1F-03 - Validation of ICD-10 Diagnosis Codes for Antepartum Anemia

Background: Iron deficiency anemia during pregnancy increases risk of adverse perinatal health outcomes, with stark racial disparities. Yet, current evidence is insufficient to support clinical guidelines for pharmacological prevention and treatment, which may include oral or intravenuous iron therapies. . Pharmacoepidemiologic research largely relies on ICD-10 diagnosis codes to identify individuals with antepartum anemia, despite unknown validity.

Objectives: Our objective was to assess the validity of ICD-10 codes for antepartum anemia compared with the diagnostic gold standard of hemoglobin or hematocrit measurements.

Methods: We used commercial insurance claims data from the Merative™ MarketScan® Research Database for pregnancies during 2018-2022. Those with hereditary anemia (e.g., sickle cell anemia, thalassemia) were excluded. Antepartum anemia was defined using clinically recommended hemoglobin and hematocrit thresholds ( < 11.0 g/dL and < 33% in the first and third and < 10.5 g/dL and < 32% in the second trimester, respectively). We calculated Cohen’s kappa, sensitivity, specificity, positive predictive value, and negative predictive value of the ICD-10 codes.

Results: Among 71,160 pregnancies, 25.0% had antepartum anemia based on hemoglobin and hematocrit values and 13.6% had at least one ICD-10 diagnosis code indicating antepartum anemia. Cohen’s kappa between anemia identified through laboratory values and ICD-10 codes was 0.258 (95% confidence interval [CI]: 0.248, 0.268). Sensitivity of the ICD-10 codes was 0.300 (95% CI: 0.294, 0.307); specificity was 0.918 (95% CI: 0.916, 0.921); positive predictive value was 0.551 (95% CI: 0.541, 0.561); and negative predictive value was 0.797 (95% CI: 0.794, 0.801). Sensitivity and negative predictive value of the ICD-10 codes increased with increasing severity of antepartum anemia (e.g., sensitivity was 0.223 (95% CI: 0.216, 0.231), 0.464 (95% CI: 0.451, 0.477), and approximately 0.95 (exact value hidden to prevent back-calculation of small cell sizes) for mild, moderate, and severe antepartum anemia, respectively). Sensitivity of the ICD-10 codes decreased from the first to second trimesters and increased from the second to third trimesters (0.158 (95% CI: 0.144, 0.173), 0.122 (95% CI: 0.115, 0.129), and 0.281 (95% CI: 0.272, 0.290), respectively), while specificity was similar, positive predictive value increased, and negative predictive value decreased across trimesters of pregnancy.

Conclusions: These findings suggest low sensitivity, but high specificity, of ICD-10 diagnosis codes for antepartum anemia. Sensitivity increased with increasing severity of antepartum anemia and was highest in the third trimester. Epidemiologic studies of prevention and treatment therapies should be aware of the limitations of relying solely on ICD-10 codes for antepartum anemia and consider conducting bias analyses when hemoglobin and hematocrit measurements are unavailable.