Renin–Angiotensin System Blocker Discontinuation and Adverse Outcomes in Chronic Kidney Disease

Carl P. Walther; Wolfgang C. Winkelmayer; Peter A. Richardson; Salim S. Virani; Sankar D. Navaneethan


Nephrol Dial Transplant. 2021;36(10):1893-1899. 

In This Article

Materials and Methods

Design Overview

We designed a retrospective cohort study of patients with non-dialysis-dependent CKD receiving care through the VA health system, with ACEI/ARB discontinuation as the exposure and incident ESKD and death as outcomes of interest. To improve comparability, we limited the cohort to people who had initiated ACEI/ARB therapy after incident CKD (new user design) and who used statin therapy during follow-up. Adjustment for discontinuation of statin therapy—chosen as a commonly used, non-ACEI/ARB medication class—was done to reduce healthy user bias, whereby patients who continue to use a therapy may be generally behaviorally different and/or healthier.

Study Population

We identified patients cared for in the VA health system who had a sustained estimated glomerular filtration rate (eGFR) <60 mL/min/1.73 m2 for >90 days (with measurements ≤410 days apart) from 1 January 2005 to 31 December 2015 estimated using the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) serum creatinine–based equation.[12,13] Incident CKD date was taken as the date of sustained reduced eGFR.[14] We defined new ACEI/ARB use as the date of the first pharmacy fill, with no previous ACEI/ARB dispensations in the VA system (for up to 5 years dating back to 2000), to identify patients with CKD who were likely to be incident ('new') ACEI/ARB users. In an attempt to exclude persons who were not true incident ACEI/ARB users, but rather had switched pharmacies to the VA system, we excluded persons who had not filled other medications through the VA within 1 year prior to the incident ACEI/ARB dispensation. We also required the use of a commonly used medication class (statin) during the period of ACEI/ARB use, so that discontinuation of that medication could be adjusted for in analyses to try to reduce confounding related to general medication discontinuation and to isolate the associations with ACEI/ARB discontinuation itself. The study was approved by the Institutional Review Board of the Baylor College of Medicine (H-38697).

Data Sources

Demographic characteristics were obtained from the VA Corporate Data Warehouse. We identified comorbid conditions based on International Classification of Diseases, Ninth and Tenth Revisions, Clinical Modification (ICD-9-CM and ICD-10-CM) diagnostic codes from the VA Inpatient and Outpatient Medical Statistical Analysis System Datasets, classified using the Deyo–Charlson comorbidity index.[15,16] For instance, diabetes was identified using the diagnostic codes 250.0–250.9, E10, E11 and E13 and cardiovascular disease was identified using the diagnostic codes 410.0, 412, 413, 414, 429 and I21.09–21.10. Outpatient laboratory tests were accessed through the Corporate Data Warehouse LabChem data files. We identified pharmacy medication fills from the VA Pharmacy Benefits Management Database.[17]


The outcomes of interest were, separately, death from any cause and ESKD. Dates of death were obtained from the VA Vital Status Files through 31 December 2016. ESKD was identified through linkage to the US Renal Data System database through 31 December 2016. For analysis of each outcome, censoring was performed at the time of the alternate outcome.


The exposure of interest was first cessation of ACEI/ARB therapy. This was defined as nonoverlap of a pharmacy prescription fill of >14 days. We divided discontinuation duration into 14–<90, 90–180 and >180 days. We did not separately categorize permanent discontinuation, as this would introduce survival bias. We investigated the most recent outpatient serum creatinine and potassium levels (measured within the VA system) within the 6 months prior to ACEI/ARB discontinuation, along with hospitalizations within 6 months prior to ACEI/ARB discontinuation.

Kidney Function

CKD was classified by GFR stages into G3a (eGFR 45–59 mL/min/1.73 m2), G3b (eGFR 30–44 mL/min/1.73 m2) and G4 (eGFR 15–29 mL/min/1.73 m2) using outpatient serum creatinine levels and the CKD-EPI creatinine equation.[18]


Demographics were age, sex and race/ethnicity categorized as white, black or other. Index year (year of incident ACEI/ARB use) was categorized as 2005–08, 2009–11 and 2012–15. Body mass index (BMI) was classified into underweight, normal weight, overweight and obese.[19] We included the following comorbidities in the model: diabetes, hypertension, congestive heart failure (CHF), chronic obstructive pulmonary disease (COPD), cerebrovascular disease, peripheral arterial disease, coronary artery disease (CAD), chronic liver disease and malignancy. Outpatient systolic blood pressure (SBP) was categorized as <120, 120–139 or ≥140 mmHg and diastolic blood pressure (DBP) as <80, 80–89 or ≥90 mmHg.

Statistical Analysis

Baseline characteristics of the study population were tabulated. We used time-updated Cox proportional hazards models to evaluate the association of ACEI/ARB discontinuation with each of the primary outcomes separately. We used univariate and sequential multivariate models with covariates chosen based on clinical considerations. In Model 1, we adjusted for age, race and sex; Model 2 added eGFR and comorbidities [hypertension, CHF, COPD, cerebrovascular disease, peripheral vascular disease (PVD), CAD, chronic liver disease and malignancy]; the final model added SBP, DBP, BMI, year of incident ACEI/ARB use and statin discontinuation. ACEI/ARB discontinuation was treated as a time-updated exposure where participants contributed risk time to the no-discontinuation category prior to the first discontinuation event and to one of the discontinuation categories following the first discontinuation event. Only the first discontinuation events were considered in the analyses. We used multiple imputation for missing baseline covariates. Age, CKD stage (based on GFR), SBP and DBP, BMI and comorbidities were time-updated at the time of ACEI/ARB initiation. Analyses were conducted using SAS Enterprise Guide version 7.1 (SAS Institute, Cary, NC, USA;