Validation of Parkinson's Disease Diagnoses in a National Register: Accuracy, Limitations, and Utility

Abstract

<p><b>Purpose:</b> To evaluate the diagnostic validity of PD identification in a national health registry requiring neurologist-confirmed diagnoses.<br><b>Patients and Methods:</b> We analyzed the Turku PD Cohort including 1626 patients diagnosed with PD between 2006 and 2020 in specialist care. Patients were identified based on ≥ 2 entries of ICD-10 code G20. Diagnoses were re-evaluated through detailed chart review after a median follow-up of approximately 10 years. Two movement disorder specialists independently classified final diagnoses based on clinical records, imaging, and treatment data. Linkage to the Finnish Register of the Entitlements to Reimbursement of Pharmaceutical Expenses identified patients granted reimbursement for antiparkinsonian medications under ICD-10 code G20. Sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) were calculated using the final clinical diagnosis as the reference.<br><b>Results:</b> Of the 1626 patients, 1550 (95.3%) had received reimbursement for PD medications. After long-term follow-up, 1314 were confirmed as PD and 236 reclassified to alternative diagnoses. Registry identification showed high sensitivity (98.2%) and PPV (84.8%). Only 24 confirmed PD cases (1.8%) were not captured. Diagnostic revisions most commonly reflected atypical or secondary parkinsonian syndromes. Specificity was lower (18.1%), reflecting diagnostic evolution during follow-up rather than systematic miscoding. The median delay between diagnosis and registry entry was 24 days.<br><b>Conclusion:</b> A national health registry requiring neurologist-confirmed diagnoses enables highly sensitive PD case identification with good PPV for epidemiological research. However, diagnostic evolution in early parkinsonism limits specificity in cross-sectional register data. Registry studies should therefore apply follow-up exclusion algorithms for alternative parkinsonian diagnoses.<br></p>