The best evidence from ACP Journal Club for general internal medicine, 1999
ACP J Club. 1999 Sept-Oct;131:A13. doi:10.7326/ACPJC-1999-131-2-A13
From July 1998 to June 1999, ACP Journal Club presented the top 150 publications for internists, as judged by explicit criteria for scientific merit and the editors views on relevance to practice. For this annual recap, we highlighted the most important of these studies for general internal medicine, using an explicit and systematic process. We used a mixed nominal-Delphi group method to select and define criteria for ranking abstracts and articles (Table). 6 teams of 4 to 6 general internists, comprising clinician-educators, clinician-researchers, and residents in internal medicine, were trained to apply these criteria. Team leaders screened all 150 titles in ACP Journal Club tables of contents; 21 titles rejected by ≥ 3 leaders were not considered further.
Table. Ranking criteria for abstracts and articles*
|Relevance: addresses a condition relevant to general internists' practices, hospital, or office|
|Validity: likely to be true, with strong methods and low potential for bias|
|Applicability: likely to apply to patients and practice settings of general internists in North America|
|Feasibility: implementation would involve few barriers, few behavior changes, and little opportunity cost|
|Impact: likely to have a clinically meaningful effect on important outcomes (e.g., large benefit-to-harm ratio, modest benefit for a very common condition, change in practice because of proved ineffectiveness, or improved efficieny without loss of benefit)|
|Knowledge context: addresses an important gap in previous knowledge or identifies a gap in current knowledge|
*4-point ranking scale for each item: 1 = fair, 2 = good, 3 = very good, 4 = excellent.
Team members independently ranked the remaining 129 ACP Journal Club abstracts (without commentary) and original published abstracts, met to discuss their rankings, and sometimes rescored abstracts based on discussion (Table). The 35 abstracts with mean summary scores ≥ 20 were selected for full-text review. Members of 2 teams independently critiqued each selected article, then discussed rankings and gave final scores. Team leaders met over espresso (n = 1), soft drinks (n = 1), and margaritas (n = 4) to select the top articles based on the mean of the 2 team scores. We selected articles with scores ≥ 20, excluded those with scores ≤ 17, and debated (bloodlessly) about those remaining in the gray zone (18 to 19) until we reached consensus.
A surprising finding about hormone replacement therapy
Hormone replacement therapy (HRT) is widely prescribed to reduce the risk for coronary artery disease (CAD) in women. Surprising results from the Heart and Estrogen/progestin Replacement Study (HERS) challenge this practice (1). In HERS, 2763 postmenopausal women with CAD (mean age 67 y, 89% white) were randomly allocated to combined estrogen and progestin or placebo. After 4.1 years of follow-up, no difference was seen between the 2 groups in nonfatal myocardial infarction (MI) or CAD mortality (12.5% vs 12.7%), secondary cardiovascular outcomes, cancer mortality, or all-cause mortality. Greater rates of venous thromboembolism (0.025% vs 0.009%) and gallbladder disease (6% vs 4%) were seen in women receiving HRT. A significant time trend for initially more (0.04% vs 0.027%), then fewer (0.024% vs 0.035%), cardiac events hinted that HRT may have prothrombotic effects early in the course of therapy but antiatherosclerotic effects later.
HERS raises the disturbing possibility of selection bias toward women with multiple healthy behaviors in the many observational studies (e.g., case-control studies and cohort studies) that have provided most of the basis for HRT recommendations. It also complicates decision making for women and clinicians. Should women with known CAD not begin HRT? Should women with known CAD who have recently begun HRT discontinue use? For women without CAD who have numerous risk factors, will HRT prevent CAD or precipitate it? The surprising finding about secondary prevention sets the stage for the first large trial of HRT in primary prevention of CAD (results expected in 2005).
Erectile dysfunction: a new drug in the spotlight
An effective new treatment has increased public awareness and our treatment options for erectile dysfunction, a problem affecting 25% of 65- and 50% of 75-year-old men (2). Men with erectile dysfunction (n = 532, mean age 58 y, range 20 to 87 y) in whom a thorough history, physical examination, and laboratory testing revealed no reversible cause were randomly allocated to sildenafil (25, 50, or 100 mg) or placebo. Men receiving nitrates were excluded because of the risk for profound hypotension when the 2 drugs are combined. Follow-up was 24 weeks. Participants achieving penetration were 5% for placebo and 60%, 84%, and 100% (number needed to treat [NNT] [defined in Glossary] for the 3 sildenafil doses were 2, 2, and 1, respectively). Men receiving sildenafil reported increased quality of orgasmic function, intercourse satisfaction, and overall satisfaction. In a dose-escalation phase of the study, 74% (n = 329) used the 100-mg dose. The most common adverse effects were headache (30% vs 6%, number needed to harm [NNH] [defined in Glossary] 4), flushing (20% vs 1%, NNH 5), and dyspepsia (16% vs 1%, NNH 7) at the 100-mg dose.
