[M2Q or Something else? The Impact of Varying Case Selection Criteria on the Prevalence Estimation of Chronic Diseases Based on Outpatient Diagnoses in German Claims Data].

Background: To determine the prevalence of chronic disease conditions based on outpatient health insurance data, we often rely on repeated occurrence of a diagnosis over the course of a year, usually in two or more quarters (M2Q). It remains unclear whether prevalence estimates change after adapting repeated occurrence of a diagnosis in different quarters of a year compared to a single occurrence or to some other case selection criteria. This study applies different case selection criteria and analyses their impact on the prevalence estimation based on outpatient diagnoses.

Methods: Administrative prevalence for 2019 was estimated for eight chronic conditions based on outpatient physician diagnoses. We applied five case selection criteria: (1) single occurrence, (2) repeated occurrence (including in the same quarter or treatment case), (3) repeated occurrence in at least two different treatment cases (including in the same quarter), (4) occurrence in two quarters and (5) occurrence in two consecutive quarters. Only information on persons with continuous insurance history within the statutory health insurance provider AOK Niedersachsen in 2019 was used (n=2,168,173).

Results: Prevalence estimates differed quite strongly depending on the diagnosis and on age group if a criterion with repeated occurrence of a diagnosis was applied compared to a single occurrence. These differences turned out to be higher among men and younger patients. The application of a repeated occurrence (criterion 2) did not show different results compared to the repeated occurrence in at least two treatment cases (criterion 3) or in two quarters (criterion 4). The application of the strict criterion of two consecutive quarters (criterion 5) resulted in further reduction of the prevalence estimates.

Conclusions: Repeated occurrence is increasingly becoming the standard for diagnosis validation in health insurance claims data. Applying such criteria results partly in a distinct reduction of prevalence estimates. The definition of the study population (e. g., repeated visits to a physician in two consecutive quarters as a mandatory condition) can also strongly influence the prevalence estimates.