Confounding by indication is common in observational studies of outcomes that treatment is intended to affect. In light of the stepped-care approach to hypertension management, we reexamined the controversy around myocardial infarction (MI) risk in relation to antihypertensive agents by considering past drug history both as a confounder and as an effect modifier.
Case–control design nested within a cohort of 19 501 adults initiating therapy with angiotensin-converting enzyme inhibitors (ACEI), calcium channel blockers (CCB) or β-blockers in Saskatchewan (1990–93) and followed up to 1997. MI cases were identified using death certificates and hospital discharge diagnoses (ICD-9 410). Four controls were matched to each case to account for duration and timing of follow-up.
812 MI cases were identified, of which 26% were fatal. At first, current use of CCB and ACEI (versus β-blockers) appeared to be associated with an increased risk of MI (RR = 2.2; 95% CI = 1.8–2.7 and RR = 1.3; CI = 1.0–1.6 respectively). Adjustment for drug use history attenuated both associations (RR = 1.6; CI = 1.1–2.2 and RR = 1.0; CI = 0.7–1.4). Moreover, the risk for CCB use disappeared when restricted to patients who had already used these agents in the past (RR = 1.1; CI = 0.77–1.7) whereas a high risk of MI for ACEI was found in digoxin users (RR = 9.4; CI = 3.2–27.5).
Past drug history can be both a confounder and an effect modifier in observational studies. We found adjustment for medication history to attenuate the associations between antihypertensive agents and MI risk. In addition, the estimates significantly varied across drug history profiles thus suggesting the presence of preferential prescribing of specific drug classes to high-risk patients. Copyright © 2001 John Wiley & Sons, Ltd.