Use of Labor Neuraxial Analgesia for Vaginal Delivery and Severe Maternal Morbidity

Key Points Question Is use of labor neuraxial analgesia for vaginal delivery associated with decreased risk of severe maternal morbidity? Findings In this cross-sectional study of 575 524 women with vaginal delivery in New York hospitals from 2010 to 2017, use of neuraxial analgesia was associated with a 14% decrease in risk of severe maternal morbidity. Meaning Increasing the access to and utilization of labor neuraxial analgesia may help reduce severe maternal morbidity and improve maternal health outcomes.

Missing information on anesthesia care (n = 176,160)

General characteristics
Age (  Variables included in the propensity score to estimate the individual probability of receiving neuraxial analgesia. The propensity score was estimated using a fixed-effect logistic regression with neuraxial analgesia as the dependent variable and the 37 variables listed in the table below as the independent variables. A complete case analysis was performed with 4967 discharges excluded (0.9%).
General characteristics Coding of continuous variables 1 Age As continuous 2 Race (missing = 3206) --3 Insurance type (missing = 3) --Interactions between age, race, and insurance 4 Rural residence (missing = 1761) --5 Comorbidity index for obstetric patients As continuous 6 Obesity --Possible contraindications to neuraxial anesthesia 7 Coagulation factor deficit, Von Willebrand disease, and thrombocytopenia --8 Fever or infection during labor --9 Chorioamnionitis --Labor and delivery characteristics 10 Admission during a weekend --11 Pregnancy resulting from assisted reproductive technology -- The total effect of labor neuraxial analgesia on severe maternal morbidity (SMM) in the presence of post-partum hemorrhage (PPH) is decomposed into 4 components: controlled direct effect, reference interaction effect, mediated interaction effect, and pure indirect effect. The sum of controlled direct effect and reference interaction effect is referred to as the direct effect, and the sum of mediated interaction effect and pure indirect effect as the indirect effect. The controlled direct effect describes the effect of neuraxial analgesia on SMM that is due to neither mediation through nor interaction with PPH. The reference interaction effect describes the effect of neuraxial analgesia that is attributable to interaction with PPH but not mediation through PPH. The mediated interaction effect describes the effect of neuraxial analgesia that is due to both mediation through and interaction with PPH. The pure indirect effect describes the effect of neuraxial analgesia that is attributable to mediation through PPH but not interaction with PPH.
The component effects are estimated through logistic regression modeling. The model for estimating the component effects of the exposure (a) on the outcome (Y) in the presence of a potential moderator (m) with adjustment for other covariates (c) can be expressed as follows: 1| , ,

1| ,
(2) Equation (1) models the effects of the exposure (a), moderator (m) and covariates (c) on the outcome (Y), where denotes the intercept, the exposure effect, θ 2m the moderator effect, θ 3am the interaction effect between the exposure and the moderator, and the effects of covariates. Equation (2) models the effect of the exposure and covariates on the moderator where denotes the intercept, the exposure effect on the moderator and the effects of covariates on the moderator. If the exposure variable and the moderator variable are binary, the total effect of the exposure (a) on the outcome (Y) can be expressed as the sum of four components as follows: Total effect of exposure= , where is the controlled direct effect, which is represented by ; is the reference interaction effect, which is equivalent to • ; is the mediated interaction effect, which is equivalent to • ; and is the pure indirect effect, which is equivalent to • . The NLMIXED procedure in SAS can be used to fit the above hierarchical model and estimate the component effects, and the delta method is used to compute 95% confidence intervals (Chihuri S and Li G "Direct and indirect effects of marijuana use on the risk of fatal 2-vehicle crash initiation.' Inj Epidemiol 2020;7(1): 49).  eTable 4. Comparison of women who did not receive neuraxial analgesia for vaginal delivery and those who did, before and after propensity-score matching (New York State hospitals, 2010-2017