What do negative associations between potential risk factors and illness in analytical epidemiological studies of infectious disease really mean?

Abstract

In epidemiological studies of infectious diseases it is not unusual to find that some potential risk factors are negatively associated with risk of illness. The mechanisms generating these associations are unclear in many cases, though one explanation is immunity due to prior exposure. We derive mathematical models for the proportion of a population who are infected with a disease and the proportion who are susceptible in any year of life when individuals are at risk of exposure through more than one route. It is shown that risk of illness declines with increasing age and that this risk declines most rapidly in those groups at increased exposure. In high exposure groups, the relative risk of illness, compared to a group with lower exposure, also declines with age, eventually becoming less than one. The threshold age at which the relative risk is less than 1, i.e., factor B becomes protective decreases with higher exposure rates. Epidemiological studies may substantially underestimate the importance of risk factors where exposure is consistent over many years.