it is clear that interdisciplinary teams are best suited to develop, implement, and carry out complex patient‐centered processes like medication reconciliation, it is crucial that all involved parties have clearly defined roles and responsibilities, including patients and their families/caregivers. It is also important to recognize that these responsibilities may change depending on the dependency or vulnerability of the patient (e.g., children or geriatric patients) or the transition of care being undertaken by the patient (i.e., admission, transfer, or discharge), thus requiring sites to develop clear policies about these roles and responsibilities and how they may change in various situations.

3. Develop Measurement Tools

Ensuring that medication reconciliation processes result in clinically meaningful outcomes requires the development and standardization of a limited number of metrics that may be used by organizations and reported centrally for benchmarking. This core set of measures should be developed by clinical, quality, accreditation, and regulatory organizations (see #10 below) through a consensus building process utilizing multi‐stakeholder input. The set should be supplemented by additional site‐specific measures determined locally that focus on steps in the process itself and allow sites to perform continuous quality improvement. Sites should be encouraged to develop tools locally to support and facilitate organizational and professional adherence to medication reconciliation processes.

4. Phased Implementation

Ultimately, comprehensive medication reconciliation processes need to be implemented in hospitals. However, to succeed in integrating complex processes like medication reconciliation into routine hospital practices, implementation may be facilitated by using a phased approach to allow for participants to adapt new processes and procedures to the local environment iteratively. While the most appropriate phased approach to implementation will vary by site and setting, options for phasing might include:

• Starting with one clinical area or service.

• Starting with either the admission or discharge reconciliation process.

• Starting with a patient population at high risk for adverse events.

• Starting with a focus on high‐risk medications.13, 14

Irrespective of the phasing strategy employed, development of a clear and pragmatic schedule for the entire implementation process should be established. Phasing decisions should be made based on organizational resources and the clinical needs of the patient population within each clinical setting. As noted, the ultimate goal is to develop comprehensive reconciliation processes occurring during all significant care transitions (i.e., admission, service or site‐of‐care transfers, and discharge) for all hospitalized patients and involving all of their medications. Flexibility in design should be encouraged to ensure the processes can work within local workflow as long as progress toward this primary goal is made.

5. Develop Risk Stratification Systems

Medication‐related adverse events related to inadequate reconciliation are more likely to occur in hospitalized patients with certain identifiable risk factors. For example, the MATCH study documented that polypharmacy and age over 65 years were independently associated with increased risk for errors at the time of hospital admission.7 Other factors that may increase the likelihood of medication‐related adverse events at care transitions in the hospital might include: patients with multiple providers, developmental/cognitive impairment, dependency/vulnerability, multiple or high‐risk medications, or poor health literacy or limited English proficiency. Research is needed to elucidate these risk factors further.

An alert system for key risk factors for complications related to incompletely, inappropriately, or inaccurately completed medication reconciliation due to patient, clinician, or system factors should be developed, tested, and broadly implemented. Additionally, an alert system would help maintain vigilance toward this patient safety issue and, potentially, help focus additional resources on high‐risk patients. Such a tool has been tested in ambulatory settings.15

6. Study Interventions and Processes

Despite having been an NPSG since 2005, there is still a relative paucity of literature about broadly applicable and effective implementation strategies and demonstrated interventions that improve medication safety related to medication reconciliation. Some strategies that have shown to reduce medication errors at transitions include the involvement of pharmacist medication review on discharge16, 17 and the usefulness of planning by multidisciplinary groups.18 Other studies have outlined the continuing barriers to successful implementation of reconciliation, including the difficulty patients have in accurately recalling their current medications19 and the high cost in nurse and pharmacist time of tracking down a patient's ongoing prescriptions.20, 21 Studies evaluating potential solutions to overcome these and other common barriers are still needed.

Future research should focus on a comprehensive review of implementation strategies, (specifically including the role of health information technology‐based innovations) clinically relevant outcomes, and best practices, while being sensitive to the different needs of varying care settings (e.g., pediatric vs. adult centers, emergency departments vs. inpatient units, community hospital vs. academic medical center, etc.) as well as the resource requirements engendered in the interventions.

7. Disseminate Success

Best practices and lessons learned, especially those rigorously tested and driven by data, stratified by patient type, care setting (emergency department, intensive care, surgical ward, etc.) and institutional type (community, teaching, safety net, critical access, etc.) need to be disseminated so others can adopt and adapt them effectively. High‐quality case studies with clear explanations of successes, failures, and lessons learned may prove valuable sources of information. This knowledge should foster a learning community approach and accelerate implementation at new sites.

