ABSTRACT
Background: Surgical volume is known to influence failure to rescue (FTR), defined as death following a complication. Robotic lung surgery continues to expand and there is variability in outcomes among hospitals. We sought to estimate the contribution of hospital-based factors on outcomes and FTR following robotic right upper lobectomy (RRUL). Methods: Using the Centers for Medicare and Medicaid Services inpatient claims database, we evaluated all patients age ≥65 years with a diagnosis of lung cancer who underwent RRUL between January 2018 and December 2020. We excluded patients who had undergone segmentectomy, sublobar, wedge, or bronchoplastic resection; had metastatic or nonmalignant disease; or had a history of neoadjuvant chemotherapy. Primary outcomes included FTR rate, length of stay (LOS), readmissions, conversion to open surgery, complications, and costs. We analyzed hospitals by tertiles of volume and Medicare Mortality Index (MMI). Defined as the institutional number of deaths per number of survivors, MMI is a marker of overall hospital performance and quality. Propensity score models were adjusted for confounding using goodness of fit. Results: Data for 4317 patients who underwent robotic right upper lobectomy were analyzed. Hospitals were categorized by volume of cases (low, <9; medium, 9-20; high, >20) and MMI (low, <0.04; medium, 0.04-0.13; high, >0.13). After propensity score balancing, patients from tertiles of lowest volume and highest MMI had higher costs ($34,222 vs $30,316; P = .006), as well as higher mortality (odds ratio, 7.46; 95% confidence interval, 2.67-28.2; P < .001). Compared to high-volume centers, low-volume centers had higher rates of conversion to open surgery, respiratory failure, hemorrhagic anemia, and death; longer LOS; and greater cost (P < .001 for all). The C-statistic for volume as a predictor of overall mortality was 0.6, and the FTR was 0.8. Hospitals in the highest tertile of MMI had the highest rates of conversion to open surgery (P = .01), pneumothorax (P = .02), and respiratory failure (P < .001). They also had the highest mortality and rate of readmission, longest LOS, and greatest costs (P < .001 for all) and the shortest survival (P < .001). The C-statistic for MMI as a predictor of overall mortality was 0.8, and FTR was 0.9. Conclusions: The MMI incorporates hospital-based factors in the adjudication of outcomes and is a more sensitive predictor of FTR rates than volume alone. Combining MMI and volume may provide a metric that can guide quality improvement and cost-effectiveness measures in hospitals seeking to implement robotic lung surgery programs.
ABSTRACT
BACKGROUND: Peripancreatic fluid collections after distal pancreatectomy and splenectomy are commonly identified on postoperative cross-sectional imaging. This study aimed to determine the incidence, natural history, and indications for intervention. METHODS: We conducted a retrospective review of patients with peripancreatic fluid collections after distal pancreatectomy with or without splenectomy between 2013 and 2018, approved by our institutional review board. The chi-square test was used for categorical variables, the Mann-Whitney U test for continuous variables, and Fisher's exact test was used for values in which the sample size was less than 5 to compare data. RESULTS: During the study period, 235 patients underwent distal pancreatectomy with or without splenectomy, and 182 patients with postoperative imaging were included. In the cohort of patients with postoperative imaging, 83 (46%) had peripancreatic fluid collections, of which 46 (55%) were symptomatic fluid collections (SFCs) and 37 (45%) were asymptomatic fluid collections (AFCs). Those with SFC had a higher incidence of postoperative morbidity (46% vs 8%; P = .0002), most commonly postoperative pancreatic fistula (90%). Of patients with SFC, 34 (74%) underwent treatment via percutaneous drainage (n = 26), endoscopic drainage (n = 7), or antibiotics alone (n = 1). AFCs (n = 37) were observed. Collections that were intervened upon resolved significantly faster than those observed, 3.5 months vs 13.2 months (P < .0001), respectively. CONCLUSION: Asymptomatic patients may be observed with or without serial imaging and the AFC will typically resolve spontaneously with time. Patients who develop symptoms should generally be intervened upon with drainage if deemed feasible, given that this reduces the time to resolution.