Table 1.
Prevalence, in-hospital mortality, population attributable fraction (PAF), and signal-to-noise ratio (SNR) for severe maternal morbidity indicators (SMM), Nationwide Inpatient Sample, 1993–2015 (weighted sample = 87,864,173, unweighted sample = 18,198,934).
Table 2.
Results of the hierarchical population attributable fraction (hPAF) approach for severe maternal morbidity indicator selection, with the selection iteration (k) of each indicator, the cumulative hPAF, and the signal-to-noise ratio (SNR) selection criterion at iteration of selection.
Fig 1.
Prevalence of severe maternal morbidity (SMM) per 10,000 delivery hospitalizations for (1) all 21 SMM indicators, (2) 20 SMM indicators (excluding blood transfusion), (3) blood transfusion alone, and (4) the top 15 ranked SMM indicators that identify the most in-hospital mortality, NIS, 1993–2019.
The blue, dotted line represents the 21 current SMM indicators developed by the Centers for Disease Control and Prevention (CDC). The green, dashed line represents the current indicator for blood transfusion alone. The red, solid line represents 20 SMM indicators (excluding blood transfusion). The purple dashed line represents the top 15 ranked SMM indicators that identify the most in-hospital mortality. Shaded areas represent the 95% confidence intervals for the regression-based SMM prevalence predictions, using 2010–2015 (ICD-9 coding) data, in 2017–2019 (ICD-10 coding). Interval colors correspond to the respective line colors. Data from the last quarter of 2015 and for the year 2016 were omitted because of changes in coding practices during the first year after ICD-10 coding implementation.
Table 3.
Observed and predicted prevalences for years 2017–2019 per 10,000 deliveries of the 15 highest ranked severe maternal morbidity (SMM) indicators for in-hospital mortality, Nationwide Inpatient Sample (weighted sample = 11,001,554, unweighted sample = 2,200,312).