A systematic review of spinal research conducted using the National Surgical Quality Improvement Program

Background Context

Over the course of the last decade, interest in the use of large data repositories for clinical research in orthopedic and spine surgery has grown substantially. Detractors maintain that the clinical relevance of research conducted using large registries is limited, and that the academic influence of such studies is minimal. Such contentions have not been empirically evaluated.

Purpose

This study sought to perform a systematic review of spinal research conducted using the National Surgical Quality Improvement Program (NSQIP).

Study Design

This is a systematic review.

Outcome Measures

Impact factor (IF) of the journal of publication and number of citations of published articles conducted using the NSQIP.

Methods

Orthopedic and spine-specific NSQIP articles published from January 1, 2007 to July 31, 2015, were identified through a query of PubMed or Medline, Embase, Web of Science, and Scopus. Articles were classified by journal of publication, year of publication, study topic, study purpose, and method of statistical analysis. Spine surgical publications were compared with other orthopedic research conducted using the NSQIP. The primary dependent variables for the purposes of statistical testing were IF of the journal in which the article was published and the number of citations for each publication. Negative binomial regression was used to evaluate the characteristics of papers associated with increased IF and number of citations.

Results

Of the 1,525 articles identified in the initial search, 114 studies were considered eligible for inclusion in the systematic review. The average IF for the journals publishing orthopedic NSQIP articles was 2.75 (standard deviation SD] 1.22, range 0–5.28), whereas it was 2.52 (SD 0.81, range 1.38–5.28) for spinal research. The average number of citations per article was 6.08 (SD 10.9, range 0–69) and 6.4 (SD 12.0, range 0–69) for spine-specific studies. Following negative binomial regression, only IF (regression coefficients RC] 0.31; 95% confidence intervals CI] 0.08, 0.55) and the year of publication (RC ?1.29; 9% CI ?1.64, ?0.95) were found to have a statistically significant association with number of citations. Among spine-specific research articles, only the year of publication was found to influence the number of citations (RC ?1.29; 95% CI ?1.94, ?0.64).

Conclusions

Our findings indicate that the academic impact of orthopedic and spine surgical research conducted using NSQIP is highly variable, with most publications found to have relatively low impact. As our evaluation of study characteristics associated with high-impact publications and increased citations were unable to uncover factors that are likely translatable, we suggest following research design guidelines that highlight best practices when using large datasets for orthopedic research.