OBJECTIVES: Real-world evidence studies on pregnancy often rely on insurance claims data, yet the accuracy of ICD-10 diagnostic codes for estimating gestational age (GA) remains unclear. Estimating GA is crucial for accurate categorization of pregnancy outcomes, like miscarriage and stillbirth. This study investigated concordance between Z3A diagnostic codes and physician-reported GA using a linked dataset of clinical and insurance claims data.

METHODS: We identified eligible patients in the Myriad Genetics-Komodo Health (MGKH) data, a linked database containing patients who received prenatal cell-free DNA (pcfDNA) screening from Myriad Genetics, Inc. from September 2015 through January 2023 and had claims in Komodo Healthcare Map™. We restricted to patients with continuous enrollment in closed claims for the estimated duration of pregnancy. We described usage patterns for Z3A codes, which specify week of pregnancy, in this real-world pregnancy cohort. We estimated mean differences and the Lin’s concordance correlation coefficient between Z3A codes and physician-reported GA from the pcfDNA intake form.

RESULTS: Of 202,565 pregnant patients eligible for analysis, 174,406 (86.1%) had at least 1 Z3A code that indicated GA. The mean difference between code-estimated GA and physician-reported GA was –0.29 weeks (SD: 0.2). GA from both sources were highly concordant (Lin’s correlation coefficient: 0.98; 95% CI: 0.98, 0.98). Among claims with a Z3A code, 93.7% were within +/-1 week of the physician-reported GA, and 96.5% were within +/-2 weeks. Codes indicating GA of 36-39 weeks had the strongest agreement with physician-reported GA: ≥95.0% of these 4 codes were within +/- 1 week of physician-reported GA.

CONCLUSIONS: Utilizing a uniquely linked database of pregnant patients, we observed strong concordance between Z3A diagnostic codes that indicate GA and physician-reported GA. These results indicate that studies utilizing claims data for pregnancy studies can rely on Z3A codes for reasonable GA estimates.