OBJECTIVES: This study assesses the impact of input precision on predicting serotype-specific vaccine effectiveness (VE) of a pneumococcal conjugate vaccine (PCV) using a method developed by Siber et al. [1], which established correlating pneumococcal anticapsular antibody responses of infants immunized with the 7-valent PCV (PCV7) with clinical protection from invasive pneumococcal disease (IPD).

METHODS: A 20-valent PCV (PCV20) was chosen for this study. The predicted VE by serotype for PCV20 compared to PCV7/PCV13 post-dose 2 of the 2+1 regimen was investigated, aiming to understand the strengths and limitations of the method. Input precision in placebo geometric mean concentration (GMC) and observed PCV7/13 VE values with rounding discrepancies were examined. To evaluate the influence of VE and placebo input precision, additional decimal places were added to the upper and lower bounds of observed serotype-specific VE and upper bound of placebo GMC values, ensuring rounding to two decimal places matched original reported values. Predicted PCV20 serotype-specific VE values were then calculated without altering other parameters.

RESULTS: Minor changes in observed PCV7/13 VE inputs had minimal effect on predicted PCV20 VE values. In contrast, precision of placebo GMC inputs significantly influenced predicted VE values. For example, adjusting the upper bound of placebo GMC for serotype 23F from 0.09 across range of values, including 0.08555 and 0.09444, resulted in predicted PCV20 VE values where the smallest lower 95% confidence interval bound was below -230%, and highest upper bound was above 76%, demonstrating high variability in estimated VE.

CONCLUSIONS: Statistically sound VE estimates and confidence intervals using this method requires comprehensive sensitivity analyses, particularly regarding input precision of placebo GMC values. Minor input changes can lead to significant variations in numerical scale outcomes, emphasizing the need for precise input data and careful interpretation of VE estimates.