The investigators analysed hospitalizations for respiratory infections in England during 2023–2024 among individuals who were vaccine‑eligible. They applied causal mediation analysis to disaggregate how much of the observed ethnic group differences in hospital admission risk could be explained by socioeconomic deprivation, vaccination uptake and comorbidities. The abstract did not include sample sizes, precise statistical models, or case definitions; those methodological specifics were not reported in the source abstract.
Data indicated that, for influenza, hospitalization odds were highest in the Pakistani and Bangladeshi ethnic groups when compared with White British individuals. The study team reported that socioeconomic deprivation explained a measurable portion of these ethnic differences: deprivation mediated 26% of the Pakistani total effect and 41% of the Bangladeshi total effect on influenza hospitalization risk. The authors noted that comorbidities contributed little to those ethnic group differences. The abstract does not state how ethnicity categories were defined or whether additional social determinants—such as household size, occupational exposure or access to primary care—were modelled, and those details were not provided.
Investigators found minimal ethnic group differences in COVID‑19 hospitalizations in the post‑pandemic period covered by the analysis. Where influenza retained clear ethnic disparities, COVID‑19 did not show the same pattern in this vaccine‑eligible cohort, according to the report. The abstract does not provide effect estimates or confidence intervals for COVID‑19 comparisons; those numeric details were not included in the source abstract.
Across both pathogens the study team observed a consistent deprivation gradient: people in the most deprived groups had disproportionately higher hospitalization than those in the least deprived groups. For influenza hospitalizations, the investigators estimated that lower vaccination uptake mediated between 17% and 26% of the effect of deprivation. For COVID‑19, once vaccination uptake and comorbidities were included in the mediation model, the authors reported no meaningful residual deprivation inequalities. The abstract did not specify the vaccination schedules used, how uptake was measured (for example, seasonally or cumulatively), or whether vaccine effectiveness was assumed constant across groups; those measurement details were not reported.
The study team reported that comorbidities made little contribution to ethnic group differences in influenza hospitalization in this cohort. That finding suggests the measured difference by ethnicity was not primarily driven by differential chronic disease burden in the analysed population. The abstract does not list which comorbid conditions were accounted for, how they were ascertained, or whether severity or multimorbidity were modelled as separate mediators; those elements were not described in the abstract.
Authors concluded that both vaccination uptake and deprivation mediated a substantial share of inequalities in hospitalization for influenza and COVID‑19 within the vaccine‑eligible cohort. They recommended that improving vaccination coverage in under‑vaccinated groups should remain a public‑health priority. The report frames vaccination access and uptake as modifiable intermediate factors that can reduce hospitalization disparities, particularly for influenza where mediation percentages were sizable.
The abstract limits interpretation by omitting many operational details: specific sample sizes, numeric estimates and uncertainty bounds for all comparisons, the exact causal mediation approach and assumptions, and the definitions used for vaccination status and comorbidity measurement were not provided. The authors did not report whether sensitivity analyses were performed to test mediation assumptions, nor did the abstract provide subgroup analyses by age strata or geographic region. These unanswered methodological questions restrict the ability to fully appraise internal validity and generalizability based solely on the abstract. Future readers should consult the full article for the analytic code, model specifications and full numerical results referenced by the authors.
End of abstract‑derived report.