Development and validation of a Modified Patient‐Generated Subjective Global Assessment as a nutritional assessment tool in cancer patients
Zhenming Fu, Rui Zhang, Kunhua Wang, Minghua Cong, Tao Li, Min Weng, Zengqing Guo, Zeng‐Ning Li, Zhaoping Li, Chang Wang, Hongxia Xu, Chunhua Song, Cheng‐Le Zhuang, Qi Zhang, Wei Li, Hanping Shi, The INSCOC Study Group
Abstract
BACKGROUND: Completing Patient-Generated Subjective Global Assessment (PG-SGA) questionnaires is time consuming. This study aimed to develop and validate an easy-to-use modified PG-SGA (mPG-SGA) for cancer patients. METHODS: Seventy professionals assessed the content validity, comprehensibility, and difficulty of the full PG-SGA. A survey including the PG-SGA and other questionnaires was completed by 34 071 adult hospitalized cancer patients with first cancer diagnosis or recurrent disease with any tumour comorbidities from the INSCOC study. Among them, 1558 patients were followed for 5 years after admission. Reliability and rank correlation were estimated to assess the consistency between PG-SGA items and to select mPG-SGA items. The external and internal validity, test-retest reliability, and predictive validity were tested for the mPG-SGA via comparison with both the PG-SGA and abridged PG-SGA (abPG-SGA). RESULTS: After deleting items that more than 50% of professionals considered difficult to evaluate (Worksheet 4) and items with an item-total correlation <0.1, the mPG-SGA was constructed. Nutritional status was categorized using mPG-SGA scores as well-nourished (0 points) or mildly (1-2 points), moderately (3-6 points), or severely malnourished (≥7 points) based on the area under curve (0.962, 0.989, and 0.985) and maximal sensitivity (0.924, 0.918, and 0.945) and specificity (1.000, 1.000, and 0.938) of the cut-off scores. The external and internal validity and test-retest reliability were good. Significant median overall survival differences were found among nutritional status groups categorized by the mPG-SGA: 24, 18, 14, and 10 months for well-nourished, mildly malnourished, moderately malnourished, and severely malnourished, respectively (all Ps < 0.05). Neither the PG-SGA nor the abridged PG-SGA could discriminate the median overall survival differences between the well-nourished and mildly malnourished groups. CONCLUSIONS: We systematically developed and validated the mPG-SGA as an easier-to-use nutritional assessment tool for cancer patients. The mPG-SGA appears to have better predictive validity for survival than the PG-SGA and abridged PG-SGA.