8. Promote the Personal Health Record

A fully integrated and transferable personal health record should be accepted as the standard for health information storage and interoperability, giving both the patient (or family/caregiver) and clinical providers access and ownership. Both the HL7 Continuity of Care Document (CCD) and the Continuity of Care Record (CCR) meet these criteria. The CCR was endorsed by the American Society for Testing and Materials22 and a coalition of other medical societies.23 Notably, CCR and CCD were recently adopted as standards for structured electronic health record (EHR) exchange through the July 2010 publication of the Final Rule of the Health Information Technology for Economic and Clinical Health Act provision of the American Recovery and Reinvestment Act of 2009 (ARRA/HITECH) and is now part of the formal US Department of Health and Human Services certification criteria for EHR technologies.24

Mandating a content exchange standard such as the CCR or the CCD should also have the desired effect of ensuring that patients (and their caregivers) become increasingly involved in maintaining an accurate list of the medications they take. Additionally, systems must be sufficiently flexible to address the unique medication management needs of children and geriatric patients. An electronic version of a personal health record is a promising method for improving consistency across care platforms, but to be implemented effectively the record must be compatible across all settings, including, where possible, the patient's home. All health care organizations, pharmacy systems, and insurers, must make medication reconciliation‐related interoperability and accessibility a priority as they pursue information technology strategies.

9. Promote Partnerships

At a broader health care system level, leveraging existing partnerships and creating new ones among health care, public/private sector‐affiliated organizations (e.g., community and mail order pharmacies, pharmaceutical organizations and manufacturers, and insurers), and public health organizations are extremely important mechanisms for broader scale impact. This view recognizes the numerous opportunities to educate and influence patients about medication safety outside the dyadic relationship of the clinician and patient in traditional clinical settings. Partnerships between health care and public entities may capitalize on these opportunities to foster adoption of healthy medication practices (e.g., maintaining an accurate and updated medication list), thereby supporting medication reconciliation efforts when individuals encounter health care settings. Partnership and information sharing could be enhanced through the use of a central coordinating body or coalition. This body could generate a shared common vision and contribute expertise to the myriad issues in medication reconciliation.

Partnerships should utilize the following:

• Social marketing techniques to engage the community. Included within this strategy must be a clear and compelling message that transmits the importance of safe medication practices. Current messages such as keep a list while important, do not offer enough of a sense of urgency or importance. A more powerful message could involve highly publicized medication errors or close calls that would resonate with a broad audience.

• Local and national champions. Such individuals should be trusted for their health knowledge (e.g., television health care reporters) or be prominent, influential, and trusted figures in other circles (e.g., clergy, politicians, movie celebrities). Indeed, taking advantage of popular media by weaving a theme into a movie or television program about medication safety may prove effective.

Relevant partnerships would include:

• Quality organizations partnering with other stakeholders to establish unambiguous and unified medication reconciliation standards across the care continuum.

• Health systems partnering with community pharmacy providers to ensure an uninterrupted communication link in both the inpatient and outpatient settings.

• Manufacturers and distributors of medications partnering with health care and public health organizations, the media, insurers and other constituents to promote the importance of maintaining and sharing an accurate list of medications.

• Public health systems partnering with community‐based organizations to encourage and promote the established standards for medication safety through messaging and educational campaigns.

All partnerships must consider issues of patient language and literacy as well as the needs of vulnerable populations in the scope of their activities.

10. Align Financial Incentives With Newly Developed Regulatory and Accreditation Requirements

Implementing and performing medication reconciliation takes time, particularly at the outset of a new program. Time requirements and associated costs are major barriers to undertaking comprehensive medication reconciliation, despite its recognized importance for reducing avoidable injury to patients. At present, systems that impede efficiency and slow hospital throughput may be discouraged due to their potential for having an adverse impact on access, finances, and other aspects of care delivery. Moreover, the changed economic climate with reduced hospital fiscal margins limits resources for new initiatives. Currently, failed medication reconciliationand the related avoidable adverse events, culminating in readmission to the hospital or emergency departmentyields additional revenue for hospitals and other providers in some reimbursement models.

Alignment of financial incentives that ensured adequate time and resources for appropriate medication reconciliation processes would facilitate implementation. Additionally, start‐up funding to create and implement these processes needs to be made available.

