Analysis of International Classification of Diseases, Tenth Revision, Coding Practices at Referral Centers and Other...

To evaluate the accuracy of International Classification of Diseases, Tenth Revision (ICD-10) codes for placenta accreta spectrum (PAS) by analyzing frequency and reasons for miscoding, with particular attention to differences between PAS referral centers and other nonreferral obstetric hospitals.

We conducted a retrospective cohort study of all deliveries with ICD-10 codes for PAS in the two large health care systems (17 hospitals, including two PAS referral centers) in one Mountain West state between September 2015 and December 2023. Medical records were evaluated independently by three reviewers to determine whether PAS was present using International Federation of Gynecology and Obstetrics clinical or histopathologic criteria (true PAS). The primary outcome was false-positive diagnosis of PAS. For each false-positive diagnosis, a reason for misapplication of the code was assigned. Composite maternal morbidity (including blood loss of 1,500 mL or more, length of stay more than 4 days, reoperation, shock, coagulopathy, genitourinary injury, thromboembolism, or mortality), as well as coding accuracy, were compared between referral centers and other hospitals.

Among 394 deliveries with PAS codes (0.15% of all deliveries), 39.6% (95% CI, 34.7-44.4%) were classified as false-positive. False-positive rates were significantly higher at nonreferral centers compared with referral centers (73.6% vs 23.8%, P <.001). The most common reason (42.9%) for a false-positive diagnosis was hemorrhage due to another etiology not meeting clinical or histopathologic criteria for PAS. In 22.4% of false-positive diagnoses, there was antenatal suspicion of PAS, but PAS was not confirmed at delivery. All patients with PAS 3 (percreta) and nearly all patients with PAS 2 (increta) were managed at referral centers. After adjusting for maternal age, prior cesarean delivery, gestational age at delivery, in vitro fertilization, and other past uterine procedures, the odds of composite maternal morbidity were higher at PAS centers (adjusted odds ratio [aOR] 2.54; 95% CI, 1.39-4.66), but this difference was not significant when analyzing only true PAS cases (aOR 1.51; 95% CI, 0.63-3.65).

The ICD-10 codes for PAS were misapplied frequently, particularly at nonreferral centers. Among true PAS cases, despite higher disease severity at referral centers, maternal morbidity was not significantly different between center types after adjustment. These findings have important implications for PAS research methodology and quality metrics that rely on administrative data.