Association Between COVID-19 Vaccination and Long COVID Symptoms in Hospitalised Survivors: Distinguishing Prevention from Reverse Causality

Abstract

Background: While COVID-19 vaccination significantly reduces acute disease severity, its impact on the incidence of long COVID remains debated, with some observational studies paradoxically suggesting higher symptom rates among vaccinated individuals. This study aimed to resolve this controversy by distinguishing between the protective effects of prior immunity and the confounding influence of reverse causality. Methods: We conducted a retrospective cohort study of 627 adults hospitalised for COVID-19 in London. Participants were stratified into two analytical cohorts based on vaccination timing: a "prevention cohort" (vaccinated ≥14 days pre-infection) and a "post-acute cohort" (vaccinated post-infection). Multivariable Bayesian logistic regression was employed to estimate Adjusted Odds Ratios (aOR) for long COVID, controlling for age, gender, BMI, comorbidities, and acute length of hospital stay (LoS). Results: In the prevention cohort, prior vaccination demonstrated a non-significant protective trend against long COVID (aOR 0.81; 95% CI 0.45-1.42; p = 0.45), with no significant difference observed between homologous and heterologous regimens. The post-acute cohort exhibited a strong, significant positive association (aOR 3.41; 95% CI 2.23-5.52; p < 0.001), indicating substantial indication bias, with symptomatic individuals more likely to seek vaccination. The strongest independent predictors of long COVID were comorbidities (aOR 2.78) and prolonged acute hospitalisation (≥4 days; aOR 1.82). Conclusions: Vaccination administered prior to infection demonstrates a protective trend against long COVID, whereas the strong association observed with post-infection vaccination reflects indication bias, with symptomatic survivors being more likely to seek immunisation. Clinical strategies to mitigate post-acute sequelae should prioritise reducing acute disease severity and managing comorbidities, which were identified as the dominant independent predictors of risk in hospitalised patients.

Keywords: Bayesian analysis; COVID-19 vaccination; heterologous vaccination; long COVID; post-acute sequelae of SARS-CoV-2 (PASC); reverse causality.

Figure 1

Association Between Vaccination Timing and long COVID. Forest plot displaying Adjusted Odds Ratios (aOR) and 95% Confidence Intervals derived from Bayesian logistic regression models. Dark blue markers (Prevention Cohort): Represent participants vaccinated ≥14 days prior to infection. These estimates show a non-significant trend toward protection (aOR < 1.0). Light Blue markers: Represent exploratory subgroups within the Prevention Cohort, stratified by vaccine regimen (Homologous vs. Heterologous). Both show similar protective trends. Orange marker (Post-Acute Cohort): Represents participants who received their first vaccine dose after the onset of acute infection. This group shows a statistically significant positive association (aOR 3.41), indicative of reverse causality (symptomatic health-seeking behaviour). The vertical dashed line at 1.0 represents no effect. All models were adjusted for age, gender, BMI, comorbidities, and acute length of hospital stay.

Figure 2

Unadjusted Prevalence of long COVID by Vaccination Timing. Bar chart displaying the raw percentage of participants reporting long COVID symptoms in each cohort, who met the clinical criteria for long COVID (symptoms persisting ≥12 weeks post-infection). Unvaccinated (Green): Baseline symptom prevalence was 30%. Vaccinated BEFORE Infection (Blue): Participants vaccinated before infection showed a slightly lower prevalence (28.3%), consistent with the protective trend observed in the adjusted regression models. Vaccinated AFTER Infection (Red): Participants who received their first dose post-infection had a markedly higher symptom prevalence (58.7%). This disparity supports the hypothesis of reverse causality, in which individuals with persistent symptoms were more likely to seek vaccination than those who had fully recovered.