Introduction

The ongoing coronavirus disease 2019 (COVID-19) pandemic caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has posed a serious threat to patients with underlying CKD. These patients are immunocompromised because of a combination of innate and adaptive immune system dysfunction, chronic inflammation, endothelial cell dysfunction, and uremia.1 In fact, infections have been the leading cause of non-cardiovascular morbidity and mortality in the CKD population.2 A recent systematic review highlighted the effect of CKD on the outcomes of COVID-19, including increased risk of hospitalization and mortality.3

With their relatively immunocompromised status, vaccination becomes crucial for patients with kidney disease. However, the ability of COVID-19 vaccines to mount a sufficient immune response in this population remains questionable because of their lower rates of seroconversion, lower antibody titers, and a less sustained response after immunization compared with healthy controls.4,5 Previous studies have also demonstrated that patients on hemodialysis and kidney transplant recipients, compared with healthy controls, have a blunted serologic response after vaccination.6,7

In the Hong Kong Special Administrative Region, China, a territory-wide vaccination program with mRNA (BNT162b2/Comirnaty, BioNTech/Pfizer/Fosun) and inactivated (CoronaVac, Sinovac Biotech HK Limited) vaccines was commenced in February 2021. Although the effectiveness of both vaccines against severe outcomes has been demonstrated in the general public in randomized controlled trials and observational studies, current data on efficacy and effectiveness after three doses of COVID-19 vaccines are limited to those with CKD on dialysis and seroconversion in the non-dialysis population.8–10 Furthermore, with the emergence of the Omicron variant of SARS-CoV-2, vaccine effectiveness seems to be reduced in the general population compared with previous strains of SARS-CoV-2. Therefore, we conducted this population-based, retrospective study to evaluate the vaccine effectiveness during the Omicron BA.2 variant–dominant pandemic wave in the CKD population in Hong Kong.

Methods Study Design and Population

This was a case–control study among patients with CKD aged 18 years or older. Patients with CKD were identified by the International Classification of Diseases, Ninth Revision, as listed in Supplemental Table 1, or eGFR <60 ml/min per 1.73 m2 measured in the previous 12 months with at least two taken ≥90 days apart. Those who contracted COVID-19 infection before the index date, had received the fourth dose of the COVID-19 vaccine, or were with a missing date of birth or sex were excluded from the analysis. The fourth dose was available on March 21, 2022, which is very close to our study end date. We, therefore, excluded patients receiving a fourth dose because of the anticipation of a limited sample size. Data were extracted from the electronic health record (EHR) from the Hong Kong Hospital Authority, which records data on patient demographics, diagnoses, prescriptions, and laboratory tests, in addition to real-time data support and monitoring across all clinics and hospitals in the Hospital Authority for routine clinical management. As a statutory administrative organization in Hong Kong, the Hospital Authority provides all public inpatient services and most public outpatient services. The COVID-19 infection and vaccination records were obtained from the Department of Health of the Government of Hong Kong Special Administrative Region, China, which manages and retains all vaccination records in Hong Kong. Datasets provided by the Department of Health were linked to the Hospital Authority database using unique Hong Kong Identity Card Numbers or other personal identification numbers. These databases have previously been applied in several COVID-19 pharmacovigilance studies.11–18

Definitions of Vaccine Exposure

In the Hong Kong territory-wide vaccination program, the public have a choice of either BNT16b2 or CoronaVac as their first dose, but the second dose has to be the same vaccine. For the booster dose, they had a choice of either a homologous booster or a heterologous booster. Therefore, the COVID-19 vaccination status could be categorized into nine groups on the basis of the types of vaccines and the number of doses administered as follows: (1) unvaccinated, (2) one dose of BNT162b2 only, (3) one dose of CoronaVac only, (4) two doses of BNT162b2 only, (5) two doses of CoronaVac only, (6) three doses of BNT162b2, (7) three doses of CoronaVac, (8) two doses of BNT162b2 followed by a CoronaVac booster, and (9) two doses of CoronaVac followed by a BNT162b2 booster.

Definitions of COVID-19 Infection and Complications

A positive COVID-19 case was defined as a positive PCR test result obtained from the Department of Health and/or Hospital Authority databases. Patients in the control group who reported positive rapid antigen test results in the voluntary reporting platform online were excluded from the analysis.

The outcomes of this study include (1) COVID-19 infection; (2) hospitalization within 28 days of COVID-19 infection; and (3) post-infection all-cause mortality, defined as all-cause mortality within 28 days after COVID-19 infection. The information on all-cause mortality was provided by the Hong Kong Deaths Registry, which officially records all registered deaths of Hong Kong residents.

