Pneumococcal Polysaccharide Vaccine May Not Prevent Hospitalization for Pneumonia in Elderly Individuals

To the Editor—The recently published cohort study by Vila-Córcoles et al. [1] adds valuable data that inform the ongoing controversy regarding the effectiveness of 23-valent pneumococcal polysaccharide vaccine (23vPPV) against hospitalization for pneumonia among the elderly population. However, confounding may have biased the estimate of effect in favor of protection from vaccination. This study presents unadjusted estimates for vaccine effectiveness that indicate no benefit against hospitalization for pneumonia (vaccine effectiveness, 12%; 95% CI, 39% to 19%). With adjustment for age and sex, the estimate increased (vaccine effectiveness, 4%; 95% CI, 19% to 22%). When additionally adjusted for factors suggesting sickness (i.e., outpatient visits, hospitalization for pneumonia, chronic lung disease, chronic cardiomyopathy, and immunoincompetence), the point estimate became much larger, statistically significant, and in favor of vaccination (vaccine effectiveness, 26%; 95% CI, 8%–41%). This model instability suggests possible confounding by indication associated with a “healthy vaccinee” effect, in which healthier subjects were more likely to have received vaccination. Residual confounding is possible, given that only a small number of factors were adjusted for in the model. It would be useful to know the stability of coefficients for other variables in the model when each factor was added or removed. In addition, the authors discuss influenza vaccination as a confounder in some models. Yet this is not adjusted for in the model for hospitalization with pneumonia. Persons who were vaccinated with 23vPPV were more likely to be vaccinated with influenza vaccine. In the discussion section, the authors refer to supplementary analyses restricted to the influenza season. These did not adjust for influenza vaccination status and revealed an even greater effect for 23vPPV, with 95% CIs that almost included no effect (vaccine effectiveness, 31%; 95% CI, 2%–80%). Vila-Córcoles et al. [1] describe considerable uncertainty around their point estimates for vaccine effectiveness. Given the potential for confounding, their conclusions that “23-valent PPV should be recommended for all subjects aged 65 years” [1, p. 867] on the basis of its ability to prevent pneumonia may be overstated.