Vaccines prevent illness both by decreasing individual risk of acquiring illness and preventing transmission. Clinical trials for testing vaccines have classically focused on disease prevention. with random assignment to individuals. However, other outcome measures and randomization to family or other units may be desirable. We will develop and apply a general methodology for assessing the consequences of vaccine trial design decisions using computer simulations that model a broad range of factors known to influence infection transmission and control. The specific vaccine to be examined is one that is under development for non-typeable Haemophilus influenzae (NTHi). a common cause of otitis media and sinusitis. Specifically, we propose to: (1) determine the power to detect useful effects of trial designs that randomize vaccine and placebo to individuals, family units, or daycare centers; (2) determine how well the effect parameters estimated from these different designs predict the ability of vaccination programs to control infection; (3) determine the sensitivity of power and predictive accuracy of vaccine trial designs to: (a) details in the trial design such as whether clinical otitis or nasopharyngeal culture is used as an outcome; (b) transmission dynamics in the study population; (c) biological aspects of NTHi infection; (d) biological effects of the vaccines; (4) determine whether standard statistical methods erroneously assess the significance and power of vaccine effect measurements. The sensitivity analysis will be performed using recent advances in the area of response surface analysis and experimental design. The models to be simulated represent a significant advance in combining analytical tractability with flexibility to include geographic and social space determinants of contact patterns, biologically defined vaccine effects, and detailed natural histories of infection. The analysis performed will provide the vaccine trial designer with comparisons of power and predictive accuracy that reveal the relative trial sizes, data completeness. and data accuracy that are needed to attain comparable utility from different trial designs. They will also provide insight regarding how different biological effects of vaccines generate different population effects of vaccine programs.
