Top POEMs of 2011: Cardiovascular Disease

Dabigatran Cost-Effective for AF Prophylaxis in Some Patients, Not Others

Clinical Question: Is dabigatran cost-effective as compared with warfarin for patients with atrial fibrillation?

Bottom Line: Dabigatran is only cost-effective for patients who have poorly controlled INR, or who are at high risk of bleeding and/or stroke. It is worth pulling the article to review Figure 3, which can be used to guide point-of-care decision-making. (Level of Evidence: 2b)

Reference: Shah SV, Gage BF. Cost-effectiveness of dabigatran for stroke prophylaxis in atrial fibrillation. Circulation 2011;123(22):2562-2570.

Study Design: Cost-effectiveness analysis

Funding Source: Foundation

Setting: Population-based

Synopsis: Dabigatran, an oral direct thrombin inhibitor, is an expensive but more convenient alternative to warfarin, since it does not require monitoring. It is argued that although the direct drug cost of dabigatran is much higher than that of warfarin, the cost is offset by not needing visits for monitoring. This cost-effectiveness analysis, funded by the American Heart Association, attempts to determine whether this is true. The authors compared 6 strategies: no treatment, aspirin once daily, aspirin plus clopidogrel, warfarin, dabigatran 110 mg twice daily, and dabigatran 150 mg twice daily. The base case was the typical 70-year-old patient with atrial fibrillation, a moderate risk of stroke, and no contraindication to anticoagulation. The authors also varied the risk of stroke (based on the CHADS2 score) and risk of major bleeding (based on the HEMORR2HAGES score) to see the impact of changing risk on the optimal strategy. The authors used reasonable estimates of benefits, harms, and costs in their model, and they performed sensitivity analyses to study the effect of varying the base estimates. For example, the risk of stroke was varied from 0.8/100 to 13.7/100 person-years; major bleeding from 1.9/100 to 10.4/100 person-years; patient age from 60 years to 80 years; and time in the therapeutic INR range from less than 57% to more than 72%. In the Markov model, patients moved from one health state to another, or stayed in the same health state, each month. This was repeated for up to 20 years, with the costs and outcomes determined for each of the 6 strategies. For the base case, warfarin cost an additional $12,000 per quality adjusted life year (QALY) compared with aspirin. (A therapy is generally considered cost-effective if it costs less than $50,000/QALY). Dabigatran 150 mg cost $86,000/QALY using the base case assumption, while dabigatran 110 mg cost $150,000/QALY. Dual therapy with aspirin and clopidogrel resulted in fewer QALYs for higher cost, so it wasn't even in the running. For the base case, dabigatran was cost-effective for patients at high risk of stroke (CHADS2 score = 3 or higher) and those ar moderate risk of stroke but at high risk of bleeding. The sensitivity analysis also showed that for patients who had less than 57% time in the therapeutic range, dabigatran was the preferred option if the CHADS2 score was 2 or higher, but if patients spent more than 72% of time in range, dabigatran was never the preferred option. For patients with a CHADS2 score of 0 or 1, treatment with aspirin or no treatment was often the preferred strategy.

Associate Professor
University of Georgia
Athens, GA

Single Blood Pressure Measurements Inaccurate

Clinical Question: How many blood pressure readings are necessary to accurately diagnose hypertension?

Bottom Line: Several blood pressure measurements, taken over time, should be averaged to obtain a patient's "true" blood pressure. Using a single measurement to determine whether blood pressure is controlled will yield false results more than 20% of the time. This study used an automated method to measure blood pressure; results are likely even worse when blood pressure is measured manually (BMJ 2011;342:d286). (Level of Evidence: 1b)

Reference: Powers BJ, Olsen MK, Smith VA, Woolson RF, Bosworth HB, Oddone EZ. Measuring blood pressure for decision making and quality reporting: Where and how many measures? Ann Intern Med 2011;154(12):781-788.

