Introduction: In community-based disease surveillance using networks of volunteer primary care professionals, estimates of the general population may be biased due to a lack of representativeness or an inaccurate characterization of the monitored population, particularly in countries where patient registration with physicians is not mandatory. Methods: In survey methodology, the Horvitz-Thomson estimator allows for the adjustment of estimates through post-stratification. We propose several post-stratified estimates to calculate the incidence of a disease based on the number of cases reported by physicians in the network. The characteristics of the monitored populations are derived from the consultation volumes of participating and non-participating physicians, provided by the health insurance system. The characteristics of the estimators are compared, with an application to data from the Sentinelles network. Results: We show that the number of cases reported by physicians is correlated with their consultation activity. Several post-stratified estimators are defined, taking into account the sampling density of physicians or consultations. Depending on the level of post-stratification, the national incidence varies by up to 3% for influenza, 6% for acute diarrhea, and 11% for chickenpox. At the regional level, the observed differences are more significant, ranging from 40% to over 55%. Post-stratification by consultation volume reduces the variability of estimates across regions. Discussion: By using administrative data on participants and non-participants in a surveillance network, we can reduce the bias in estimated incidence rates, an important step in improving the measurement of population health. (R.A.)