Abstract
Cesarean delivery is used when there are problems with the placenta or umbilical cord, for twin pregnancies, and breech births. However, research has found that Cesarean delivery increases the risk of maternal complications like blood transfusions and admission to the intensive care unit. Here, using an instrumental variables study design to reduce bias from unobserved confounders, we study whether Cesarean delivery increases the risk of maternal complications. We use a variant of matching—near-far matching—to render our study design more plausible. In a near-far match the investigator seeks to strengthen the effect of the instrument on the exposure while balancing observable characteristics between groups of subjects with low and high values of the instrument. Extant near-far matching methods are computationally intensive for large data sets, and computing time can be very lengthy. To reduce the computational complexity of near-far matching in large observational studies, we apply an iterative form of Glover’s algorithm for a doubly convex bipartite graph to determine an optimal reverse caliper for the instrument which reduces the number of candidate matches and allows for an optimal match in a large but much sparser graph. We also incorporate a variety of balance constraints, including exact matching, fine and near-fine balance, and covariate balance prioritization. We illustrate this new matching method using medical claims data from Pennsylvania, New York, and Florida. In our application we match on physician’s preferences for delivery via Cesarean section which is the instrument in our study. We compare the computing time from our match to extant methods, and we find that we can reduce the computational time required for the match by more than 11 hours. If our matched sample came from a paired randomized experiment, we could conclude that Cesarean delivery elevates the risk of maternal complications and increases the time spent in the hospital. Sensitivity analysis shows that the estimates for complications could be the result of a minor amount of confounding due to an unobserved covariate. The effects on the length of stay outcome, however, are more insensitive to hidden confounders.
Acknowledgments
The dataset used for this study was purchased with a grant from the Society of American Gastrointestinal and Endoscopic Surgeons. Although the AMA Physician Masterfile data is the source of the raw physician data, the tables and tabulations were prepared by the authors and do not reflect the work of the AMA. The Pennsylvania Health Cost Containment Council (PHC4) is an independent state agency responsible for addressing the problems of escalating health costs, ensuring the quality of health care, and increasing access to health care for all citizens. While PHC4 has provided data for this study, PHC4 specifically disclaims responsibility for any analyses, interpretations, or conclusions. Some of the data used to produce this publication was purchased from or provided by the New York State Department of Health (NYSDOH) Statewide Planning and Research Cooperative System (SPARCS). However, the conclusions derived, and views expressed herein are those of the author(s) and do not reflect the conclusions or views of NYSDOH. NYSDOH, its employees, officers, and agents make no representation, warranty, or guarantee as to the accuracy, completeness, currency, or suitability of the information provided here. This publication was derived, in part, from a limited data set supplied by the Florida Agency for Health Care Administration (AHCA) which specifically disclaims responsibility for any analysis, interpretations, or conclusions that may be created as a result of the limited data set. The authors declare no conflicts.
For comments and suggestions, we thank Paul Rosenbaum.
Citation
Ruoqi Yu. Rachel Kelz. Scott Lorch. Luke J. Keele. "The risk of maternal complications after cesarean delivery: Near-far matching for instrumental variables study designs with large observational datasets." Ann. Appl. Stat. 17 (2) 1701 - 1721, June 2023. https://doi.org/10.1214/22-AOAS1691
Information
