S. Parikh1, G. Sugiyama1, C. Choy1, G. Coppa1, P. Chung2,3 1Zucker School of Medicine Hofstra Northwell,Department Of Surgery,Manhasset, NY, USA 2SUNY Downstate Medical Center,Department Of Surgery,Brooklyn, NY, USA 3Coney Island Hospital,Department Of Surgery,Brooklyn, NY, USA
Introduction:
Cancer of the head of the pancreas is an aggressive disease with high rates of mortality. At this time surgical resection via a pancreaticoduodenectomy is the only potentially curative procedure. Although the procedure has in recent times become safer, the role that operative time has on postoperative outcomes is not well defined. Using the American College of Surgeons National Surgical Quality Improvement Program data, we sought to determine the role that operative time might play in length of stay.
Methods:
Using the 2010-2015 ACS NSQIP Participant Use Files (PUF) we identified cases in which pancreaticoduodenectomy was performed (CPT code 48150) in the setting of a postoperative diagnosis of pancreatic cancer (ICD9 code 157.0). We excluded cases that had emergency admissions, intraoperative wound classification of III or IV, and disseminated cancer. Cases with missing preoperative albumin, operative time, and total length of stay (LOS) data were excluded. We also excluded cases that had an operative time <15 minutes, and LOS >30 days. Multiple imputation for missing sex, race, functional status, and ASA classification was performed. Operative time was divided into quartiles (1st Quartile: <292.8 minutes; 2nd Quartile: 292.8 – 373.0 minutes; 3rd Quartile: 373.0 – 465.0 minutes; 4th Quartile: >465.0 minutes). Primary outcome was length of stay, for which negative binomial regression adjusting for age, sex, race, obesity, history of diabetes, chronic obstructive pulmonary disease (COPD), congestive heart failure (CHF), dyspnea, bleeding disorder, ascites, hypertension, renal failure, dialysis dependent, functional status, smoking status, steroid use, weight loss, preoperative transfusion within 72 hours of surgery, preoperative albumin levels, ASA class, and operative time, was performed.
Results:
3,424 patients that met inclusion/exclusion criteria were identified. Of these, 1,758 (51.3%) were male, majority were White (88.1%), and mean age was 66.1 years. Median LOS was 9.0 (SD 5.4) days. Negative binomial regression showed that presence of bleeding disorder (IRR 1.17, p=0.0035), 4th vs 1st quartile operative time (IRR 1.16, p<0.0001), and history of COPD (IRR 1.10, p=0.022) were associated with increased LOS. Higher preoperative albumin status was associated with decreased LOS (IRR 0.88, p<0.0001).
Conclusion:
We performed a large observational study using a national database. We found that increased operative time, even after adjusting for multiple preoperative and intraoperative risk factors, is independently associated with increased LOS in patients that undergo pancreaticoduodenectomy for pancreatic cancer. Further prospective studies are warranted to determine whether operative time should be used as a quality metric for patients undergoing pancreaticoduodenectomy.