One example illustrating efforts to align payment policy with medication safety efforts occurred when the Office of the National Coordinator (ONC), in publishing its Final Rule under the 2009 HITECH Act,24 endorsed the importance of financially supporting proper medication reconciliation, particularly at first encounter and transitions in care, by requiring EHR systems seeking certification under the rule to support the care team in the task of reconciliation. For example, vendors will have to support the ability to compare 2 or more medication lists electronically, create medication lists, drug allergy lists, perform drug formulary look‐ups, drug‐drug and drug‐allergy checks, and support creating patient summaries after each visit or post discharge that include medication lists. The ONC, in defining Meaningful Use for eligible health care organizations, included in that definition the goal of exchanging meaningful clinical information among the professional health care teams. This goal is demonstrated through organizations reporting that they performed medication reconciliation for at least 50% of transitions of care in which the patient is transitioned into the care of the eligible professional or admitted to the eligible hospital's or Critical Access Hospital's inpatient or emergency department. Organizations able to demonstrate this level of compliance, along with other Meaningful Use requirements, will be eligible to receive stimulus funds through 2015 and avoid financial penalties that begin after that period.

Addressing Barriers to Medication Reconciliation

In order to implement successful medication reconciliation processes, one must build the steps with the patient and family/caregiver as the focus and demonstrate an understanding of the intent of these processes. At its roots, medication reconciliation was developed to ensure that clinicians do not inadvertently add, change, or omit medications and that changes made are communicated to all relevant caregivers.

A number of key issues with respect to successful medication reconciliation processes surfaced in discussions with stakeholders. We believe addressing these issues is necessary before meaningful and standardized implementation can be achieved. After each discussion below, we provide suggested first steps to address these issues.

1. Achieve Consensus on the Definition of Medication and Reconciliation

Despite proposed definitions of these terms by various organizations, there was little agreement about them in the healthcare community. This ambiguity contributed to general confusion about what actually constitutes medication reconciliation. There needs to be a single, clear, and broadly accepted definition of what constitutes a medication. For the purposes of medication reconciliation, the term medication should be broadly inclusive of substances that may have an impact on the patient's care and treatments as well as those substances that may interact with other therapies potentially used during the medical care episode. Illicit or recreational substances may also have impact on therapies considered and therefore may influence this definition.12 Concretely, this definition should encompass prescription and over‐the‐counter medications as well as herbal and dietary supplements.

The term reconciliation in its simplest form implies the process of verifying that a patient's current list of medications (including dose, route, and frequency) are correct and that the medications are currently medically necessary and safe. Reconciliation suggests a process which, by necessity, will vary based on clinical context and setting. Further defining this termand the process of reconciliation itselfshould be carried out using patient safety principles with a focus on patient‐ and family‐centeredness.

Designing hospital‐based medication reconciliation processes should:

• Employ a multidisciplinary approach that involves nurses, pharmacists, and other appropriate personnel from the inpatient setting as well as ambulatory and community/retail areas, both ambulatory and inpatient physicians, and a patient/family representative;

• Involve hospital leaders who support, provide guidance, and remove barriers for the multidisciplinary team working to implement the processes;

• Clearly define the roles of each participant in the processes developed;

• Include methods to assess and address any special needs due to the developmental stage, age, dependency, language or literacy levels of patients and their family/caregiver;

• Use clinically relevant process measures (e.g., adherence to procedural steps) and outcome measures (e.g., change in the number of ADEs, unnecessary hospitalizations, or emergency department visits) where appropriate to assess the impact of the process;

• Include feedback systems to allow for clinically significant process improvement.

Once a common understanding of the terms and intent of medication reconciliation is achieved, it will be important for accrediting organizations, medical societies, quality improvement organizations, and other interested parties to adopt the same language.

2. Clarify Roles and Responsibilities

Given the differences in organizational and practice structures in hospitals and the varying numbers of health professionals involved in a patient's care, no one process design will meet the needs of all sites. As it is clear that interdisciplinary teams are best suited to develop, implement, and carry out complex patient‐centered processes like medication reconciliation, it is crucial that all involved parties have clearly defined roles and responsibilities, including patients and their families/caregivers. It is also important to recognize that these responsibilities may change depending on the dependency or vulnerability of the patient (e.g., children or geriatric patients) or the transition of care being undertaken by the patient (i.e., admission, transfer, or discharge), thus requiring sites to develop clear policies about these roles and responsibilities and how they may change in various situations.

3. Develop Measurement Tools

Ensuring that medication reconciliation processes result in clinically meaningful outcomes requires the development and standardization of a limited number of metrics that may be used by organizations and reported centrally for benchmarking. This core set of measures should be developed by clinical, quality, accreditation, and regulatory organizations (see #10 below) through a consensus building process utilizing multi‐stakeholder input. The set should be supplemented by additional site‐specific measures determined locally that focus on steps in the process itself and allow sites to perform continuous quality improvement. Sites should be encouraged to develop tools locally to support and facilitate organizational and professional adherence to medication reconciliation processes.