Since publication of the trial, several reports of serious cardiovascular events and some deaths have been noted, usually in men with pre-existing cardiovascular disease. Thus, knowledge about the frequency of serious adverse effects is evolving. For now, sildenafil offers a relatively safe, effective, and convenient option for men with erectile dysfunction.
Broader scope of effectiveness for statins
Results of the large Air Force/Texas Coronary Atherosclerosis Prevention Study (AFCAPS/TexCAPS) strengthened the evidence that statins reduce major coronary events in individuals without known CAD (3). This study randomly allocated 6605 middle-aged or older healthy adults with "average" low-density lipoprotein (LDL) cholesterol levels (85% men, mean age 58 y, mean LDL cholesterol levels 9.9 mmol/L [150 mg/dL]) and low high-density lipoprotein (HDL) cholesterol levels (mean 1.0 mmol/L [38 mg/dL]) to diet plus either lovastatin or placebo. Treatment reduced LDL cholesterol levels by 25% and incidence of a first major coronary event (MI, unstable angina, or sudden cardiac death) by 37%. Absolute benefit was modest; NNT over 5.2 years to avoid an additional major coronary event was 49. No adverse events were reported. Cardiac and total mortality were not reduced, probably because of a lower baseline risk for CAD than that of any other statin trial.
2 major trials of lipid-lowering therapy in patients with established CAD and average LDL cholesterol levels were also published. The Long-term Intervention with Pravastatin in Ischemic Heart Disease (LIPID) study group trial randomly allocated 9014 patients (83% men, mean age 62 y, mean LDL cholesterol levels 3.9 mmol/L [150 mg/dL]) with acute MI or unstable angina to pravastatin, 40 mg/d, or placebo (4). Overall mortality was decreased by 22% with treatment (NNT over 6.1 years to prevent 1 additional death was 33), as were all cardiovascular end points. In the Cholesterol and Recurrent Events (CARE) trial, a subgroup analysis of 1283 elderly patients with MI (88% men, mean age 69 y, mean LDL cholesterol levels 3.6 mmol/L [138 mg/dL]) showed that pravastatin prevented recurrent coronary events (CAD death or nonfatal MI) by 39% over 5 years (NNT 15) (5). Finally, a meta-analysis of mainly secondary CAD prevention trials using statins showed reductions in stroke (risk ratio 0.76) and overall mortality (risk ratio 0.80) (6).
These studies suggest that statins are safe and effective for primary prevention of CAD events in men with low HDL cholesterol levels and for secondary prevention of atherosclerotic events and mortality in patients with average LDL cholesterol levels. Patients with a higher baseline risk for CAD clearly derive greater benefit from statins. Practitioners must continue to individualize treatment decisions, considering many factors, such as baseline risk, patient preferences, and drug costs.
Migraine: treatment and prophylaxis get simpler
Migraine affects approximately 18% of women and 6% of men in the United States; prevalence peaks between 25 and 55 years of age. Although many patients self-medicate with nonprescription drugs, most trials compare single-agent nonsteroidal anti-inflammatory drugs with placebo, prescription drugs, or drugs that are not available in the United States. 2 recent reports give us new, usable information about migraine management.
One report compared 2 tablets of Excedrin Extra-Strength (which contained acetaminophen, 500 mg; aspirin, 500 mg; and caffeine, 130 mg) with placebo for relief of a single episode (7). The study combined 3 randomized, double-blind, placebo-controlled trials of 1357 participants (mean age 37 y, 79% women) who met International Headache Society (IHS) criteria for migraine and had frequent (2 to 6/mo) but not disabling (vomiting < 20% of the time and requiring bed rest) migraines. The drug combination reduced pain intensity at 2 and 6 hours after dosing and increased the proportion of patients whose pain was mild or absent at 2 and 6 hours after dosing (NNT 4 for all).
For prophylaxis, a small (n = 55, mean age 36 y, 78% women), randomized, double-blind, placebo-controlled trial compared riboflavin, 400 mg/d, with placebo (8). Participants met IHS diagnostic criteria, had a history of migraine for ≥ 1 y, and had frequent (2 to 8 per mo) migraines. Riboflavin decreased the frequency of episodes and headache days in months 2, 3, and 4 compared with month 1. The NNT was 3 for 1 additional person to have a ≥ 50% reduction in migraine frequency or headache days.
