[Prevalence of thyroid abnormality: a comparison of ambulatory claims data with data from a population-based study].

Background: Billing diagnoses are used for quality assurance, estimates of prevalence and resource allocation. Validity studies showed relevant limitations. In Germany, there are no population-based data on the agreement of outpatient billing diagnoses with clinical data of thyroid disorders.
Objectives: The study investigated the agreement of ICD-diagnosed thyroid nodules, goitre, hyperthyroidism, hypothyroidism and thyroiditis with clinical and self-reported data from the population-based cohort study called the Study of Health in Pomerania (SHIP).
Materials and Methods: Billing data from the Association of Statutory Health Insurance Physicians Mecklenburg-Vorpommern were linked on an individual level for the period from 2002-2016 with data from SHIP. The agreement was evaluated using sensitivity, specificity and positive and negative predictive value (PPW, NPW). Data were weighted to ensure population representativeness.
Results: The data of 5746 participants were analysed (46% male, average age 55 years, SD [standard deviation] ± 15, min: 20 years, max: 93 years). Based on clinical data, 63% (3451/5511, missing values n = 235) and based on billing data 25% (1421/5746) of the participants had thyroid disorders. The sensitivity was 12-36%, the specificity was 84-98%, the PPW was highest for thyroid nodules (75%) and hypothyroidism (70%) and the NPW was between 63 and 94%, depending on the investigated thyroid disorder.
Conclusions: Thyroid disorders are common and often undiagnosed. Billing data have a low sensitivity to identify clinically relevant thyroid disorders.