Back to Annual Meeting Program

Background: There is now increased awareness of the detrimental effects of transfusion in elective general surgical procedures. Our objectives are to determine 1) which preoperative clinical variables can predict the need for intraoperative transfusion and 2) the impact of transfusion on hospital costs and complications in pancreatectomy.

Methods: Using our prospective institutional and ACS-NSQIP database, we identified 173 patients who had elective pancreatectomy from 9/2007 to 9/2011. Univariate and multivariate analyses were performed using 24 preoperative clinical variables to identify risk factors associated with transfusion. Preoperative severity of illness (SOI) and mortality risk were determined using the Agency for Health Research and Quality (AHRQ) Risk Adjustment Score, a standardized metric used by the University Health System Consortium. Hospital costs and operative complications were also evaluated.

Results: Patients had left pancreatectomy (n=60) or pancreaticoduodenectomy (n=113) to treat malignant (n=134) or benign (n=39) disease. Median OR time was 7.4 hours (2.4-12.3). Median LOS was 10 days (4-77) and 51 patients (29%) spent at least one night in the ICU. 98 patients (56.6%) had a complication and 90-day mortality was 2.9% (n=5). SOI at admission was minor in 21 patients (12.1%), moderate in 59 (34.1%), and major/extreme in 43 (24.8%). Risk of mortality at admission was: minor (n=91, 53%), moderate (n=58, 34%), and major (n=24, 14%).

There were 78 patients (45%) who received at least 1 unit of blood and the median number of intraoperative transfusions was 3.0 units (1-55); 11 of these patients (6.4%) also received plasma. Mean total hospital costs observed was \,434 (\,285-\,157). Compared to patients who did not receive a transfusion, those who received at least one blood product had a higher mean hospital cost and hospital charges (Table). Among transfused patients, 65% (n=51) experienced at least one complication vs. 49% (n=47) of patients not transfused (p=0.036), including infectious complications and pancreatic fistula/leak/abscess (Table). In multivariate analysis, independent predictors of increased transfusion likelihood included lower hematocrit, increased BMI, and worse AHRQ SOI and mortality risk scores. Age, gender, comorbidities, diagnosis, ASA class, procedure, OR time, and ICU stay were not independent predictors of transfusion risk.

Conclusions: Blood transfusion is associated with increased hospital cost and morbidity in pancreatectomy. Factors associated with increased risk for transfusion such as preoperative hematocrit, BMI and AHRQ scores can be utilized to stratify patients in clinical trials and inform patients of their risk for transfusion. Further research is needed to determine the extent to which transfusion contributes to morbidity and cost independently of SOI.

Table. Cost and complications associated with transfusion in pancreatectomy

Transfusion	Hospital Cost	p-value(a)	Hospital Charges	p-value (a)	Infectious Complication	p-value (b)	Grade III Pancreatic Fistula/Leak/Abscess	p-value (b)
Yes No	\,239 \,099	<0.0001	\,978 \,023	<0.0001	46% 18%	<0.0001	22% 3%	0.001

*p-values determined using (a) two-sample t-tests and (b) chi-square tests.