Validity of the Patient-Generated Subjective Global Assessment (PG-SGA) in Colorectal Cancer Patients in China

Abstract: Objective The patient-generated subjective global assessment (PG-SGA) is a nutritional assessment tool specially designed for cancer patients. This study tested the validity of the PG-SGA for the nutritional assessment of colorectal cancer (CRC) patients in China. Methods A total of 8,093 Chinese patients with any stage CRC were enrolled in this cross-sectional, observational study. Within the first 48 hours of admission, patients were evaluated using the PG-SGA, nutritional risk screening 2002 (NRS 2002), Karnofsky Performance Status (KPS), and some anthropometric parameters, including the triceps skinfold thickness (TSF), mid-arm circumference (MAC), mid-arm muscle circumference (MAMC), hand grip strength (HGS), maximum left calf circumference (MLCC), and maximum right calf circumference (MRCC), among others. Spearman’s correlation test was performed to analyze item-total score correlations and correlations between the total PG-SGA score and other parameters or performance scores, further testing convergent validity. The discriminative ability of the PG-SGA was measured by comparing different characteristics between several nutritional groups. A principal component analysis was performed with selected parameters to evaluate the construct validity. Results The average total PG-SGA score of all patients was 5.69 ± 4.46, and they had a median age of 59 years (interquartile range (IQR), 51-67 years) and an average body mass index (BMI) of 22.57 ± 3.29 kg/m2. The item-total correlations of the total PG-SGA score between the global PG-SGA rating (Spearman, r = 0.94) and PG-SGA score (patient-generated) (Spearman, r = 0. 97) were strong. Significant correlations were also found between the total PG-SGA score and other nutritional screening tools, including the BMI (Spearman, r = -0.26), KPS (Spearman, r = -0.36), and NRS 2002 (Spearman, r = -0.47). Well-nourished (41.2%), mildly or moderately malnourished (35.3%) and severely malnourished (23.5%) groups defined according to the PG-SGA had significantly different characteristics. Patients with a worse nutritional status tended to have a decreased BMI (well-nourished, 23.32 kg/m2 vs. mildly or moderately malnourished, 22.52 kg/m2 vs. severely malnourished, 21.35 kg/m2; P < 0.001), hemoglobin, body fat mass, muscle mass, protein mass, TSF, MAC, HGS, MAMC, and MLCC, and a dramatically increased C-reactive protein level. The Kaiser-Meyer-Olkin measure was > 0.7, and the P value of Bartlett’s test of sphericity was < 0.001. One component was extracted by the principal component analysis, and the analysis showed that the total PG-SGA score explained the total variances of 97.3%. Conclusion The PG-SGA is a valid tool for evaluating the nutritional status for Chinese CRC patients.