Validation of Lactational Mastitis Diagnosis Codes in Electronic Health Care Data

Background: Electronic health record data are an underused source for lactation-related research. The validity of the International Classification of Diseases, 10th Revision Clinical Modification (ICD-10-CM)-coded lactational mastitis is unknown. Methods: We assessed lactational mastitis diagnosis code validity by medical record review. We included patients from three health care systems with a live birth between December 2020 and September 2022 whose infant had ≥1 well visit and for whom there was electronic health record documentation of lactation in patient or infant records. We used ICD-10-CM diagnosis codes (N61.0 and O91.2) to identify patients with suspected lactational mastitis and assessed antibiotic dispensings. We performed medical record reviews on a random sample to determine whether suspected lactational mastitis cases met definitions for “probable” (breast symptoms with systemic symptoms) or “possible” (breast symptoms without systemic symptoms) lactational mastitis. We report positive predictive values (PPV) with 95% confidence intervals (CI). Results: Among 19,660 eligible patients, 1,023 (5.2%) had either N61.0 or O91.2 diagnosis code and 768 (3.9%) had a diagnosis code and antibiotic dispensed. Chart reviews of 119 identified PPV of 76% (95% CI: 67.3, 82.9) for probable and 97% (95% CI: 91.6, 98.7) for probable or possible lactational mastitis. Restricting to those dispensed an antibiotic (n = 87), PPVs improved to 80% (95% CI: 69.6, 87.4) for probable and 100% (95% CI: 95.8, 100) for probable or possible lactational mastitis. Conclusions: Diagnosis codes alone have good PPV for lactational mastitis. PPV for lactational mastitis improves when including antibiotic data, although case numbers decrease. Future research may consider the use of ICD-10 codes alone for the identification of lactational mastitis.