Challenges of Missing Specific Disease Codes: Estimating Incidence and Prevalence of Immunoglobulin A Nephropathy (IgAN) Using Health Insurance Claims Data

OBJECTIVES: Disease identification usually relies on diagnostic, pharmaceutical and procedural codes. However, coding practices can be inconsistent and specific diagnosis codes are not always available in health insurance datasets. This study presents an approach how to overcome these challenges to generate epidemiological data for IgAN in Germany.

METHODS: A large German claims database was used to estimate the 2022 incidence and prevalence of IgAN using two coding algorithms. The lower bound (LB) included specific ICD‑10‑GM codes (N00.3, N02.3, N06.3) associated with histologically confirmed IgAN. In contrast, the upper bound (UB) expanded to codes potentially used to record IgAN or related symptoms (N00.3, N02.3, N02.5, N02.7, N02.8, N02.9, N06.3, N06.8). Patients with codes for diseases different from IgAN and comorbidities of secondary IgAN were excluded. Incident patients required a documented kidney biopsy around the time of diagnosis.

RESULTS: The study identified a prevalence of IgAN ranging from 5 to 38 per 100,000 individuals. More than 90% of cases identified in both the LB and UB received their IgAN codes during outpatient care. In the UB, a high proportion of codes were nonspecific, with N02.8 and N02.9 accounting for 38% and 49% of cases, respectively. Incidence rates in 2022 varied from 0.19 to 0.64 per 100,000. In the UB, 74% of cases received inpatient codes, and 48% received outpatient codes. The most frequently used ICD-10-GM codes in the UB were N02.8 (61%) and N02.3 (30%).

CONCLUSIONS: Estimating a range of incidence and prevalence helps address the challenges of missing specific diagnostic codes and variability in coding practices. However, more precise coding could significantly enhance the accuracy of epidemiological studies and aid physicians in making informed treatment decisions. Future efforts should focus on developing specific diagnostic codes to reduce uncertainties and improve data reliability.