Sensitivity of Administrative Coding for Heart Disease in Pregnancy

Purpose: Heart disease complicating pregnancy is a major contributor to maternal morbidity and mortality worldwide, yet its true burden remains underexplored. Electronic medical records may help address this gap. However, the accuracy of the data that can be obtained needs to be established before undertaking estimations. This study evaluates the sensitivity of administrative coding in identifying pregnancies complicated by heart disease during childbirth admissions in a tertiary Australian setting. Patients and Methods: Using a prospectively maintained clinical database of 155 women who received specialized cardio-obstetric care, we compared ICD-10-AM codes recorded in hospital administrative data related to their delivery admission against a cardiologist-adjudicated reference standard. Case and individual code-level comparisons were undertaken. Results: The sensitivity of administrative coding for detecting any cardiac disease was 66.5%. Only 25.2% of cases were completely matched to the clinical diagnosis, and 41.3% mismatched (partial or incorrect). Notably, 33.5% of cases were not coded with any cardiac-related code. At the individual code level, sensitivity varied widely across disease categories: the lowest sensitivity (19%) was shown for arrhythmias. Congenital heart disease, the most common etiology in this cohort, was frequently not coded, particularly when surgically corrected. Conclusion: In addition to incorrect coding, findings reflect the structure of ICD-10-AM, which captures only diagnoses requiring active management during the hospital encounter. As most cardio-obstetric care is delivered antenatally, mainly to prevent peripartum complications, well-managed conditions with uncomplicated delivery may not be coded, leading to underestimation that is independent of the severity of the cardiac condition and associated risk. While exact estimates of coding sensitivity may vary by context, most of the identified issues would be common across Australia and the world. Epidemiological studies and service planning based on ICD-10-AM data should consider enhanced case-ascertainment strategies, including patient-level data linkage and diagnostic look-back periods, to improve accuracy.