FIGURE 4.
Observed relative risks obtained with different proportion of exposure captured for subjects with and without risk factor, in absence of a true effect (RRtrue = 1.0). Observed relative risk for differential misclassification to NSAIDs caused by age was 1.18, assuming a relation of RR 2.0 for older subjects compared to younger subjects, 25% captured exposures for younger subjects, and 75% captured exposures for older subjects