BACKGROUND: Molecular testing in NSCLC is essential for treatment selection, yet routine implementation remains inconsistent across institutions. Clinical evidence suggests that variability in testing may not be explained by patient or tumor characteristics but might be driven by institutional factors, potentially leading to adverse outcomes. We examined the extent to which variability in AGA testing is attributable to the treating institution. METHODS: We analyzed 6437 adults with stage IIIB/C or IV NSCLC enrolled in the prospective German real-world registry CRISP (2016-2022). Logistic mixed-effects models with AGA testing as the primary outcome were used to determine institutional variability across 171 institutions. Models included patient, tumor, and treatment-related fixed effects with institutions as random effects. Intraclass correlations (ICC) quantified institutional variability unexplained by other covariates. Institution type was tested in secondary analysis, and overall survival in exploratory analysis. FINDINGS: AGA testing was performed in 77.9 % of patients (n = 5016). Predicted probabilities for testing use ranged from 30.5 % to 93.2 % across institutions. Institutions significantly influenced testing use (p < 0.001), accounting for 21.4 % of the total variance. Variability significantly differed by institution type and was more pronounced in subgroups, e.g., squamous histology (ICC 29.5 %) and KRAS testing (ICC 34.4 %). Absence of AGA testing was independently associated with inferior survival (HRadj 1.11, 95 % CI 1.01-1.23, p = 0.029). INTERPRETATION: Substantial institutional variability exists in AGA testing for NSCLC, which was unexplained by patient or tumor characteristics. This objective evaluation of institution-based variability in testing may emphasize the importance of practice patterns on patient care and may therefore provide an avenue for change.
Keywords
