Provider versus patient factors impacting hospital length of stay after pancreaticoduodenectomy

Background: Studies reporting perioperative outcomes after pancreaticoduodenectomy (PD) have focused on morbidity and mortality. Understanding factors that impact hospital duration of stay may have cost-saving implications. We sought to examine variation in duration of stay after PD occurring at the patient, surgeon, and hospital levels. Methods: Year-specific PD volumes for both surgeons and hospitals were determined from the 2003-2009 Nationwide Inpatient Sample and grouped into terciles. Patient age, gender, and comorbidities were examined. Median duration of stay was calculated and modified Poisson regression examined factors associated with duration of stay. Results: Among 5,190 individuals undergoing PD, median age was 65 years and 49.3% were female. Median duration of stay was 13 days (range, 0-234). Older patients and those with comorbid illness were more likely to have duration of stay of ≥14 days (P <.001). Median annual surgeon volume was 8 (interquartile range [IQR], 2-19; range, 1-54). Annual hospital volume ranged from 1 to 129 (median, 18; IQR, 6-52). Both low surgeon and hospital PD volume were associated with longer durations of stay (P <.001). In multivariable modeling, age remained associated with duration of stay (relative risk [RR], 1.007 per year; P <.001); however, comorbidity did not. Patients operated on by high-volume surgeons (RR, 0.67) or at high-volume hospitals (RR, 0.75) had a reduced risk of a prolonged duration of stay of ≥14 days (both P <.001). Conclusion: PD patients treated by higher volume surgeons and at higher volume hospitals had a shorter duration of stay. Although some patient-level factors impact duration of stay after PD, nonclinical factors such as surgeon and hospital volume were also important contributors to duration of stay.