6ER-002 Validation of ICD-10 diagnosis codes for Guillain-Barré syndrome in Taiwan’s national health insurance claims database

<h3>Background and Importance</h3> Secondary healthcare databases provide extensive population-level health data, making them valuable for both timely disease surveillance and outcome research related to Guillain-Barré Syndrome (GBS). While GBS is often identified in these databases using diagnosis codes, the accuracy and appropriateness of these codes for research purposes remain uncertain. <h3>Aim and Objectives</h3> To validate ICD-10 codes for GBS in Taiwan’s National Health Insurance (NHI) claims database, electronic medical records (EMR) data were used to verify the accuracy of the diagnoses. <h3>Material and Methods</h3> We analysed EMR data from adult patients hospitalised at eight Chang Gung Memorial hospitals in Taiwan between 2017 and 2022. Patients were identified using the ICD-10 code G61.0 listed in any of the five discharge diagnosis positions, indicating a possible diagnosis of GBS. Two clinical doctors independently validated the suspected GBS diagnosis using EMR data of the identified patients, based on the diagnostic criteria established by the National Institute of Neurological Disorders and Stroke. We evaluated the positive predictive values (PPV) of various definitions, including the position of the ICD-10 code in the discharge diagnosis, nerve conduction studies (NCS) claims, and/or specific GBS treatments. <h3>Results</h3> Among 484 patients with the ICD-10 GBS code, 368 were confirmed to have true GBS. The positive predictive value (PPV) for identifying GBS using the code in any discharge diagnosis position was 76.0%. More restrictive criteria, such as using the code in the primary diagnosis position or requiring additional nerve conduction study (NCS) and/or treatment claims, increased the PPV but identified fewer true GBS cases. Specifically, using the ICD-10 GBS code as the primary diagnosis plus NCS and treatment claims achieved the highest PPV of 98.3%, though it missed 140 (38.0%) true GBS patients. Conversely, using the ICD-10 GBS code in any diagnosis position along with NCS claims provided a good PPV of 85.8% with minimal loss of true GBS cases (13, or 3.5%). <h3>Conclusion and Relevance</h3> The PPV of the ICD-10 code for GBS in Taiwan’s NHI claims data were high. For identifying GBS as a vaccine safety outcome, our findings recommend a balanced approach: using the ICD-10 GBS code in any discharge position along with NCS claims in Taiwan’s NHI claims data. <h3>References and/or Acknowledgements</h3> <h3>Conflict of Interest</h3> No conflict of interest