Matching Method

To assess the effectiveness of BNT162b2 and CoronaVac during the Omicron BA.2 outbreak, the inclusion period of each outcome in this study was from January 1, 2022, to March 31, 2022.19 The matching procedure was applied for each outcome independently. The index date was the date of outcome of interest for the case and the date of hospitalization or attendance at an outpatient clinic for the control. Controls were selected from individuals with CKD who attended any hospital authority service and were not cases. For each case, up to ten matched controls were selected on the basis of age (5-year band), sex, date of attendance (within three calendar days), and Charlson Comorbidity Index (0, 1–2, 3–4, ≥5).20 The Charlson Comorbidity Index is a morbidity score that reflects mortality risk. A score of zero indicates that no comorbidities were found. The higher the score, the more likely the predicted outcome will result in mortality.20

Statistical Analyses

The association between vaccination and each outcome was evaluated using conditional logistic regression, adjusted for chronic comorbidities, including hypertension, cancer, CKD, respiratory disease, coronary heart disease, stroke, heart failure, and dementia, along with the use of chronic medications, including renin-angiotensin system agents, β-blockers, calcium channel blockers, diuretics, nitrates, lipid-lowering agents, oral anticoagulants, antiplatelets, and immunosuppressants. Vaccine effectiveness was calculated using the formula: (1−adjusted odds ratio)×100%. Cochran–Armitage test was performed to assess the association between vaccine effectiveness and number of doses of vaccine.

There were three sensitivity analyses in this study. In the first sensitivity analysis, patients who developed infection <14 days after each dose of vaccine were excluded because the time for full vaccine effect is generally regarded as 14 days.21,22 In the second sensitivity analysis, patients who received their last dose of vaccine more than 180 days before the index date were excluded because waning immunity after vaccination is well recognized after 6 months.23 In the third sensitivity analysis, a COVID-19 case was defined by either a positive PCR test or a positive rapid antigen test result. Subgroup analysis stratified by baseline eGFR level (dialysis, <15, 15–29, 30–44, 45–59 ml/min per 1.73 m2) was conducted.

All statistical tests were two-sided, and a P value of <0.05 was considered statistically significant. Statistical analysis was conducted using R version 4.0.3 (www.R-project.org). At least two investigators (V.K.C. Yan, C.I.Y. Chan, B. Wang, and F.W.T. Cheng) conducted the statistical analyses independently for quality assurance.

Post Hoc Analysis

To examine the effects of modification of dose by vaccine type and baseline eGFR category, we conducted post hoc analyses stratified by vaccine type with the addition of interaction terms using generalized estimating equations.24

Ethical Approval

This study was approved by the Central Institutional Review Board of the Hospital Authority of Hong Kong (CIRB-2021-005-4) and the Department of Health Ethics Committee (LM171/2021).

Results

The selection of cases and controls for COVID-19 infection and complications are shown in Figure 1. A total of 20,570 COVID-19 cases, 6604 cases of COVID-19–related hospitalization, and 2267 all-cause mortality were matched to 81,092, 62,803, and 21,348 controls, respectively.

Figure 1:

Flowchart of selection of cases and controls. COVID-19, coronavirus disease 2019. A&E, accident and emergency; IP, inpatient; CCI, Charlson Comorbidity Index; OP, Outpatient.

The characteristics of cases and controls are summarized in Table 1. For COVID-19 infection, there was no noticeable difference between groups in most of the medical and medication history, except for the use of diuretics (case 6459 [31%]; control: 17,474 [22%]). As for COVID-19–related hospitalization, there was a higher proportion of case patients on diuretics (2805 [42%] versus 14,874 [24%]) and antiplatelets (3069 [46%] versus 23,698 [38%]), with fewer case patients on lipid-lowering agents (4073 [62%] versus 45,123 [72%]) and anti-diabetic drugs (2547 [39%] versus 30,365 [48%]). A similar pattern was observed for all-cause mortality, with more case patients on diuretics (1207 [53%] versus 5471 [26%]), antiplatelets (1188 [52%] versus 8806 [41%]), immunosuppressants (166 [7%] versus 216 [1%]), and insulin (1007 [44%] versus 3171 [15%]) but fewer patients with hypertension (1548 [68%] versus 16,443 [77%]) and diabetes mellitus (1228 [54%] versus 13,602 [64%]) prescribed with renin-angiotensin system agents (1028 [45%] versus 12,286 [58%]), lipid-lowering agents (1308 [58%] versus 14,949 [70%]), and anti-diabetic drugs (788 [35%] versus 9849 [46%]).

