Validation of diagnostic codes for identifying chlamydial infections from TriNetX electronic health record data in...

The accuracy of diagnostic codes for chlamydial infections in electronic health records (EHR) is understudied. TriNetX, a global federated database, provides real-time access to patient-level EHRs, but its suitability for studying chlamydial infections is unknown. This study aimed to validate ICD-10-CM diagnostic codes for chlamydial infections in TriNetX USA data.

A retrospective cohort study was conducted using the TriNetX USA Network from 2016 to 2024. Patients aged 14-44 years with chlamydia diagnosis codes (including C. trachomatis, C. psittaci, and C. pneumoniae) or known Nucleic Acid Amplification Test (NAAT) results were included. All eligible patients were categorized into four mutually exclusive groups based on the presence of diagnosis code (yes or no) and NAAT results (gold standard, positive or negative): true positive, false positive, false negative, and true negative. Positive predictive value (PPV), negative predictive value (NPV), sensitivity, and specificity and their 95% confidence intervals (CI) were calculated for chlamydia diagnosis, including stratified analyses by age, sex, race, ICD codes and calendar year.

Among 4,501,240 eligible patients (75.9% female), chlamydial infection diagnosis codes showed: PPV: 84.3% (95% CI: 84.2-84.5%); NPV: 95.5% (95% CI: 95.5-95.5%); sensitivity: 42.8% (95% CI: 42.6-42.9%); specificity: 99.3% (95% CI: 99.3-99.4%). Stratified analyses revealed PPV: 80.9-87.1%, NPV: 91.7-97.3%, specificity: 99.1-99.9%, and sensitivity: 10.4-49.7% with lowest sensitivity observed in C. trachomatis code analysis.

Most diagnostic codes can accurately identify chlamydial infections in the TriNetX USA Network. However, low sensitivity suggests that many cases could be missed without laboratory test results.