In hospital outcomes after pancreatectomies: An analysis of a national database from 1996 to 2004

Kiran Turaga, Manas Kaushik, R. Armour Forse, Aaron R. Sasson

Research output: Contribution to journalArticle

21 Scopus citations

Abstract

Introduction: National complication rates following pancreatectomies have not been systematically reported. Methods: We queried the national hospital discharge survey (NHDS) database to analyze risk factors associated with mortality and length of stay after pancreatectomies. Results: An estimated 49,346 pancreatectomies were performed from 1996 to 2004. The national mortality rate is 9% with an average length of stay 15 days (Interquartile range 10-23) while the morbidity is 35%. Size of the hospital (<300 beds) (OR 2.76 (95% CI 1.14-6.70, P = 0.02)), post-operative pulmonary edema (OR 2.80 (95% CI 1.28-6.12, P = 0.01)) and sepsis (OR 5.22 (95% CI 1.94-14.11, P = 0.001)) are associated with higher mortality. Patients in larger hospitals (>500 beds) (Rate ratio 0.87 (95% CI 0.83-0.91, P < 0.001)) had a shorter hospital stay. Temporal trends reveal a shorter hospital length of stay in 2004 (Rate ratio 0.86 (95% CI 0.78-0.94, P = 0.001)) as compared to 1996. The percentage of pancreatectomies performed at larger hospitals in 1996 (40%) and 2004 (41%) has remained constant. Conclusion: The national mortality and morbidity rates after pancreaticoduodenectomy are 9% and 35%, respectively. Larger hospital size and absence of pulmonary edema and sepsis improves mortality. Larger hospitals have better outcomes although the trend for regionalization is not apparent.

Original languageEnglish (US)
Pages (from-to)156-160
Number of pages5
JournalJournal of Surgical Oncology
Volume98
Issue number3
DOIs
StatePublished - Sep 1 2008
Externally publishedYes

Keywords

  • High volume
  • Outcomes
  • Pancreatic resection
  • Whipple

ASJC Scopus subject areas

  • Surgery
  • Oncology

Fingerprint Dive into the research topics of 'In hospital outcomes after pancreatectomies: An analysis of a national database from 1996 to 2004'. Together they form a unique fingerprint.

  • Cite this