Back to 2008 Program and Abstracts
Optimizing Outcome Measurement in Pancreatic Surgery: Can Nsqip Measure Up?
Craig P. Fischer*1,2, Thomas a. Aloia1,2, Bridget N. Fahy1,2, Stephen L. Jones1,2, Barbara L. Bass1,2
1Surgery, The Methodist Hospital, Houston, TX; 2Weill Medical College of Cornell University, New York, NY
Intro: A validated system to measure and compare outcomes in complex GI and HPB surgery has not been established. The ACS NSQIP has been criticized as inadequate, citing lack of procedure-specific outcomes and limitations inherent to sampling methodology. Institutional registries have been an alternative, but do not allow site to site comparison. To assess the accuracy of NSQIP data collection, we investigated the utility of our institutional NSQIP data to capture procedure specific outcomes after a complex abdominal procedure - pancreatectomy.
Methods: From July 2006 - July 2007 31,332 surgical procedures were performed at our medical center. 1,638 (5.2%) were captured by the sampling method of NSQIP, including 32 pancreatic resections. Our institutional pancreatic database captured 75 patients having pancreatic resection during the same time period and included NSQIP occurrence categories and pancreatic fistula (PF) rates (ISGPF grade A,B, C). NSQIP nurses were instructed in ISGPF definitions and asked to retrospectively categorize patients with PF. Postoperative occurrences were compared.
Results: NSQIP captured equivalent outcomes to the pancreatic surgery database including rates of pancreatic fistula (table 1). However, grading of pancreatic fistula by NSQIP increased the ISGPF grade in 3 cases. Two cases were upgraded from A to B, due to the discovery of a drain in-situ in the outpatient setting. NSQIP discovered readmission of one patient, upgrading this patient from grade A to B. All three upgrades were not detected by the pancreatic registry, which only focused on in-hospital morbidities.
Conclusions: The data collected within the NSQIP at our institution compares favorably to a procedure specific clinical database. The sampling methodology inherent to NSQIP did not compromise the validity of outcome observations. The quality of data collection by NSQIP nurse reviewers may lead to more accurate appraisal and interpretation of outcomes due to lack of self reporting bias. The collected PF data were sufficient to allow accurate grading according the ISGPF system. These findings suggest that NSQIP may be further improved by consensus development of procedure specific outcome variables.
NSQIP Pancreatic Surgery Outcomes
| Category | Pancreatic Registry | NSQIP | P value |
| N | 75 | 32 | NA |
| Death | 2.66% | 3.13% | NS |
| Wound Infection | 9.33% | 12.5% | NS |
| Organ/Space SSI | 13.33% | 9.38% | NS |
| Respiratory | 8.0% | 6.25% | NS |
| Urinary | 5.33% | 9.36% | NS |
| Pancreatic Fistula (total) | 21.3% | 15.63% | NS |
SSI - surgical site infection.