Study Design: Cohort (prospective)

Funding Source: Government

Setting: Outpatient (any)

Synopsis: There are many things that can falsely elevate blood pressure -- the "white coat effect," poor technique, bad equipment, and so forth. In addition, blood pressure varies from moment to moment in an individual. In this study, the investigators evaluated repeated blood pressure measurements over time in a hypertension study, comparing results obtained in normal clinical practice, home measurement results, and measurements as part of the research protocol. The study was conducted in 444 men undergoing hypertension treatment (average age = 64 years) who had a total of 111,181 blood pressure determinations over 18 months. All patients had blood pressures measured by all 3 methods, though it doesn't appear that results were obtained in a masked manner. A patient's "true" blood pressure was calculated by modeling the average blood pressure over time with intrapatient variability. All 3 methods used automated devices. Over time, the average variation in systolic blood pressure in an individual was 10%, regardless of method of measurement. The research assessment, which probably had the closest adherence to protocol, resulted in patients being classified as having blood pressure controlled at baseline 68% of the time, compared with 47% with home measurement and 28% with clinic measurement. A single blood pressure measurement using any method was quite inaccurate when systolic blood pressure was between 120 mmHg and 157 mmHg. To obtain at least 80% certainty of an accurate blood pressure, several blood pressure readings must be taken over time (weeks to months) and averaged.

Professor of Family Medicine
Tufts University
Boston, MA

Automated Blood Pressure More Accurate in the Office

Clinical Question: Is an automated measure of systolic blood pressure more predictive of home blood pressure than manual measure?

Bottom Line: The use of an automated blood pressure measurement, with the patient alone in an examination room, provides readings closer to ambulatory measures and eliminates the "white-coat hypertension" that occurs when physicians take a patient's blood pressure. (Level of Evidence: 1b)

Reference: Myers MG, Godwin M, Dawes M, et al. Conventional versus automated measurement of blood pressure in primary care patients with systolic hypertension: randomised parallel design controlled trial. BMJ 2011;342:d286.

Study Design: Randomized controlled trial (nonblinded)

Funding Source: Government

Setting: Outpatient (primary care)

Synopsis: Blood pressure measurement, if done exactly right, takes 14 minutes -- a far cry from time it takes to do a typical office-setting measurement. As a consequence, blood pressure measurements in the office are often spuriously raised and not representative of a patient's blood pressure the other 23.95 hours of the day. The Canadian investigators conducting this study enrolled 555 patients of 67 family medicine practices. The practices, not the patients, were randomized into a group that used either automated office blood pressure measurement or typical, manual blood pressure measurement for patients with pre-established systolic hypertension, either treated or not treated. Physicians in the manual blood pressure group were not given specific instruction on technique. Results from both groups were compared with awake blood pressures measured with a 24-hour ambulatory blood pressure monitor. The automated measurement was closer to ambulatory measurement for systolic but not diastolic blood pressure, overestimating blood pressure by an average 2.3 mmHg as compared with 6.5 mmHg for the manual measurement (P = .006). Diastolic blood pressure differences were similar in both groups. Part of the issue is the presence of the physician; average readings dropped significantly with the automated measurement after the physician left the room.

Professor of Family Medicine
Tufts University
Boston, MA

Statin Therapy Decreases Cardiac Events

Clinical Question: Is statin therapy effective in preventing cardiac events?

Bottom Line: In this analysis of patient data pooled from multiple studies, intensive statin therapy was more effective than less intensive statin therapy in reducing the rate of major cardiovascular events. Additionally, statins are more effective than controls in preventing major cardiovascular events. In spite of the authors' attempts to link these data to low-density lipoprotein (LDL) concentrations, none of the trials randomly assigned patients to specific LDL targets. (Level of Evidence: 1a)

Reference: Cholesterol Treatment Trialists' (CTT) Collaboration, Baigent C, Blackwell L, Emberson J, et al. Efficacy and safety of more intensive lowering of LDL cholesterol: a meta-analysis of data from 170,000 participants in 26 randomised trials. Lancet 2010;376(9753):1670-1681.

