Evaluation of the Congruency of National Surgical Quality Improvement Program Outcomes With a Propensity Matched Anterior Cervical Discectomy and Fusion, Posterior Lumbar Decompression and Fusion Cohort From an Institutional Registry.

STUDY

DESIGN

Retrospective cohort.

OBJECTIVE

To compare institutional and National Surgical Quality Improvement Program (NSQIP) patients who underwent either anterior cervical discectomy and fusion (ACDF) or posterior lumbar decompression and fusion (PLDF) to determine the congruency of baseline characteristics and postoperative outcomes.

BACKGROUND

Large database studies have grown in prominence owing to their inexpensive nature, and ability to draw on large sample sizes. However, concerns regarding the accuracy and limitations of database studies have generated significant debate.

MATERIALS AND

METHODS

A total of 954 patients who underwent ACDF/PLDF at our institution, and 9875 ACDF/PLDF patients from NSQIP were included. The NSQIP cohort was propensity matched to the institutional cohort for single-level ACDF and single-level PLDF procedures. Statistical analysis included either t-tests or Mann-Whitney U tests for continuous variables and χ 2 tests or Fisher exact tests for categorical variables.

RESULTS

The majority of postoperative complication rates did not differ significantly between the matched cohorts. Institution ACDF patients had shorter operative times and length of stay (LOS) compared with the NSQIP cohort. The institutional PLDF cohort had a higher 30-day readmission rate (3.13% vs 0.22%, P = 0.002) compared with the NSQIP PLDF cohort, while no differences were noted in 30-day readmission for ACDF. Institutional 90-day ACDF readmission rates were 2.36%, and institutional PLDF 90-day readmission rates were 7.14%, while NSQIP does not track 90-day readmission rates.

CONCLUSIONS

The current study offers a rigorous comparative analysis of ACDF and PLDF outcomes from NSQIP and a single institution. Although large databases are valuable in understanding trends and rare events, clinicians should be advised of the limitations of database research, such as missing data, selection bias, and coding variations, and their relevance to the unique population encountered within institutions.

LEVEL OF EVIDENCE: Level III.