Sample Selection
The American College of Surgeon’s National Surgery Quality Improvement Program (ACS NSQIP) database was queried to identify all patients undergoing TAA between January 1st, 2007 and December 31st, 2017. Both primary and revision procedures were considered using Current Procedural Terminology (CPT) codes 27702 and 27703. The ACS NSQIP database reports de-identified patient data and has been deemed HIPPA compliant. The Institutional Review Board (IRB) of Northwestern University approved this study as a retrospective cohort study.
The data provided in the ACS NSQIP database have been extensively investigated in various surgical fields, often to determine the incidence of short-term complications, identify risk factors for adverse short-term outcomes, and risk stratify patients for various procedures (22-25). ACS NSQIP reports over 150 variables, including patient demographics, comorbidities, lifestyle factors, preoperative laboratory values, operative variables, 30-day operative and medical complications, and 30-day disposition outcomes (e.g. return to OR, reoperation, and readmission). Given the substantial number of cases, the database is ideal for assessing low incidence complications after various procedures. ACS NSQIP has been shown to have excellent validity, reliability, and a consistently low rate of reporting error (26-28). Data sampling methodologies at participating institutions are routinely monitored, and interrater reliability audits are regularly performed to ensure data accuracy.
Measures
The total number of TAAs performed were identified for each year between 2011 and 2017. A one-way analysis of variance (ANOVA) with Games-Howell post-hoc test was used to compare the mean postoperative LOS after TAA for each year. For analysis of outcomes, only data between 2011 and 2017 were considered because 2011 was the first year that ACS NSQIP began reporting certain 30-day outcomes, including unplanned readmission, reoperation, and return to the operating room.
Prior to propensity-score matching, several patient variables – sex, age group, race/ethnicity, BMI classification, ASA classification, and comorbidities (diabetes, smoking, COPD, congestive heart failure, hypertension, dialysis, chronic steroid use) – were compared for patients with outpatient or short-stay hospitalization (LOS ≤ 1 day) versus standard inpatient hospitalization (LOS > 1 day). Pearson’s chi-squared test or Fisher’s exact test, where appropriate, were used to compare these categorical variables and various 30-day outcome measures between the two samples, including non-home discharge, mortality, return to the OR, readmission, reoperation, operative complications (surgical site infection, dehiscence, bleeding), and medical complications (wound infection, pneumonia, reintubation, failure to wean intubation, pulmonary embolism, renal insufficiency, renal failure, urinary tract infection, cerebral vascular accident, cardiac arrest, myocardial infarction, deep venous thrombosis, systemic sepsis, septic shock).
Propensity Score Matching and Statistical Analysis
Propensity-score matching was used to control for the differences in both modifiable and non-modifiable risk factors between the two disposition groups (LOS ≤ 1 day and LOS > 1 day). Specifically, patients from the two groups were paired in a 1:1 manner using a balanced, nearest neighbor approach based on the following variables: sex, age, BMI, ASA classification, and comorbidities (diabetes, smoking, COPD, CHF, hypertension, and chronic steroid use). All categorical variables (sex, ASA classification, and comorbidities) were matched exactly between the two groups. Continuous variables (age and BMI) were matched with a pre-defined tolerance of ± 10 years and ± 5 kg/m2, respectively. Additionally, age was divided into five groups: under 50, between 50 and 59, between 60 and 69, between 70 and 79, and 80 and over. BMI was divided into six groups: underweight (BMI < 18.5), normal (BMI 18.5 to 24.9), overweight (BMI 25.0 to 29.9), obesity class I (BMI 30.0 to 34.9), obesity class II (BMI 35.0 to 39.9), and obesity class III (BMI ≥ 40.0). After propensity-score matching, Pearson’s chi-squared test and, where appropriate, Fisher’s exact test were used to compare the rate of the patient and operative variables described previously to ensure these factors were statistically equivalent between the two groups. Lastly, the same statistical tests described previously were used to compare the rate of the 30-day complications between the matched groups. All statistical analyses were completed using IBM SPSS Version 24 (IBM Corp., Armonk, NY). The criterion for statistical significance was set at p ≤ 0.05.