Persistence in risk and effect of COVID-19 vaccination on long-term health consequences after SARS-CoV-2 infection

The study investigates whether the increased risk of long-term health consequences after SARS-CoV-2 infection persists over time and how COVID-19 vaccination modifies these risks. In the context of extensive global COVID-19 transmission and vaccination campaigns, prior evidence has shown elevated risks of acute and post-acute sequelae affecting multiple organ systems (cardiovascular, neurological, psychiatric, metabolic) and increased mortality following infection, sometimes persisting up to two years. Definitions of post-COVID conditions vary, but generally acute sequelae occur within 30 days and post-acute sequelae after 30 days. The purpose of this study is to quantify the progression of risk over one year post-infection, stratified by vaccination status, and to assess whether complete primary vaccination and booster doses mitigate these risks. The findings aim to inform clinical management and public health strategies regarding long-term outcomes and the protective role of vaccination.

Existing studies report elevated risks of incident cardiovascular disease, neurological and psychiatric disorders, diabetes, and mortality following SARS-CoV-2 infection, with some risks persisting up to two years. There is variability in definitions of post-COVID-19 condition and heterogeneity in methodologies assessing vaccination effects, age, timing since infection, and prior infection history, leading to inconsistent results on vaccine protection against long-term sequelae. Vaccination, including booster doses (e.g., bivalent boosters), has consistently reduced severe disease and mortality, often by over 80% in high-risk populations, and broadened immunity against Omicron sublineages. Evidence on protection against post-acute sequelae is mixed, and mechanisms may involve prevention of severe acute disease, thereby lowering long-term organ damage. This study builds on these findings to provide territory-wide, real-world evidence across multiple post-infection windows and vaccination strata.

Design: Retrospective, territory-wide cohort study in Hong Kong using routinely collected electronic medical records. Data sources: Hong Kong Hospital Authority (HKHA) electronic medical records (inpatient, outpatient, emergency), Hong Kong Deaths Registry for mortality, Department of Health vaccination records, and Centre for Health Protection SARS-CoV-2 case confirmations. Databases were linked via anonymized unique patient identifiers. Population: Individuals with first positive SARS-CoV-2 test between April 1, 2020 and October 31, 2022; non-infected controls without COVID-19 during the study period were also included. Patients aged <18 years and those who died before index date were excluded. Cohort size included 1,175,277 infected patients (stratified by vaccination status: unvaccinated; 1, 2, or ≥3 doses) and 2,649,116 non-COVID-19 controls (after exclusions). Exposure: Vaccination status prior to infection categorized as 0, 1, 2, or ≥3 doses of BNT162b2 (BioNTech) or CoronaVac. Follow-up windows: Outcomes assessed across 0–30, 31–90, 91–180, 181–270, and 271–365 days post-index (infection date for cases; matched index for controls). Outcomes: Clinical diagnoses based on ICD-9-CM, including composite major cardiovascular disease (stroke, myocardial infarction, heart failure, coronary artery disease), coronary artery disease, deep vein thrombosis, chronic pulmonary disease, acute respiratory distress syndrome, end-stage renal disease, acute kidney injury, pancreatitis, cardiovascular mortality, and all-cause mortality. Statistical analysis: Inverse Probability of Treatment Weighting (IPTW) using propensity scores incorporating age, sex, Charlson Comorbidity Index (CCI), medication histories (renin–angiotensin system agents, beta-blockers, calcium channel blockers, diuretics, lipid-lowering agents, immunosuppressants), healthcare utilization, and vaccination status. Weighted Cox proportional hazards models estimated hazard ratios (HRs) with 95% CIs for each time window, comparing infected groups by vaccination status to non-infected controls. Balance was assessed using standardized mean differences (SMD ≤ 0.1 indicating good balance). Sensitivity analyses included extended follow-up and assessments across subgroups (e.g., age, CCI). Ethics and transparency: Ethical approvals obtained (UW210-556, UW21-I49; Department of Health approvals L/M21/2021, LM/175/2022). Data are not publicly shareable due to restrictions; code available on Zenodo (DOI: 10.5281/zenodo.10132693). Reporting followed RECORD/STROBE guidelines.

