M. Kheng1, A. Manzella1, T. Kravchenko1, H. Pitt2, A. M. Laird1,2, T. Beninato1,2 1Rutgers Robert Wood Johnson Medical School, General Surgery, New Brunswick, NJ, USA 2Rutgers Cancer Institute of New Jersey, New Brunswick, NJ, USA
Introduction: The ACS-NSQIP database provides de-identified data on surgical procedures that is used for both secondary analysis and quality improvement efforts by participating hospitals. Since 2016, procedure-targeted (PT) data containing more detailed, procedure-relevant variables have been collected for thyroid operations. Total case volume-to-participant ratios are higher in the thyroidectomy PT dataset compared to the Participant Use Data File (PUF) containing all cases, suggesting potential volume-outcome differences between hospitals that provide targeted data for thyroidectomies and those that provide only generalized data for the main ACS database. Therefore, we aimed to determine whether patient variables and outcomes were similar between the thyroid PT and general NSQIP PUFs.
Methods: The 2016-2020 ACS-NSQIP general and PT PUFs were used to identify hemithyroidectomies, total thyroidectomies, and thyroidectomies with limited neck dissection, which constituted 33.9%, 38.6%, and 10.5% of all thyroidectomy cases, respectively. To control for differences in clinical practice across subspecialties, only cases attributed to general surgery were included. Patient demographics, perioperative variables, and rates of post-operative complications between cases in the PT data set and cases included only in the general PUF data set were analyzed. Pearson’s Chi square test and student’s t test were used where appropriate.
Results: Case volumes totaled 15,957 hemithyroidectomies, 23,659 total thyroidectomies, and 5,072 thyroidectomies with limited neck dissection. Procedure-targeted cases constituted 24.3-25.4% of cases within each operation type. There was a slightly higher proportion of males within the PT group compared to the main group (22.4% vs 20.8%, 95% CI 0.1-3.0%, p=0.04); patient characteristics including mean age, gender, ethnicity, ASA class, and comorbidities were otherwise the same across both groups and all three operations. Within the PT group, a slightly higher proportion of hemithyroidectomies were performed for malignancy (22.6% vs 20.7%, 95% CI 0.3-3.3%, p=0.015) and conversely, a slightly lower proportion of hemithyroidectomies were done for benign single nodules (26.9% vs 29.1%, 95% CI 0.6-3.8%, p=0.007). Differences in operative times, admission status, and hospital length of stay were not statistically significant. Rates of post-operative complications, unplanned reoperations, and readmissions were also similar for both groups and across all three surgery types.
Conclusion: Any group differences that exist between the hospitals that provide procedure-targeted data and those that provide generalized data for NSQIP are largely not reflected in patient demographics or outcomes for thyroidectomies. As such, findings from analyses of thyroidectomy procedure-targeted NSQIP data sets are likely applicable to the larger population of general surgeons who perform thyroidectomies at participating NSQIP hospitals.