Assessment of potential confounding bias in historical new users of ace inhibitors (ACEIS)

Document Type

Conference Proceeding

Publication Date


Publication Title

Pharmacoepidemiology and Drug Safety


Background: As a FDA post-marketing requirement, a cohort study is needed to assess the incidence of angioedema in Black heart failure (HF) patients treated with sacubitril/valsartan using an active control group (defined as naïve to ACEI). A sufficiently large contemporaneous cohort that is naïve to ACEI may be difficult to accrue as many HF patients are treated with an ACEI prior to their HF diagnosis due to other comorbidities. Using historical ACEI control patients, prior to market approval of sacubitril/valsartan, is one solution.

Objectives: To evaluate the potential for bias introduced by historical controls.

Methods: A cohort study was conducted in adult HF patients from 5 sites within the Cardiovascular Research Network. Patients were naïve to ACEI in the year preceding their first diagnosis of HF in the database, but subsequently were dispensed an ACEI. We defined 9 yearly cohorts based on patients' first ACEI dispensing (index date). The first cohort covered Jul 2006 to Jun 2007, the last Jul 2014 to Jun 2015. Standardized differences ([SDif] mean difference/standard deviation) assessed bias in patient characteristics for each cohort as compared to the year preceding sacubitril/valsartan launch (July 2015) with SDif >0.2 indicative of imbalance. Characteristics assessed included the following key factors determined a priori for their relationship with angioedema: age, sex, smoking status, HF severity, diabetes, and allergic reactions, and several non-key factors.

Results: Of the 360,000 HF patients, 37,102 met the inclusion criteria for the yearly cohorts, with 5,909 being Black. None of the key factors showed imbalance at SDif >0.2. For example, among Black patients in the first compared to last cohort, SDif were: 0.04 (age), -0.06 (sex), -0.14 (smoking status), -0.07 (HF severity), -0.03 (diabetes), and 0.05 (allergic reactions). Non-key factors suggested a similar trend with 2 (of 35) characteristics reporting an SDif >0.2 (largest was 0.26).

Conclusions: Results suggest key factors of Black HF patients newly initiating ACEI are balanced over time and indicate historical controls have low potential to bias results of the planned study for measured predictors of angioedema.




Supplement 2

First Page