- Sample: 1,175,277 infected patients: unvaccinated 124,443; 1 dose 101,379; 2 doses 457,896; ≥3 doses 491,559. Controls: 2,649,116. Median time from last vaccine dose to infection: 18 days (IQR 12–42; 1 dose), 175 (170; 2 doses), 101 (37–170; ≥3 doses). - Progressive reduction in risk over time: Compared to controls, the risk of all-cause mortality among infected patients declined across windows over one year (e.g., 271–365 days HR 2.12, 95% CI 1.92–2.33), with the sharpest decline between 31–90 days versus 0–30 days. - Vaccination effect: Patients with complete primary vaccination or booster doses had lower risks of multiple sequelae (including major CVD and all-cause mortality) 30–90 days post-infection than unvaccinated or incompletely vaccinated patients. From 271 days onward (fully vaccinated) and from 91 days onward (booster), there was no significant excess risk of clinical sequelae versus controls, whereas unvaccinated and incompletely vaccinated individuals continued to show elevated risks up to one year. - Specific outcomes: Persistent but diminishing elevated risks were observed for certain outcomes. For example, heart failure risk remained elevated in early post-acute periods (dose-2 HR 1.43 [1.13–1.85]; dose-3 HR 1.35 [1.05–1.75]). All-cause mortality remained modestly elevated in some vaccinated groups in early windows (e.g., dose-2 HR 1.11 [1.02–1.21]). - Subgroups: Patients aged >65 years or with CCI ≥4 had higher risks of acute and post-acute sequelae, particularly if unvaccinated or partially vaccinated. Risks were broadly comparable between males and females after accounting for comorbidities. - Additional observations: Sensitivity analyses supported primary findings. A moderate increase in risk of lung cancer and lymphoma was noted in the first 30 days post-infection. - Overall: There was a dose–response pattern where greater numbers of vaccine doses correlated with lower risk of post-infection sequelae and mortality, and risk attenuated over time.

The study addresses whether the elevated risks of post-COVID clinical sequelae persist over one year and whether COVID-19 vaccination mitigates these risks. Findings show a graded decline in hazards over successive post-infection windows and a clear protective association of complete primary vaccination and booster doses against a broad set of sequelae, including major cardiovascular events and mortality. These results align with the hypothesized mechanism whereby vaccination reduces severity of acute illness and subsequent organ damage, thereby lowering long-term risks. The attenuation of risk was most pronounced among the fully vaccinated and boosted, with no significant excess risk beyond 91 days (booster) or 271 days (fully vaccinated) compared to controls, while unvaccinated individuals continued to have elevated risks up to one year. The study also highlights that high-risk groups—older adults and those with high comorbidity burdens—retain higher susceptibility to adverse outcomes, underscoring the importance of vaccination and vigilant follow-up in these populations. In the context of evolving variants (e.g., Omicron sublineages with milder severity) and improved vaccine coverage, the overall disease burden from long-term sequelae may be shifting downward, though vulnerable groups remain at risk.

In this territory-wide cohort, risks of long-term clinical sequelae after SARS-CoV-2 infection declined progressively over one year. Complete COVID-19 vaccination and booster doses were associated with substantial reductions in the incidence and risk of most sequelae and all-cause mortality beyond the acute phase. Fully vaccinated individuals showed no significant excess risk from 271 days onward, and boosted individuals from 91 days onward, whereas unvaccinated or incompletely vaccinated patients had persistent elevated risks up to one year. These findings provide real-world evidence supporting vaccination as an effective strategy to reduce long-term health consequences after COVID-19. Future research should refine understanding of causal pathways, assess variant-specific and dose-timing effects, and address potential biases (e.g., selection, residual confounding) and unmeasured factors (obesity, smoking, socioeconomic status, education, viral strain) using enriched datasets and alternative designs.

- Detection and surveillance bias: Under-reporting of pre-existing conditions before SARS-CoV-2 diagnosis and increased healthcare contacts around vaccination and infection may inflate identification of conditions post-infection. - Variant–vaccination confounding: Later variants (e.g., Delta/Omicron) coincided with higher vaccination coverage and lower severity, potentially confounding associations despite sensitivity analyses. - Selection bias in period-specific HRs: Censoring after incident sequelae and survivor differences between cohorts can bias later-window HR estimates, possibly attenuating effect sizes. - Residual confounding: Unmeasured factors (e.g., obesity, smoking, socioeconomic status, education, viral strain) were unavailable and may bias results even after IPTW. - Data constraints: Some outcomes had limited events in certain strata/time windows, reducing precision; data sharing restrictions limit external validation using the same source.