Accuracy of isolated nutritional indicators in diagnosing malnutrition and their prognostic value to predict death in patients with gastric and colorectal cancer: a prospective study.

BACKGROUND The study aims to evaluate the accuracy of isolated nutrition indicators in diagnosing malnutrition in patients with gastric and colorectal cancer and their association with mortality. METHODS Prospective cohort study involving patients with cancer (n = 178) attending a reference center of oncology at any point in the disease trajectory or treatment. Nutrition status was evaluated in a unique moment by body mass index (BMI), Patient-Generated Subjective Global Assessment (PG-SGA), handgrip strength (HGS), and calf circumference (CC). Kappa coefficient, accuracy, sensitivity, specificity, positive and negative predictive values, and area under the receiver operating characteristic curve (AUC) were calculated for each nutrition indicator (PG-SGA as the reference method). The Cox proportional hazards model was used to assess hazard ratio (HR) and CI of mortality. RESULTS From the total patients, 11% were underweight, 48% were malnourished (PG-SGA B or C), 43% had low HGS, and 55% presented low CC. There were 46 deaths (25.8%). BMI, HGS, and CC showed poor and fair agreements (κ < 0.30 for all ) and poor accuracy (AUC < 0.70 for all) in identifying malnutrition by PG-SGA. After the adjustment for confounders (age, treatment performed, site, and stage of cancer), PG-SGA (HR, 2.9; 95% CI, 1.5-5.9) and low CC (HR, 2.4; 95% CI, 1.1-5.2) were independent predictors of mortality. CONCLUSION The nutrition indicators are not accurate in diagnosing malnutrition, whereas PG-SGA and low CC could predict mortality in gastric and colorectal cancer patients. Thus, CC should be combined with PG-SGA in nutrition assessments.