Background: World Health Organization recently reviewed the possible nonspecific effects of diphtheria-tetanus-pertussis (DTP) vaccine. The results were considered inconsistent though most studies suggested deleterious effects. We examined whether inconsistencies in results reflected differences in effect of DTP or differences in the methodology used in different studies.
Methods: If children remain unvaccinated because they are frail or if children (including dead ones) with no information on vaccination status are classified as "unvaccinated," the mortality rate becomes unnaturally high among "unvaccinated" controls. To measure this bias, we defined the "bias index" as the mortality rate ratio (MRR) between unvaccinated and vaccinated children.
Results: Five studies had frail or poorly defined control groups and survival bias, the bias index being 2.0-8.0; in these studies DTP was associated with an MRR of 0.39 (0.18-0.83). Eight studies determined "unvaccinated" by vaccination card and the bias index was 0.5-1.7; in these studies DTP was associated with an MRR of 2.00 (1.50-2.67).
Conclusion: The observed inconsistencies in results were because of methodologic differences between studies. Bias does not seem to explain why DTP is associated with higher mortality.