Composite Measurement of Outcomes in Medicare Inpatient Laparoscopic Cholecystectomy

• Published in: Journal of the American College of Surgeons Vol. 221; no. 1; pp. 102 - 109
• Main Authors: Fry, Donald E., MD, FACS, Pine, Michael, MD, MBA, Locke, David, BS, Pine, Gregory, BA
• Format: Journal Article
• Language: English
• Published: United States Elsevier Inc 01.07.2015
• Subjects: Adult; Aged; Aged, 80 and over; Cholecystectomy, Laparoscopic - mortality; Decision Support Techniques; Female; Hospital Mortality; Humans; Length of Stay - statistics & numerical data; Logistic Models; Male; Medicare; Middle Aged; Odds Ratio; Outcome Assessment (Health Care) - methods; Patient Readmission - statistics & numerical data; Postoperative Complications; Risk Adjustment; Risk Assessment; Surgery; United States; United States; Medicare Severity-Diagnosis Related Group; present-on-admission; prLOS; Centers for Medicare and Medicaid Services; POA; CMS; prolonged postoperative length of stay; 90-DpD; 90 days postdischarge; MS-DRG
• ISSN: 1072-7515; 1879-1190
• DOI: 10.1016/j.jamcollsurg.2014.12.061

Background Objective measurement of outcomes in surgical care lack standard definitions, effective and consistent surveillance, and identification of significant postdischarge events. Study Design Using the Medicare Inpatient file (2009 to 2011), we designed logistic prediction models for inpatient mortality, prolonged length of stay (prLOS) as a measure of serious inpatient complications, and all-cause 90-day postdischarge (90-DPd) deaths and hospital readmissions for elective and nonelective laparoscopic cholecystectomy (LC). Qualifying hospitals had more than 20 cases for the study period and met rigorous present-on-admission coding standards. Results A total of 902 hospitals had 64,021 LCs. There were 509 inpatient deaths (0.8%) and 4,624 (7.2%) were prLOS. At 90-DPd, 729 patients died without readmission with a prediction model of 15 variables (C-statistic = 0.848), and 11,052 patients (17.4% of live discharges) were readmitted (1,165 died) with a prediction model of 36 variables (C-statistic = 0.674). Among significant (p < 0.0001) odds ratios (ORs), 90-DPd deaths were associated with age greater than 84 years (OR 3.7), prLOS (OR 7.8), operations performed on day 3 or thereafter in the index hospitalization (OR 1.6), and other chronic disease variables. Similar variables were associated with 90-DPd readmissions. A composite measure of all inpatient and 90-DPd deaths, prLOS for the index hospitalization, and 90-DPd readmissions resulted in an overall adverse outcome rate of 23.7% (15,195 of 64,021). Conclusions Adverse outcomes of inpatient deaths, prLOS, and 90-DPd readmissions and deaths provide an objective target for care redesign and improvement. The postdischarge period is the greatest source of adverse outcomes in LC. Composite measurement of adverse outcomes becomes a meaningful tool for the design of surgical warranties for episode-based payment initiatives.