For patients with uncomplicated migraines, both of these over-the-counter interventions offer easy, relatively inexpensive, and safe additions to proved measures for migraine.
Blood pressure targets and aspirin trade-offs identified
Studies reported landmark news for patients with diabetes mellitus and hypertension, 2 of the most common comorbid conditions managed by generalists. The U.K. Prospective Diabetes Study (UKPDS) trials confirmed a subset analysis from the Hypertension Optimal Treatment (HOT) trial showing that tight control of blood pressure (trial targets < 140 to 150 diastolic and 80 to 85 mm Hg systolic) reduced cardiovascular events and microvascular complications among patients with diabetes and hypertension (9, 10). UKPDS had 1148 participants (mean age 56 y, 55% men) and showed an NNT of about 6 for a period of 10 years to prevent 1 additional complication; to prevent 1 additional death from a diabetes-related cause, the NNT was about 15. Both trials, which evaluated first-line use of captopril, atenolol, or felodipine, indicated that 2 to 3 drugs are often needed to achieve tight control.
What about blood pressure targets in hypertensive patients without diabetes mellitus? The HOT trial showed neither significant reductions nor increases in cardiovascular events with tight control (the first group, which had a diastolic target blood pressure < 80, achieved a diastolic blood pressure of 81; the second group, which had a diastolic target < 90 mm Hg, achieved a diastolic blood pressure of 85 mm Hg). Most clinical benefit was seen by lowering pressure to about 140/90 mm Hg. This trial, which involved 18 790 older, hypertensive patients, also showed that approximately 200 such persons would need to be treated for about 4 years with aspirin, 75 mg daily, to prevent 1 additional MI. NNH for the occurrence of 1 additional major or fatal bleeding episode was about 160.
So what is the bottom line for practitioners? Tight blood pressure control is clearly beneficial in patients with diabetes mellitus. Because very low blood pressures were not achieved in either trial, the optimal lower limit in patients with diabetes is not clear. For patients without diabetes, little benefit was achieved from lowering blood pressure below 140/90 mm Hg. Aspirin prevents and precipitates different types of events with no clear-cut overall effects on total mortality.
Some patients with mild-to-moderate carotid stenosis may benefit from endarterectomy
An important report from the North American Symptomatic Carotid Endarterectomy Trial (NASCET) clarified our understanding of which patients benefit from carotid endarterectomy. Previous studies have shown that patients with symptomatic severe stenosis (70% to 99%) had improved outcomes with endarterectomy. This randomized trial evaluated 5-year outcomes in 2267 patients (median age 66 y, 70% men) with nondisabling stroke or transient ischemic attack and < 70% stenosis (11). Participating surgeons were selected for low surgical complication rates. Patients with 50% to 69% stenosis who had surgery had decreased rates of ipsilateral stroke (15.7% vs 22.2%, NNT 15) and any stroke or death (43.2% vs 33.3%, NNT 10). However, the NNT to prevent 1 additional ipsilateral stroke during a 5-year period was twice that for patients with ≥ 70% stenosis and even higher for women than for men. For patients with < 50% stenosis, the groups did not differ for any outcome.
Patients with symptomatic, 50%-to-69% carotid stenosis may benefit when surgery is done in centers with highly experienced surgeons. Because the benefit of endarterectomy for moderate stenosis is eliminated when perioperative mortality or disabling stroke exceeds 2%, local risks and benefits must be considered carefully before recommending endarterectomy.
β-Blockers for congestive heart failure: the pendulum swings
2 systematic reviews examined the benefit of β-blockers in patients with congestive heart failure (CHF) who are already treated with diuretics, angiotensin-converting enzyme (ACE) inhibitors, and digoxin (12, 13). Both reviews appraised 18 small, randomized, controlled trials, representing about 3000 patients (most with ischemic disease or dilated cardiomyopathy), in which any of several β-blockers or placebo was added very slowly to standard treatment of compensated CHF. Both found that β-blockers reduced mortality (NNT 40), hospitalization for CHF (NNT 24), and proportions of persons with functional status deterioration (NNT 25). Although robust overall, the available evidence does not yet fully define which patients benefit most or which β-blockers work best. Most of the studies were of metoprolol or carvedilol, and carvedilol is the only β-blocker currently approved by the U.S. Food and Drug Administration for compensated CHF.
After many years of contradicting theoretical rationale (benefit of decreased sympathetic tone compared with the risk for further myocardial depression), evidence is pushing the pendulum toward a new role for β-blockers in CHF. For patients with stable or compensated heart failure who are receiving diuretics and ACE inhibitors, it is time to consider cautiously adding β-blockers.