This study was performed under an exempt status granted by the Institutional Review Board of Rhode Island Hospital (IRB#1532652). The IRB determined that the study qualified for exemption under 45 CFR 46.101(b). The exemption was granted because the study involved a retrospective review of existing data recorded in such a manner that subjects cannot be identified, directly or through identifiers linked to the subjects. Clinical information of the subjects was obtained for the years between 2015 and 2018 from the American College of Surgeons (ACS) National Surgical Quality Improvement Program (NSQIP) database. The study is reported following the STOBE guidelines for reporting observational studies (Avery and Rotondi 2020).

The ACS-NSQIP database is a national prospective database that compiles voluntarily reported data from over 680 institutions in the United States. Over 1 million cases were submitted as part of the 2017 and 2018 update to the NSQIP database. Data is collected on over 300 variables that include preoperative risk factors, intraoperative variables and post-operative outcomes including complications up to 30 days after surgical procedures. Data collection has been previously described in detail (Raval and Pawlik 2018; Jiang et al. 2018). In brief, data are collected in 8-day cycles, with the first 40 procedures in the cycle included in the dataset. The most commonly performed procedures are capped at 5 within each cycle to increase procedure heterogeneity. Trained clinical nurses assigned at each site collect data for 30 days postoperatively using isolated telephone interviews and operative and clinical notes. Interrater reliability audits of selected participating sites help ensure the collected data are of the highest quality possible. The combined results of inter-rater reliability audits completed to date revealed an overall inter-rater disagreement rate of approximately 1.8% for all assessed program variables (Raval and Pawlik 2018; Jiang et al. 2018).

De-identified patient information is freely available to all institutional members who comply with the ACS NSQIP Data Use Agreement. The Data Use Agreement implements the protections afforded by the Health Insurance Portability and Accountability Act of 1996 and the ACS NSQIP Hospital Participation Agreement. The ACS NSQIP and the hospitals participating in this program are the sources of the data used in this study; however, these entities have not verified and are not responsible for the statistical validity of the data analysis or the conclusions derived by the authors.

The 2015 through 2018 NSQIP Participant Use Data Files were queried to extract all patients scheduled to undergo total thyroidectomy. Patients who underwent primary, total thyroidectomy were identified using the Current Procedural Terminology (CPT) codes 60220, 60225, 60240 and 60252. We excluded CPT codes 60210 (partial total lobectomy) and 60212 (partial total lobectomy with contralateral subtotal lobectomy), and 60260 (completion thyroidectomy) because we wanted to only include patients undergoing primary, total thyroidectomy for comparison. We also excluded CPT codes 60254, 60270, and 60271 as these represent more extensive surgery (e.g., neck dissection) and will very often require patient admission.

Preoperative demographic variables such as age, sex, body mass index, American Society of Anesthesiologists (ASA) classification, smoking status, hypertension, diabetes, congestive heart failure, disseminated cancer, bleeding disorder and COPD were compared between the two cohorts. Intraoperative factors including surgical duration and relative value units (RVUs) were also compared between the cohorts. RVUs reflect the relative level of time, skill, training, and intensity required of a physician to provide a given service. RVUs therefore are a method for calculating the volume of work or effort expended by a physician in treating patients. The primary independent variable was if the surgical procedure was performed in an outpatient (length of stay  < 1 day) versus an inpatient (length of stay  ≥ 1 day) stetting.

The primary outcomes of interest included any 30-day adverse events, defined as any surgical or medical complication within 30 days of surgery (Khavanin et al. 2014, 2015). Other outcomes of interest included surgical complications (e.g., overall surgical site infection (SSI), [which includes superficial SSI, deep incisional SSI, organ space SSI], wound dehiscence), medical complications (e.g., pneumonia, unplanned intubation), VTE [deep venous thrombosis, pulmonary embolism], failure to wean, progressive renal insufficiency, acute renal failure, urinary tract infection, stroke or cerebrovascular accident, cardiac arrest, myocardial infarction, bleeding, sepsis/septic shock, death, readmission, and return to the operating room.

Due to the observational (non-randomized) nature of this data, propensity score matching was used to minimize the effects of confounding when assessing differences in patient demographics and surgical characteristics between outpatient and inpatient thyroidectomies. The propensity score is the probability of treatment group conditional on observed baseline characteristics.

In this study, the probability for undergoing an inpatient thyroidectomy procedure (propensity score) was calculated for each patient based on age, sex, body mass index, diabetes, smoking status, dyspnea, chronic obstructive pulmonary disease, congestive heart failure, hypertension, disseminated cancer, steroid use, bleeding disorder, ASA classification, RVUs and surgical duration. Inpatient thyroidectomies were one-to-one matched without replacement to an outpatient thyroidectomy with the nearest propensity score, using a caliper of 0.10. If such a match was not available, the patient was eliminated.

Prior to matching, pre-operative demographics, comorbidities and surgical characteristics were compared using unpaired Student’s t test for continuous variables, and Chi-Square Test for binary variables. Pre-operative demographics were compared in the matched cohorts using paired t-tests for continuous variables, and McNemar’s Test for binary variables.

After propensity score matching, differences in outcome rates of the matched cohorts were assessed using McNemar’s test for matched data. Relative risks were calculated, as were risk differences. The rates of events between outpatient thyroidectomy and inpatient thyroidectomy were compared for outcomes at 30 days post procedure. To adjust for multiple endpoint testing, an adjusted p-value was calculated to correct for False Discovery Rate (FDR).

All statistical analyses were conducted with the use of SAS software version 9.4 (SAS Institute Inc., Cary, North Carolina).