4. Phased Implementation

Ultimately, comprehensive medication reconciliation processes need to be implemented in hospitals. However, to succeed in integrating complex processes like medication reconciliation into routine hospital practices, implementation may be facilitated by using a phased approach to allow for participants to adapt new processes and procedures to the local environment iteratively. While the most appropriate phased approach to implementation will vary by site and setting, options for phasing might include:

• Starting with one clinical area or service.

• Starting with either the admission or discharge reconciliation process.

• Starting with a patient population at high risk for adverse events.

• Starting with a focus on high‐risk medications.13, 14

Irrespective of the phasing strategy employed, development of a clear and pragmatic schedule for the entire implementation process should be established. Phasing decisions should be made based on organizational resources and the clinical needs of the patient population within each clinical setting. As noted, the ultimate goal is to develop comprehensive reconciliation processes occurring during all significant care transitions (i.e., admission, service or site‐of‐care transfers, and discharge) for all hospitalized patients and involving all of their medications. Flexibility in design should be encouraged to ensure the processes can work within local workflow as long as progress toward this primary goal is made.

5. Develop Risk Stratification Systems

Medication‐related adverse events related to inadequate reconciliation are more likely to occur in hospitalized patients with certain identifiable risk factors. For example, the MATCH study documented that polypharmacy and age over 65 years were independently associated with increased risk for errors at the time of hospital admission.7 Other factors that may increase the likelihood of medication‐related adverse events at care transitions in the hospital might include: patients with multiple providers, developmental/cognitive impairment, dependency/vulnerability, multiple or high‐risk medications, or poor health literacy or limited English proficiency. Research is needed to elucidate these risk factors further.

An alert system for key risk factors for complications related to incompletely, inappropriately, or inaccurately completed medication reconciliation due to patient, clinician, or system factors should be developed, tested, and broadly implemented. Additionally, an alert system would help maintain vigilance toward this patient safety issue and, potentially, help focus additional resources on high‐risk patients. Such a tool has been tested in ambulatory settings.15

6. Study Interventions and Processes

Despite having been an NPSG since 2005, there is still a relative paucity of literature about broadly applicable and effective implementation strategies and demonstrated interventions that improve medication safety related to medication reconciliation. Some strategies that have shown to reduce medication errors at transitions include the involvement of pharmacist medication review on discharge16, 17 and the usefulness of planning by multidisciplinary groups.18 Other studies have outlined the continuing barriers to successful implementation of reconciliation, including the difficulty patients have in accurately recalling their current medications19 and the high cost in nurse and pharmacist time of tracking down a patient's ongoing prescriptions.20, 21 Studies evaluating potential solutions to overcome these and other common barriers are still needed.

Future research should focus on a comprehensive review of implementation strategies, (specifically including the role of health information technology‐based innovations) clinically relevant outcomes, and best practices, while being sensitive to the different needs of varying care settings (e.g., pediatric vs. adult centers, emergency departments vs. inpatient units, community hospital vs. academic medical center, etc.) as well as the resource requirements engendered in the interventions.

7. Disseminate Success

Best practices and lessons learned, especially those rigorously tested and driven by data, stratified by patient type, care setting (emergency department, intensive care, surgical ward, etc.) and institutional type (community, teaching, safety net, critical access, etc.) need to be disseminated so others can adopt and adapt them effectively. High‐quality case studies with clear explanations of successes, failures, and lessons learned may prove valuable sources of information. This knowledge should foster a learning community approach and accelerate implementation at new sites.

8. Promote the Personal Health Record

A fully integrated and transferable personal health record should be accepted as the standard for health information storage and interoperability, giving both the patient (or family/caregiver) and clinical providers access and ownership. Both the HL7 Continuity of Care Document (CCD) and the Continuity of Care Record (CCR) meet these criteria. The CCR was endorsed by the American Society for Testing and Materials22 and a coalition of other medical societies.23 Notably, CCR and CCD were recently adopted as standards for structured electronic health record (EHR) exchange through the July 2010 publication of the Final Rule of the Health Information Technology for Economic and Clinical Health Act provision of the American Recovery and Reinvestment Act of 2009 (ARRA/HITECH) and is now part of the formal US Department of Health and Human Services certification criteria for EHR technologies.24

Mandating a content exchange standard such as the CCR or the CCD should also have the desired effect of ensuring that patients (and their caregivers) become increasingly involved in maintaining an accurate list of the medications they take. Additionally, systems must be sufficiently flexible to address the unique medication management needs of children and geriatric patients. An electronic version of a personal health record is a promising method for improving consistency across care platforms, but to be implemented effectively the record must be compatible across all settings, including, where possible, the patient's home. All health care organizations, pharmacy systems, and insurers, must make medication reconciliation‐related interoperability and accessibility a priority as they pursue information technology strategies.

