Current issues of ACP Journal Club are published in Annals of Internal Medicine


Poor treatment adherence increased risk for death after myocardial infarction in women

ACP J Club. 1994 Mar-April;120:46. doi:10.7326/ACPJC-1994-120-2-046

Source Citation

Gallagher EJ, Viscoli CM, Horwitz RI. The relationship of treatment adherence to the risk of death after myocardial infarction in women. JAMA. 1993 Aug 11;270:742-4.



To determine if treatment adherence is a risk factor for death after a myocardial infarction (MI) in women.


Cohort of women participating in a randomized, double-blind, placebo-controlled trial, the β-Blocker Heart Attack Trial (BHAT).


Multicenter trial in North America.


602 women (age range 30 to 69 y) enrolled in the BHAT. At 5 to 21 days after a MI, patients were randomly assigned to receive propranolol hydrochloride or placebo.

Assessment of risk factors

The cohort was followed up at quarterly intervals for a median of 26 months. At each visit, patients were asked to return all unused medication. Adherence was calculated as the proportion of prescribed medication taken based on recorded pill counts. Overall adherence for each patient was the mean of all adherence estimates. Adherence was dichotomized into good (taking ≥ 75% of prescribed medication) or poor (taking < 75% of prescribed medication). Other potential risk factors documented were severity of MI; history of congestive heart failure (CHF); age; smoking during the trial; marital status; educational level; race; and 2 composite variables, 1 for clinical severity and another for sociodemographic risk.

Main outcome measures

Mortality from all causes.

Main results

Adherence data were available for 505 women (84%). Among these 505 women, 32 deaths (6%) occurred. 6 of 44 poor adherers (14%) died compared with 26 of 461 good adherers (6%) (relative risk [RR] 2.4, 95% CI 1.1 to 5.6). Mortality for patients with a history of CHF was 16% compared with 6% for those with no history of CHF (RR 2.9, CI 1.3 to 6.6). No association was noted between mortality and any of the other variables considered. The effect of treatment adherence on all-cause mortality remained unchanged after adjustment for treatment category (propranolol or placebo), age, severity of MI, CHF, smoking history, marital status, education level, and race (adjusted RR for death for poor adherence 2.5 to 3.0, P ≤ 0.02).


Poor treatment adherence after a myocardial infarction was an independent risk factor for all-cause mortality in women.

Sources of funding: {Robert Wood Johnson Clinical Scholars Program and Yale University School of Medicine.}

For article reprint: Dr. E.J. Gallagher, 1-W-20, Bronx Municipal Hospital, Bronx, NY 10461, USA. FAX 718-918-7459.

*Information supplied by author.


The relation between treatment adherence and clinical prognosis appears straightforward. Patients who consume a sufficient quantity of an effective medication should have better outcomes than those who do not. The reports by Gallagher and colleagues and others (1) describe a more complex relation: High adherence may improve prognosis even if the medication is a placebo.

Recent research suggests that pill counts, the adherence measure in this study and in most clinical trials, are often manipulated by the patient and can overestimate medication consumption (2). Thus, the excellent adherence of participants in this study, 91% of whom seemed to consume ≥ 75% of their medication, may be more apparent than real. In any event, the clinical trial setting limits the generalizability of these findings to the more heterogeneous world of clinical practice. This limitation, however, does not invalidate the methods and conclusions of the study.

The patients with high adherence rates in this study may also have improved unmeasured behaviors, such as successful lifestyle changes, or may have been less troubled by depression (3), which reduced their mortality after MI. Thus, medication adherence is likely part of a constellation of patient attributes that can be powerful determinants of clinical outcomes. The power is evident in the observation that patients highly adherent to taking placebo in this study had a mortality rate similar to that of those poorly adherent to taking propranolol.

The message of this study is that poor medication adherence identifies a high-risk group of patients who require careful assessment and special efforts to modify their behavior. Whether successful adherence counseling can improve prognosis is an important but unresolved question.

John F. Steiner, MD
University of ColoradoDenver, Colorado, USA


1. Influence of adherence to treatment and response of cholesterol on mortality in the Coronary Drug Project. N Engl J Med. 1980;303:1038-41.

2. Rudd P, Ahmed S, Zachary V, Barton C, Bonduelle D. Improved compliance measures: applications in an ambulatory hypertensive drug trial. Clin Pharmacol Ther. 1990;48:676-85.

3. Frasure-Smith N, Lespérance F, Talajic M. Depression following myocardial infarction: impact on 6-month survival. JAMA. 1993;270:1819-25.