OBJECTIVETo compare the value of three preoperative nutritional assessment methods, European nutrition risk screening 2002(NRS 2002), mini-nutrition assessment(MNA) and subjective global assessment(SGA), in predicting postoperative complications of gastrointestinal cancer patients.

METHODSA total of 235 patients with gastrointestinal cancers, including 31 esophageal cancers, 82 gastric cancers, and 122 colorectal cancers, in our hospital from January 2012 to June 2013 were prospectively enrolled. Preoperative nutritional status was evaluated with above 3 methods respectively. Postoperative complication rates were compared among different preoperative nutritional status.

RESULTSAccording to SGA score, the morbidity of severe-moderate, mild and no malnourished patients was 40.5%(17/42), 25.3%(22/87) and 14.2%(15/106) respectively(P<0.01). According to MNA score, the morbidity of patients with malnutrition, at risk of malnutrition and without malnutrition was 32.9%(23/70), 24.7%(18/73) and 14.1%(13/92) respectively(P<0.05). According to NRS 2002, the morbidity of patients at malnutrition risk and without malnutrition risk was 27.6%(27/98) and 19.7%(27/137) respectively(P>0.05). Multiple regression analysis revealed that both SGA and MNA scores were predictive factors for the development of postoperative complications(both P<0.01). The sensitivity of SGA score for predicting complications was higher compared to MNA score (90.7% vs. 79.6%), while the specificity was similar(49.7% vs. 50.8%).

CONCLUSIONSBoth SGA and MNA scores can effectively predict the development of postoperative complications in gastrointestinal cancer patients, and SGA score has better sensitivity. SGA score is recommended for decision-making regarding preoperative nutrition support.