Validity of subjective global assessment as a nutritional marker in end-stage renal disease

Subjective global assessment (SGA) is a widely available clinical tool established to be of prognostic value in patients with end-stage renal disease (ESRD). Although it is reported to reflect a patient's nutritional status, its direct relationship to nutrition has not been determined. The aim of this study is to compare SGA with the gold standard for nutrition, total-body nitrogen (TBN) level, and thus determine the validity of SGA as a marker of nutritional status in patients with ESRD. Seventy-six consecutive dialysis patients referred from the renal service for routine measurement of TBN underwent simultaneous assessment of SGA determined by two independent examiners. Both examiners were blinded to TBN results. Only a moderate level of agreement was found in SGA score between the two examiners (weighted kappa score, 0.6). When patients were stratified into three nutritional groups determined by their SGA score, mean TBN values for each group by observer differed significantly (P = 0.0008 and P = 0.02, respectively). However, a significant statistical trend of worsening nutrition across SGA strata was found only for observer 2 (P = 0.049). Test performance of SGA as a predictor of malnutrition using a cutoff score of B (SGA = B or C) or C (SGA = C) was poor (positive likelihood ratios [LRs], 0.7 to 2.3; negative LRs, 0.5 to 1.0). Therefore, SGA appears not to improve the posttest probability of detecting malnutrition. SGA may differentiate severely malnourished patients from those with normal nutrition, but is not a reliable predictor of degree of malnutrition.