The box plot shows the median hospital readmission rate and surrounding interquartile range (IQR). Whiskers extend 1.5 times the IQR beyond the box. Rates outside of the whiskers are considered outliers (34 of 1228 hospitals).
Customize your JAMA Network experience by selecting one or more topics from the list below.
Clapp MA, Little SE, Zheng J, Kaimal AJ, Robinson JN. Hospital-Level Variation in Postpartum Readmissions. JAMA. 2017;317(20):2128–2129. doi:10.1001/jama.2017.2830
Readmission rates are used as a quality indicator and linked to reimbursement for certain medical and surgical conditions.1 Obstetric maternal readmissions have not been rigorously studied as a quality measure, though their use has been proposed.2 The goal of this study was to determine the potential utility of this metric and its ability to accurately reflect quality by quantifying (1) the variance in hospital postpartum readmission rates and (2) the percentage of the variance that was attributed to the effect of the hospital after controlling for case mix.
The Partners HealthCare institutional review board exempted this study from review. Childbirth hospitalizations were identified using a previously validated method in the 2013 National Readmission Database, which contains discharge data from 14 million hospitalizations from 21 states.3,4 The following hospitalizations were excluded: those in December (per database use guidelines), patients with more than 1 childbirth hospitalization, and those in hospitals performing less than 100 deliveries per year. The primary outcome was any readmission within 30 days of the childbirth hospitalization discharge. Logistic regression models using patient-level data were constructed to calculate risk-adjusted readmission rates for each hospital, accounting for clustering at the level of the hospital. The rates were adjusted for patient age, payer, median income quartile of their zip code, and comorbidities (obesity, hypertensive disorder, pregnancy-related hypertensive disorder, diabetes, asthma, smoking, thyroid disease, psychiatric disease, and seizure disorder), which were identified by International Classification of Disease, Ninth Revision, codes. To understand the amount of variation explained by the hospital, a series of hierarchical random-effects linear regression models were created. Although a binary outcome, this linear approximation allows the within-hospital variability to be estimated from the within-hospital residuals; this method has been previously described.5 The percentage of overall variability attributed to the random effect of the hospital was calculated before and after case-mix adjustment. A sensitivity analysis was performed that accounted for intrapartum events, including mode of delivery and delivery complications (infection, hemorrhage, operative injury, uterine rupture), because these events may be on the causal pathway for readmission. Attributable primary indications for readmission were tabulated to provide clinical context. Analyses were performed using SAS, version 9.4, and JMP 13 software (both from SAS Institute).
Of the 1 664 472 childbirth hospitalizations identified, 1 517 683 from 1228 hospitals met the inclusion criteria. There were 17 508 30-day readmissions. The median unadjusted hospital readmission rate was 1.01% (interquartile range [IQR], 0.67%-1.42%). There was little change after adjustment. The Figure shows the distribution of the adjusted readmission rates. The median was 1.06% (IQR, 0.73%-1.43%). There was little hospital-attributable variance prior to case mix adjustment; 0.11% of the variation was attributed to the hospital (Table). This percentage variance was unchanged after case-mix adjustment and in the sensitivity analysis. The most common primary indications for readmission were hypertension (21.6%), wound infection (13.0%), endometritis (10.2%), hemorrhage (5.8%), urinary tract infection (3.7%), sepsis (3.1%), thrombotic disease (3.1%), mastitis (2.8%), and psychiatric disease (2.5%).
In this study, postpartum readmissions were rare events and attributable to a variety of causes. Fifty percent of hospitals had postpartum readmission rates of 1% or less; in contrast, well-studied medical and surgical readmission rates exceed 20% for some conditions.6 The low frequency of readmissions resulted in rate data that were unstable for analysis, especially for lower volume facilities. Furthermore, of the little variability that existed between the hospital readmission rates, less than 1% of this variation was attributed to the hospital, limiting their use as a quality metric.
This study was limited in that diagnoses were dependent on individual hospital coding practices, and the database lacks the clinical and geographic details to perform more granular analyses. Although a percentage of some indications for readmission may be potentially avoidable, the rarity of events would make studying preventability challenging.
In the search for appropriate metrics, these findings caution against the assumption that postpartum readmission rates accurately reflect obstetrical care quality. The adoption of an insufficient quality metric may negatively affect patient care and reimbursement.
Accepted for Publication: March 2, 2017.
Corresponding Author: Mark A. Clapp, MD, MPH, Department of Obstetrics and Gynecology, Massachusetts General Hospital, 55 Fruit St, Boston, MA 02114 (email@example.com).
Author Contributions: Drs Clapp and Zheng had full access to all of the data in the study, performed the statistical analyses, and take responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: Clapp, Little, Robinson.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Clapp, Little, Robinson.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Clapp, Zheng.
Supervision: Little, Kaimal, Robinson.
Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest and none were reported.
Funding/Support: This study was supported in part by the Department of Obstetrics and Gynecology at Brigham and Women’s Hospital through an intradepartmental research grant, which was used for data acquisition, database processing, and statistical analysis support.
Role of the Funder/Sponsor: The funder was not involved in the design and conduct of the study; the collection, management, analysis, and interpretation of the data; the preparation, review, or approval of the manuscript; or the decision to submit the manuscript for publication.
Create a personal account or sign in to: