A National Database Analysis Comparing the Nationwide Inpatient Sample and American College of Surgeons National Surgical Quality Improvement Program in Laparoscopic vs Open Colectomies: Inherent Variance May Impact Outcomes.

BACKGROUND: Clinical and administrative databases each have fundamental distinctions and inherent limitations that may impact results.
OBJECTIVE: This study aimed to compare the American College of Surgeons National Surgical Quality Improvement Program and the Nationwide Inpatient Sample, focusing on the similarities, differences, and limitations of both data sets.
DESIGN: All elective open and laparoscopic segmental colectomies from American College of Surgeons National Surgical Quality Improvement Program (2006-2013) and Nationwide Inpatient Sample (2006-2012) were reviewed. International Classification of Diseases, Ninth Revision, Clinical Modification coding identified Nationwide Inpatient Sample cases, and Current Procedural Terminology coding for American College of Surgeons National Surgical Quality Improvement Program. Common demographics and comorbidities were identified, and in-hospital outcomes were evaluated. SETTINGS: A national sample was extracted from population databases. PATIENTS: Data were derived from the Nationwide Inpatient Sample database: 188,326 cases (laparoscopic = 67,245; open = 121,081); and American College of Surgeons National Surgical Quality Improvement Program: 110,666 cases (laparoscopic = 54,191; open = 56,475). MAIN OUTCOME MEASURES: Colectomy data were used as an avenue to compare differences in patient characteristics and outcomes between these 2 data sets.
RESULTS: Laparoscopic colectomy demonstrated superior outcomes compared with open; therefore, results focused on comparing a minimally invasive approach among the data sets. Because of sample size, many variables were statistically different without clinical relevance. Coding discrepancies were demonstrated in the rate of conversion from laparoscopic to open identified in the National Surgical Quality Improvement Program (3%) and Nationwide Inpatient Sample (15%) data sets. The prevalence of nonmorbid obesity and anemia from National Surgical Quality Improvement Program was more than twice that of Nationwide Inpatient Sample. Sepsis was statistically greater in National Surgical Quality Improvement Program, with urinary tract infections and acute kidney injury having a greater frequency in the Nationwide Inpatient Sample cohort. Surgical site infections were higher in National Surgical Quality Improvement Program (30-day) vs Nationwide Inpatient Sample (8.4% vs 2.6%; p < 0.01), albeit less when restricted to infections that occurred before discharge (3.3% vs 2.6%; p < 0.01). LIMITATIONS: This is a retrospective study using population-based data.
CONCLUSION: This analysis of 2 large national databases regarding colectomy outcomes highlights the incidence of previously unrecognized data variability. These discrepancies can impact study results and subsequent conclusions/recommendations. These findings underscore the importance of carefully choosing and understanding the different population-based data sets before designing and when interpreting outcomes research.