Table 1 - Baseline characteristics of cases and controls Baseline Characteristics COVID-19 Infection COVID-19–Related Hospitalization All-Cause Mortality Case Control Case Control Case Control N 20,570 81,092 6604 62,803 2267 21,348 Age, yr 75 (14) 75 (14) 79 (14) 79 (13) 84 (11) 84 (10) Sex, male 12,006 (58) 47,464 (59) 3852 (58) 36,998 (59) 1410 (62) 13,425 (63) Charlson Comorbidity Index 5.01 (2.23) 4.93 (2.13) 5.72 (2.11)a 5.56 (1.98)a 6.53 (2.15)a 6.29 (1.93)a Hypertension 13,348 (65) 54,607 (67) 4345 (66)a 45,532 (72)a 1548 (68)a 16,443 (77)a Cancer 1383 (7) 4314 (5) 511 (8) 4226 (7) 189 (8) 1823 (9) Diabetes mellitus 11,625 (57) 49,404 (61) 3629 (55)a 39,669 (63)a 1228 (54)a 13,602 (64)a Respiratory disease 1280 (6) 3986 (5) 573 (9) 3972 (6) 245 (11) 2039 (10) Coronary heart disease 3487 (17) 11,625 (14) 1373 (21)a 10,379 (17)a 612 (27)a 4117 (19)a Stroke 3250 (16) 11,705 (14) 1371 (21) 11,301 (18) 643 (28)a 4818 (23)a Heart failure 2739 (13)a 7754 (10)a 1295 (20)a 7722 (12)a 572 (25)a 3666 (17)a Dementia 505 (2) 1002 (1) 309 (5)a 1211 (2)a 144 (6)a 628 (3)a eGFR  Dialysis 768 (4)a 87 (0)a 656 (10)a 69 (0)a 121 (5)a 16 (0)a  <15 ml/min per 1.73 m2 2158 (11)a 5756 (7)a 1303 (20)a 4028 (7)a 466 (21)a 1144 (5)a  15–29 ml/min per 1.73 m2 2891 (14) 10,707 (13) 1210 (18)a 8921 (14)a 568 (25)a 3408 (16)a  30–44 ml/min per 1.73 m2 5591 (27) 23,130 (29) 1682 (26)a 19,283 (31)a 575 (25)a 7336 (35)a  45–59 ml/min per 1.73 m2 6143 (30)a 29,274 (36)a 1272 (19)a 22,327 (36)a 388 (17)a 7267 (34)a  ≥60 ml/min per 1.73 m2 2872 (14) 11,744 (15) 467 (7)a 7801 (13)a 166 (7) 2011 (10) Renin-angiotensin system agents 12,111 (59) 50,365 (62) 3471 (53)a 38,018 (61)a 1028 (45)a 12,286 (58)a β-blockers 8541 (42) 32,422 (40) 2880 (44)a 24,258 (39)a 943 (42) 7885 (37) Calcium channel blockers 13,191 (64) 52,879 (65) 4218 (64) 41,333 (66) 1431 (63) 13,968 (65) Diuretics 6459 (31)a 17,474 (22)a 2805 (42)a 14,874 (24)a 1207 (53)a 5471 (26)a Nitrates 3147 (15)a 8597 (11)a 1304 (20)a 7667 (12)a 537 (24)a 3067 (14)a Lipid-lowering agents 13,614 (66)a 57,662 (71)a 4073 (62)a 45,123 (72)a 1308 (58)a 14,939 (70)a Oral anticoagulants 1813 (9) 5909 (7) 705 (11) 5366 (9) 260 (11) 2127 (10) Antiplatelets 7846 (38) 27,935 (34) 3069 (46)a 23,698 (38)a 1188 (52)a 8806 (41)a Immunosuppressants 810 (4) 2866 (4) 225 (3) 1335 (2) 166 (7)a 216 (1)a Insulin 4242 (21)a 12,954 (16)a 1623 (25)a 9951 (16)a 1007 (44)a 3171 (15)a Anti-diabetic drugs 8962 (44)a 39,359 (49)a 2547 (39)a 30,365 (48)a 788 (35)a 9849 (46)a Vaccination status  Unvaccinated 7932 (39)a 23,776 (29)a 3600 (55)a 20,614 (33)a 1552 (68)a 7632 (36)a  One dose of BNT162b2 695 (3) 3508 (4) 216 (3) 2566 (4) 45 (2) 700 (3)  One dose of CoronaVac 3745 (18) 12,869 (16) 1315 (20) 11,093 (18) 392 (17) 4148 (19)  Two doses of BNT162b2 2174 (11) 10,209 (13) 389 (6)a 6854 (11)a 52 (2)a 1877 (9)a  Two doses of CoronaVac 4158 (20) 18,377 (23) 846 (13)a 14,234 (23)a 204 (9)a 4731 (22)a  Three doses of BNT162b2 591 (3)a 4908 (6)a 89 (1)a 2765 (4)a 7 (0.3)a 775 (4)a  Three doses of CoronaVac 1013 (5) 5590 (7) 115 (2)a 3702 (6)a 13 (0.6)a 1226 (6)a  B-B-C 6 (0.0) 57 (0.0) 0 (0.0) 29 (0.0) 0 (0.0) 5 (0.0)  C-C-B 256 (1) 1798 (2.2) 34 (0.5) 946 (2) 2 (0.0)a 254 (1)a Time from vaccination, d 66.0 (70.7)a 58.4 (66.2)a 51.1 (61.3) 54.1 (62.8) 54.3 (63.2)a 47.6 (57.0)a