9. Promote Partnerships

At a broader health care system level, leveraging existing partnerships and creating new ones among health care, public/private sector‐affiliated organizations (e.g., community and mail order pharmacies, pharmaceutical organizations and manufacturers, and insurers), and public health organizations are extremely important mechanisms for broader scale impact. This view recognizes the numerous opportunities to educate and influence patients about medication safety outside the dyadic relationship of the clinician and patient in traditional clinical settings. Partnerships between health care and public entities may capitalize on these opportunities to foster adoption of healthy medication practices (e.g., maintaining an accurate and updated medication list), thereby supporting medication reconciliation efforts when individuals encounter health care settings. Partnership and information sharing could be enhanced through the use of a central coordinating body or coalition. This body could generate a shared common vision and contribute expertise to the myriad issues in medication reconciliation.

Partnerships should utilize the following:

• Social marketing techniques to engage the community. Included within this strategy must be a clear and compelling message that transmits the importance of safe medication practices. Current messages such as keep a list while important, do not offer enough of a sense of urgency or importance. A more powerful message could involve highly publicized medication errors or close calls that would resonate with a broad audience.

• Local and national champions. Such individuals should be trusted for their health knowledge (e.g., television health care reporters) or be prominent, influential, and trusted figures in other circles (e.g., clergy, politicians, movie celebrities). Indeed, taking advantage of popular media by weaving a theme into a movie or television program about medication safety may prove effective.

Relevant partnerships would include:

• Quality organizations partnering with other stakeholders to establish unambiguous and unified medication reconciliation standards across the care continuum.

• Health systems partnering with community pharmacy providers to ensure an uninterrupted communication link in both the inpatient and outpatient settings.

• Manufacturers and distributors of medications partnering with health care and public health organizations, the media, insurers and other constituents to promote the importance of maintaining and sharing an accurate list of medications.

• Public health systems partnering with community‐based organizations to encourage and promote the established standards for medication safety through messaging and educational campaigns.

All partnerships must consider issues of patient language and literacy as well as the needs of vulnerable populations in the scope of their activities.

10. Align Financial Incentives With Newly Developed Regulatory and Accreditation Requirements

Implementing and performing medication reconciliation takes time, particularly at the outset of a new program. Time requirements and associated costs are major barriers to undertaking comprehensive medication reconciliation, despite its recognized importance for reducing avoidable injury to patients. At present, systems that impede efficiency and slow hospital throughput may be discouraged due to their potential for having an adverse impact on access, finances, and other aspects of care delivery. Moreover, the changed economic climate with reduced hospital fiscal margins limits resources for new initiatives. Currently, failed medication reconciliationand the related avoidable adverse events, culminating in readmission to the hospital or emergency departmentyields additional revenue for hospitals and other providers in some reimbursement models.

Alignment of financial incentives that ensured adequate time and resources for appropriate medication reconciliation processes would facilitate implementation. Additionally, start‐up funding to create and implement these processes needs to be made available.

One example illustrating efforts to align payment policy with medication safety efforts occurred when the Office of the National Coordinator (ONC), in publishing its Final Rule under the 2009 HITECH Act,24 endorsed the importance of financially supporting proper medication reconciliation, particularly at first encounter and transitions in care, by requiring EHR systems seeking certification under the rule to support the care team in the task of reconciliation. For example, vendors will have to support the ability to compare 2 or more medication lists electronically, create medication lists, drug allergy lists, perform drug formulary look‐ups, drug‐drug and drug‐allergy checks, and support creating patient summaries after each visit or post discharge that include medication lists. The ONC, in defining Meaningful Use for eligible health care organizations, included in that definition the goal of exchanging meaningful clinical information among the professional health care teams. This goal is demonstrated through organizations reporting that they performed medication reconciliation for at least 50% of transitions of care in which the patient is transitioned into the care of the eligible professional or admitted to the eligible hospital's or Critical Access Hospital's inpatient or emergency department. Organizations able to demonstrate this level of compliance, along with other Meaningful Use requirements, will be eligible to receive stimulus funds through 2015 and avoid financial penalties that begin after that period.