COVID-19 Vaccine Effectiveness Among Patients With Maintenance Dialysis; Observations From Population Level Cohort Studies in 2 Large Canadian Provinces

Abstract

Background: It was unknown if the effectiveness of COVID-19 vaccines could vary between regions. Objective: To explore key differences in COVID-19 pandemics in British Columbia (BC) and Ontario (ON) and to investigate if the vaccine effectiveness (VE) among maintenance dialysis population could vary between these 2 provinces. Study Design: Retrospective cohort. Setting and Patients: This retrospective cohort study included patients from population-level registry in BC who were on maintenance dialysis from December 14, 2020, to December 31, 2021. The COVID-19 VE among BC patients were compared to the previously published VE among similar patient population in ON. Two-sample t-test for unpaired data were used to investigate if the VE estimates from BC and ON were statistically significantly different. Exposure: Exposure to COVID-19 vaccines (BNT162b2, ChAdOx1nCoV-19, mRNA-1273) was modeled in a time-dependent fashion. Outcome: Reverse transcription polymerase chain reaction (RT-PCR) confirmed COVID-19 infection and related severe outcome defined by hospitalization or death. Analytical Approach: Time-dependent Cox regression analysis. Results: This study using BC data included 4284 patients. Median age was 70 years and 61% was male. Median follow-up time was 382 days. 164 patients developed COVID-19 infection. The ON study by Oliver et al included 13 759 patients with a mean age of 68 years. 61% of the study sample was male. Median follow-up time for patients in the ON study was 102 days. A total of 663 patients developed COVID-19 infection. During the overlapped study periods, BC had 1 pandemic wave compared to 2 in Ontario with substantially higher infection rates. Vaccination timing and roll out among the study population were substantially different. Median time between first and second dose was 77 days (interquartile range [IQR] 66-91) in BC compared to 39 days (IQR = 28-56) in Ontario. Distribution of COVID-19 variants during the study period appeared to be similar. In BC, compared to pre-vaccination person-time, risk of developing COVID-19 infection was 64% (aHR [95% CI] 0.36 [0.21, 0.63]), 80% (0.20 [0.12, 0.35]) and 87% (0.13 [0.06, 0.29]) less when exposed to 1 dose, 2 doses, and 3 doses, respectively. In contrast, risk reduction among Ontario patients was 41% (0.59 [0.46, 0.76]) and 69% (0.31 [0.22, 0.42]) for 1 dose and 2 doses, respectively (patients did not receive the third dose by study end date of June 30, 2021). VE against COVID-19 infection in BC and ON was not statistically significantly different, the P values for exposure to 1 dose and 2 doses comparisons were 0.103 and 0.163, respectively. Similarly, in BC, risk of developing COVID-19-related hospitalization or death were 54% (0.46 [0.24, 0.90]), 75% (0.25 [0.13, 0.48]) and 86% (0.14 [0.06, 0.34]) less for 1 dose, 2 doses, and 3 doses, respectively. Interestingly, exposure to second dose appeared to provide better protection against severe outcomes in Ontario versus BC, risk reduction was 83% (aHR = 0.17, 95% CI [0.10, 0.30]) and 75% (aHR = 0.25, 95% CI [0.13, 0.48]), respectively. However, the adjusted hazard ratios were not statistically significantly different between BC and ON, the P values were 0.676 and 0.369 for exposure to 1 dose and 2 doses, respectively. Limitations: Infection rate, variant distribution, and vaccination strategies were compared using publicly available data. VE estimates were compared from 2 independent cohort studies from 2 provinces without patient-level data sharing. Conclusions: Health Canada approved COVID-19 vaccines were highly effective among patients with maintenance dialysis from BC and ON. Although there appeared to be between province differences in pandemic waves and vaccination strategies, the VE against COVID-19 infection as well as related severe outcome appeared to be not statistically significantly different. A nationally representative VE could be estimated using pooled data from multiple regions.