Study Design: Meta-analysis (randomized controlled trials)

Funding Source: Government

Allocation: Concealed

Setting: Outpatient (any)

Synopsis: These authors pooled the patient-level data from 26 studies each involving at least 1000 patients and having at least 2 years of follow-up. Five of the trials compared intensive statin therapy with less intensive statin therapy and 21 compared statins with controls. Of the 5 trials comparing statin intensity, 2 evaluated 8659 patients with acute coronary syndromes (2.1 years of follow-up) and 3 evaluated 30,953 patients with stable coronary artery disease (5.8 years of follow-up). After 1 year of treatment, the LDL cholesterol levels decreased by an average 0.51 mmol/L (20 mg/dL). The annual rate of major vascular events (cardiovascular death, nonfatal myocardial infarction, revascularization, or stroke) was 4.5% in the intensive therapy group and 5.3% in the less intense therapy group (number needed to treat [NNT] = 200 per year). Of the 14 trials comparing statin therapy with control (128,596 patients with 4.8 years of follow-up), 6 appear to be primary prevention studies and the remainder were for secondary prevention. In these 14 studies, after 1 year of treatment the LDL cholesterol levels decreased by 1.07 mmol/L (41 mg/dL). The annual rate of major vascular events was 2.8% in the patients taking statins compared with 3.6% in patients taking a control agent (NNT = 125 per year). Although they don't report the annual rate of death from any cause for each treatment group, the authors report the total death rate for the intensively treated patients plus the statin-treated patients (2.1%) compared with the rate of the less intensively treated patients plus control patients (2.3%). This works out to a number needed to treat of 500 per year. The authors also try to correlate the outcomes data with the LDL levels achieved by the various interventions. However, since none of the trials actually randomized patients to specific lipid targets, this information should be interpreted cautiously and is best used to generate hypotheses. If you subscribe to the lipid theory of atherogenesis, you will love this part of the study. If you subscribe to alternate theories (eg, inflammation, plaque stability, and so forth) or are a methodologic purist, you will be annoyed by the authors' extrapolations.

Michigan State University
East Lansing, MI

Statins for Hyperlipidemia Reduce All-Cause Mortality

Clinical Question: Does statin therapy for primary and secondary prevention of cardiovascular disease among patients with elevated low-density lipoprotein levels reduce all-cause mortality?

Bottom Line: Statins reduce all-cause mortality among patients with hyperlipidemia, whether used for primary or secondary prevention of cardiovascular disease. This is a class effect. (Level of Evidence: 1a)

Reference: Mills EJ, Wu P, Chong G, et al. Efficacy and safety of statin treatment for cardiovascular disease: a network meta-analysis of 170,255 patients from 76 randomized trials. QJM 2011;104(2):109-124.

Study Design: Meta-analysis (randomized controlled trials)

Funding Source: Unknown/not stated

Setting: Various (meta-analysis)

Synopsis: This is a methodologically thorough meta-analysis of 76 randomized trials, including 170,255 patients, in which a statin was compared with placebo, standard therapy, or no treatment for primary or secondary prevention of cardiovascular disease (CVD). The authors included all identified randomized controlled trials of atorvastatin, fluvastatin, lovastatin, pitavastatin, rosuvastatin, and simvastatin. Head-to-head trials of statins were excluded. Women comprised 26% of study participants. The mean low-density lipoprotein (LDL) level at baseline was 180 mg/dL. Trials ranged in duration from 6 months to 6.1 years (mean = 2.7 years). The all-cause mortality rate was 8.1% in the statin groups and 9.5% in the control groups (relative risk [RR] = 0.90; 95% CI, 0.86-0.94; P < .0001; number needed to treat [NNT] = 67). A dose-response relationship was found with change in absolute LDL levels and did not differ between statins in univariate analysis; however, this effect was not observed in the multiple regression analysis. The reduction in all-cause mortality was largely attributable to a reduction in CVD deaths (4.1% vs 5.1%; RR = 0.80; 0.74-0.87; P < .001; NNT = 100). There were significant reductions in death due to myocardial infarction, and nonsignificant reductions in deaths due to stroke and non-CVD causes. Incident cancers and rhabdomyolosis were not significantly increased in statin groups. There was a slight increase in new incident diabetes (3.8% vs 3.5%; odds ratio = 1.09; 1.02-1.16; P = .008. Although relative risk reduction was consistent among primary prevention and secondary prevention populations, the absolute risk reduction would not be similar (absolute risk reduction was not reported).

Professor and Chair, Department of Family Medicine
University of Toledo
Toldeo, OH

POEMs are provided by Essential Evidence Plus, a point-of-care clinical decision support system published by Wiley-Blackwell. For more information, please see