The Positive Predictive Value of Hospital Discharge Data for Identifying Severe Maternal Morbidity With and Without Blood Transfusion.

Abstract

Blood transfusion is 1 of the 21 indicators for severe maternal morbidity (SMM) as defined by the Centers for Disease Control and Prevention (CDC) using administrative data. The CDC SMM definition is being prepared to measure hospital quality of care; however, transfusion coding reliability has been questioned. The authors assessed the positive predictive value (PPV) of administrative data for identifying gold standard SMM using the CDC SMM definition, with and without the transfusion indicator. A retrospective cohort study of one hospital's childbirth admissions (2016-2019) was performed. Data were screened for CDC SMM, and subgroups were created for those with transfusion as the sole indicator for SMM (transfusion-only SMM) versus those with at least one other SMM indicator (other SMM). Medical chart review classified CDC SMM cases based on gold standard SMM criteria. Gold standard SMM was defined by validated indicators identified by internal hospital quality reviews and confirmed by expert consensus. The PPV was calculated for all CDC SMM cases and the subgroups. Of 4,212 eligible people, 278 (6.6%) had CDC SMM. Chart review identified 110 gold standard SMM cases among screen-positive cases, yielding an overall PPV of the CDC SMM definition for gold standard SMM of 39.6%. CDC SMM cases identified solely by administrative coding for transfusion were half as likely to meet gold standard criteria, compared to cases identified by other SMM administrative codes (25.9% vs. 49.4%). Blood transfusion, coded as an independent risk factor, had a poor PPV for gold standard SMM. Given efforts to use CDC SMM for quality comparisons, more research is needed to reliably identify cases of SMM without relying on blood transfusion codes.