Kevin R Kasten, MD1, Konstantinos Spaniolas, MD2. 1Carolinas HealthCare System, 2Stony Brook Medical Center
Introduction: The last decade yielded myriad studies of surgical outcomes using large datasets. Unfortunately, the National Surgical Quality Improvement Program (NSQIP) and Thoracic Morbidity and Mortality System (TMMS) datasets were developed with institutional outcomes in mind. In an attempt to quantify individual surgeon quality, these platforms are increasingly queried but with minimal supporting data. The American College of Surgeons (ACS) Surgeon Specific Registry (SSR) is an online database whereby surgeons conveniently track cases and outcomes for maintenance of certification (MOC). Here, we identify potential shortcomings of outcome measurement using this reporting system and identify methods to overcome them using commercially available statistical software.
Methods and Procedures: One surgeon (KRK) maintained 100% case capture within ACS SSR from July 2014 through August 2016. Additional variable fields were generated within SPSS including “referring physician”, “insurance status”, “BMI” and others. Variables were recoded as necessary for statistical analysis. CPT codes were grouped into procedure bundles (i.e. “lap_right_colectomy”) for outcome measurement. As a project goal was ease of use, a single coding script was generated and run on a monthly basis after the online SSR database was exported into the SPSS database (Version 23.0. Armonk, NY: IBM Corp.). Pre-defined reports were run using ACS SSR and compared to SPSS analysis.
Results: The study dataset included 536 cases completed at a tertiary referral center. The SSR “Outcomes by Frequency” report was identical to SPSS frequency reporting for “death”, “return to operating room” and “readmission to hospital” variables. The SSR “Wound Infection Report” only reported “Superficial Incisional SSI”, ignoring organ and deep space infections. Predefined SSR reports were unable to analyze anything. To attempt analysis, the user must run the “Post-Op Occurrence by CPT Code” report, find each complication occurrence and click through to the associated patient record. This proves inadequate without methods for statistical analysis. Further, grouping by “emergency” field or comorbidity is not possible, precluding ability for risk-adjustment. In contrast, SPSS provides all necessary functionality for effective analysis of outcomes. On average, 1 hour a month was required to import data into SPSS, update values for user-defined variables, run the previously-defined script, and evaluate outcomes.
Conclusions: Minimal data supports use of non-100% capture data for individual surgeon outcomes. Data entry by non-surgeons may lead to misidentification of outcomes or risk-adjustment variables. Using the ACS SSR with a commercially available statistical package enables surgeons to control their data and effectively demonstrate outcomes.
Presented at the SAGES 2017 Annual Meeting in Houston, TX.
Abstract ID: 79241
Program Number: P654
Presentation Session: Poster (Non CME)
Presentation Type: Poster
sages_adbutler_leaderboard