Comparison of Definitions of Severe Maternal Morbidity: A National Cohort Study Using Administrative Data
Ian Henderson, Dorothea Geddes‐Barton, Raph Goldacre, Marian KnightABSTRACT
Objective
To understand agreement between definitions of severe maternal morbidity (SMM), the reasons for disagreement and the impact the outcome definition may have on the measurement of ethnic and socioeconomic inequalities.
Design
A national cohort study using administrative data.
Setting
The English National Health Service.
Population
Maternities leading to birth between 1st January 2013 and 31st March 2023 at a gestational age at birth of ≥ 20 weeks.
Methods
The characteristics of maternities were described overall and according to SMM definition. Agreement and chance‐adjusted agreement between definitions were calculated. Poisson regression was used to estimate relative risks (RRs) with 95% confidence intervals (95% CI) for SMM across ethnic and socioeconomic groups.
Main Outcome Measures
SMM definitions were the USA's Centers for Disease Control and Prevention (CDC) without blood transfusion, the Canadian Perinatal Surveillance System (CPSS) definition, and the English Maternal Morbidity Outcome Indicator (EMMOI).
Results
There were 5 939 919 included maternities. Overall, SMM occurred in 1.4% (80 191/5 939 919), 2.7% (158 759/5 939 919) and 1.1% (67 745/5 939 919) of maternities according to the CDC, CPSS and EMMOI definitions, respectively. Chance‐adjusted agreement was K = 0.67 for the CDC and EMMOI; 0.50 for the CPSS and EMMOI; and 0.52 for the CDC and CPSS, with 0.55 between all three definitions. Ethnic and socioeconomic inequalities varied by definition, with RR 1.96 (95% CI 1.90–2.01), 1.71 (1.67–1.74) and 1.98 (1.92–2.04) for the CDC, CPSS and EMMOI definitions, respectively, for Black women compared to White, and RR 1.37 (1.33–1.40), 1.18 (1.16–1.21) and 1.22 (1.19–1.26), respectively, for women living in most vs. least deprived areas. Among maternities with SMM, 55% (44 193/80 191), 41% (64 982/158 759) and 35% (23 905/67 745) had a diagnostic code for sepsis within the CDC, CPSS and EMMOI definitions, respectively. Sepsis was the only component of the SMM indicator for 51%, 44% and 33% of maternities within each of the definitions, respectively. Only 0.3%–0.6% of sepsis codes were associated with a code for shock. The same percentages for acute renal failure overall were 18% (14 159/80 191), 9% (14 842/158 759), and 22% (14 842/67 745), respectively, with acute renal failure representing the only component of the SMM indicator for 13%, 5% and 17%, respectively. Only 1.1%–1.2% of renal failure codes were associated with a procedural code for dialysis.
Conclusions
The prevalence of SMM differed between national definitions and chance‐adjusted agreement was only moderate. The majority of maternities defined as having SMM had only a single diagnostic code for either sepsis or acute renal failure. Only a small minority of maternities with a diagnosis of sepsis or acute renal failure had additional diagnoses or procedures indicative of life‐threatening disease. Validation of the common diagnostic codes is needed to ensure that the measurement of SMM reliably captures and can inform the equitable prevention of severe complications.