COVID-19, coronavirus disease 2019.

aIndicate a standardized mean difference >0.1 between cases and controls. All parameters are expressed in either frequency (percentage) or mean (SD).

A positive dose-response relationship of vaccine effectiveness between the number of BNT162b2 or CoronaVac doses received is summarized in Table 2. Vaccine effectiveness among patients with CKD against COVID-19 infection after the first dose of BNT162b2 and CoronaVac were 42% (95% confidence interval [CI], 37% to 47%) and 7% (3% to 11%), respectively. Much higher effectiveness was shown in vaccine recipients who received three doses of BNT162b2 (64% [95% CI, 60% to 67%]) or CoronaVac (42% [95% CI, 38% to 47%]). A similar dose-response relationship was observed, with more doses of vaccine administered associated with higher vaccine effectiveness in preventing hospitalization and all-cause mortality after COVID-19 infection. Vaccine effectiveness against hospitalization and all-cause mortality after COVID-19 infection were 82% (95% CI, 77% to 85%) and 94% (95% CI, 88% to 97%) after three doses of BNT162b2 and 80% (95% CI, 76% to 84%) and 93% (95% CI, 88% to 96%) after three doses of CoronaVac, respectively. Compared with three doses of CoronaVac, patients who received two doses of CoronaVac with BNT162b2 as a booster had higher vaccine effectiveness against COVID-19 infection (57% [95% CI, 50% to 62%]) but had similar vaccine effectiveness against hospitalization and all-cause mortality. Owing to the small number of people who received CoronaVac after two doses of BNT162b2 (n=63), the vaccine effectiveness against different outcomes could not be compared. The adjusted and unadjusted odds ratios of COVID-19 infection and related morbidity are presented in Supplemental Table 2.

Table 2 - Vaccine effectiveness for preventing coronavirus disease 2019 infection and related morbidity by vaccine type Outcomes Unvaccinated BNT162b2 CoronaVac One Dose Only Two Doses Only Three Doses P Value for Trend One Dose Only Two Doses Only Three Doses P Value for Trend COVID-19 infection  Case (%) 7932 (36) 695 (3) 2174 (11) 591 (3) <0.001 3745 (18) 4158 (20) 1013 (5) <0.001  Control (%) 23,776 (29) 3508 (4) 10,209 (13) 4908 (6) 12,869 (16) 18,377 (23) 5590 (7)  Vaccine effectiveness % REF 42 35 64 7 29 42  (95% CI) (37 to 47) (31 to 39) (60 to 67) (3 to 11) (26 to 32) (38 to 47) COVID-19–related hospitalization  Case (%) 3600 (55) 216 (3) 389 (6) 89 (2) <0.001 1315 (20) 846 (13) 115 (2) <0.001  Control (%) 20,614 (33) 2566 (4) 6854 (11) 2765 (6) 11,093 (18) 14,234 (23) 3702 (6)  Vaccine effectiveness % REF 54 66 82 25 63 80  (95% CI) (46 to 60) (62 to 80) (77 to 85) (19 to 30) (60 to 66) (76 to 84) All-cause mortality  Case (%) 1552 (68) 45 (2) 52 (2) 7 (0) <0.001 392 (17) 204 (9) 13 (1)