Results
Baseline Characteristics
We identified a total of 1836 consecutive chronic dialysis patients who were treated with one of the study exposure drugs (504 beta-blocker, 570 calcium channel blocker and 762 statin-only users, as shown in Figure 1). The largest group of patients excluded from the analysis (11 644 patients), consisted of those not meeting criteria for incident medication use following the initiation of dialysis (this included patients without evidence of two prescriptions for the study medications or those with evidence of use of the study medications prior to the initiation of dialysis—i.e. prevalent medication users). Beta-blocker users were similar to calcium channel blocker and statin-only users on most baseline characteristics (Table 1), with the exception of higher rates of diabetes, hypertension, cerebrovascular and peripheral vascular disease in beta-blocker users than statin-only users. Beta-blocker users also had higher rates of cardiac comorbidities including coronary artery disease, heart failure and atrial fibrillation/flutter. Beta-blocker users and calcium channel blocker users were more likely to receive hemodialysis compared to peritoneal dialysis. With respect to Charlson comorbidity index, calcium channel blocker users were noted to have slightly lower values than the other two groups; however, the groups were similar in terms of number of distinct drugs used in the previous year. The beta-blocker subclasses demonstrated that 89% of patients (449 patients) were receiving low-dose regimens and 6% (29 patients) high-dose therapy. The majority of patients, 83% (417 patients) were found to be receiving highly dialyzable beta-blockers. Similarly, the majority of patients, 89% (447 patients), were using cardioselective beta-blockers. Most, 78% (394 patients), were using a hydrophilic beta-blocker and 7% (37 patients) were prescribed carvedilol. Concurrent statin use was evident in 54% of beta-blocker users (271 patients) and 32% of calcium channel blocker users (183 patients).
(Enlarge Image)
Figure 1.
Study flow diagram. *Patients considered to not have evidence of incident medication use include those without evidence of two prescriptions for the study medications and those patients with evidence of prescriptions in the 180 days prior to the first of the two identifying prescriptions (i.e. prevalent medication users).
We used the period between 365 and 720 days following the index date to assess durability of medication use and crossover. Most patients remained on the study medications, with 74–87% demonstrating a repeat prescription for the index medication. Crossover measures demonstrated that 17% of calcium channel blocker users and 10% of statin-only users initiated a beta-blocker during this period. Among beta-blocker and statin-only users, 14 and 11%, respectively, had evidence of calcium channel blocker use. Lastly, statin use was seen in 49% of beta-blocker users and 31% of calcium channel blocker users during this period.
Outcomes
The mean follow-up time was 2.1 years for beta-blocker users (1034 total person-years), 2.5 years for calcium channel blocker users (1417 total person-years) and 2.2 years for statin users (1658 total person-years). There was a total of 1233 primary outcome events in all three groups, with 360, 473 and 400 events, respectively, in the beta-blocker, calcium channel blocker and statin-only groups, respectively. Compared to statin-only and calcium channel blocker use, beta-blocker use was not associated with a decrease in the primary outcome in unadjusted or adjusted analyses [adjusted hazard ratio (aHR) 1.07 compared to the statin-only group, 95% CI 0.92–1.23, Table 2]. Similarly, there was no association between beta-blocker use and any of the components of the primary outcome (Table 2).
Additional Analyses
The results were consistent across all patient characteristic subgroups (Figure 2). With respect to the beta-blocker characteristics, the only attribute associated with reductions in the primary outcome was beta-blocker dose, with a higher dose associated with better outcomes (aHR 0.50 compared to low dose, 95% CI 0.29–0.88, Table 3).
(Enlarge Image)
Figure 2.
Effect modification by age, year of cohort entry, coronary artery disease, heart failure and dialysis modality. (Beta-blocker users could also be receiving a statin).