Maternal hematological and coagulation measurements and neonatal outcomes in placental abruption: a two-center retrospective cohort study

Abstract Background Placental abruption can compromise neonatal condition, but the relevance of routinely available maternal hematological and coagulation measurements remains uncertain. We evaluated these associations using ordinal and sparse-event methods. Methods We retrospectively analyzed records from two centers (January 2020 to January 2026). All 135 records met clinical eligibility criteria; the fixed complete-data cohort included 124 pregnancies and 122 live births after two intrauterine fetal deaths were excluded from neonatal analyses. Model-specific sensitivity analyses included up to 129 live births. Suzhou data extraction and related analyses began after ethics approval and waiver on 8 June 2026. Primary proportional-odds models adjusted for maternal age, recorded body mass index (BMI), gestational age, diabetes, hypertensive disorders, and center. Measurement-expanded, Firth, bootstrap, D-dimer, and birth-weight analyses were exploratory, with false-discovery-rate correction where specified. Results Among 122 live births, 15 (12.3%) and five (4.1%) had 1- and 5-minute Apgar scores < 7. Sampling times were unavailable for Nantong. In Suzhou, selected complete blood count (CBC) and coagulation panels were obtained a median of 32.5 and 34.6 days before delivery, and only 10 and 8 were within 14 days. Gestational age was associated with higher 1- and 5-minute scores (odds ratio (OR) per standard deviation (SD), 2.17 [95% confidence interval (CI): 1.46 to 3.21] and 2.59 [95% CI: 1.60 to 4.20]) and was bootstrap-stable. Measurement blocks did not improve gestational-age-adjusted ordinal or birth-weight models, and no individual measurement remained significant after multiplicity correction. The n = 129 clinical sensitivity analysis confirmed the gestational-age associations. Without gestational-age adjustment, the hematology block improved the 5-minute model, but nominal hemoglobin (Hb) and raw-scale D-dimer findings did not survive correction; the D-dimer association disappeared after log transformation. Conclusions Gestational age was the most stable correlate of Apgar scores. No individual maternal measurement showed a robust association after multiplicity correction. Because sampling times were heterogeneous or unavailable, these measurements cannot be interpreted as acute-abruption biomarkers. Prospective studies require standardized sampling and harmonized assays.