Background: The National Surgical Quality Improvement Program (NSQIP) and National Readmission Database (NRD) are 2 widely used databases that provide valuable information regarding the quality of health care. However, the 2 differ in sampling methodology, which may result in conflicting findings when used for research studies. The objective of this study is to evaluate the differences regarding predictors of 30-day readmissions after anterior cervical discectomy and fusion (ACDF) and posterior lumbar fusion (PLF).
Methods: In this case-control study, NSQIP and NRD were queried for patients undergoing elective ACDF and PLF between 2014 and 2015. The outcome of interest was 30-day readmissions following ACDF and PLF, which were unplanned and related to the index procedure. Subsequently, univariable and multivariable analyses were conducted to determine the predictors of 30-day readmissions using both databases.
Results: For ACDF procedures, diagnosis, outpatient status, American Society of Anesthesiologists class, and length of hospital stay were found to be significant predictors of 30-day readmissions in NSQIP, whereas only age and hypertension were significant in NRD. Among patients undergoing PLF procedures, body mass index, functional status, smoking, steroid use, diabetes, dyspnea, dialysis, emergency, discharge to rehab facility, and length of hospital stay were found to be significant predictors of 30-day readmissions in NSQIP, whereas only alcohol abuse and obesity were significant predictors in NRD.
Conclusions: Two databases differed in terms of significant predictors of 30-day readmissions following ACDF and PLF. This difference may emphasize the differences in the sampling methodology. Further analyses, potentially with an institutional validation, are needed to draw conclusions regarding the accuracy of the 2 databases for predictive analytics.
Keywords: Big data; Databases; NRD NSQIP; Predictive analytics; Readmissions.
Copyright © 2020 Elsevier Inc. All rights reserved.