The proliferation of Big Medical Data has revolutionized observational epidemiology but introduced a significant risk of selection bias. This note proposes a primary validation protocol: the Consistency Delta Test. By comparing the baseline incidence of a control cohort against established national epidemiological benchmarks, researchers can identify fundamental discrepancies before comparative analysis. We demonstrate through three clinical cases, oncology, dermatology, and psychiatry, that when a control group deviates by more than a value of Delta (e.g., 20%) from national standards, the resulting risk estimates lose biological plausibility. This check is proposed as a mandatory prerequisite for ensuring that statistical signals are anchored in epidemiological reality
Marco Roccetti (Sun,) studied this question.