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


Patient Characteristics

Among 1 371 075 veterans with CKD we identified 141 252 who met the inclusion criteria (Figure 1) from 2005 to 2015. Among them, 97% were males with a mean age of 73.7 years (SD 10.4) and a mean eGFR of 49.3 mL/min/1.73 m2. Diabetes mellitus was diagnosed prior to ACEI/ARB initiation in 42.5% and CHF in 17.4%. Other patient characteristics at the time of ACEI/ARB initiation are detailed in Table 1.

Figure 1.

Flow chart showing how study participants were selected for this analysis.

ACEI/ARB Discontinuations

We identified 135 346 first discontinuation events, which occurred at a median of 0.52 years [interquartile range (IQR) 0.25–1.49] after ACEI/ARB initiation. About 14.5% of discontinuation events (19 289) were preceded by hospitalization within 6 months (Table 2). Serum potassium level was checked within the 6 months prior to discontinuation in 52 285 discontinuation events (38.6%). Serum potassium was ≥5.0 mEq/L in 5466 (10.5%) and ≥5.5 mEq/L in 1194 (2.3%) (Table 2). About 61% of patients who discontinued ACEIs/ARBs restarted within 6 months of discontinuation. Among them, 46.0% of patients had both creatinine and potassium tested in the first 90 days of ACEI/ARB resumption. An additional 0.8% had only creatinine measured in the same time period and an additional 2.3% had only potassium measured in that 90-day period.


Death. Over 829 095 patient-years of follow-up there were 68 699 deaths and 6152 incident ESKD events. The median follow-up was 4.47 years (IQR 2.20–7.31). ACEI/ARB discontinuation was associated with a more than doubling of the risk of death in the univariate model for all durations of discontinuation studied (Table 3). Adjustment attenuated these associations slightly and in the fully adjusted model the hazard ratio (HR) for death ranged from a low of 1.74 [95% confidence interval (CI) 1.70–1.78] for discontinuations >180 days to a high of 2.30 (95% CI 2.21–2.39) for those who discontinued for 14–30 days. In all the models, the strength of the association with death was higher for shorter durations of discontinuation and the associations were statistically significant. Full model results are shown in Table 3.

ESKD. The pattern of associations between discontinuation and risk of ESKD was similar to the associations of discontinuation with death. In the univariate model, the risk was more than doubled for all durations of discontinuation, with attenuation in the adjusted models, and risk ranged from 1.47 (95% CI 1.26–1.71) to 1.65 (95% CI 1.47–1.85) in the fully adjusted model for different durations of discontinuation (Table 3).