Subjective Global Assessment in Chronic Kidney Disease: A Review

Nutritional assessment of patients with chronic kidney disease is a vital function of health care providers. Subjective Global Assessment (SGA) is a tool that uses 5 components of a medical history (weight change, dietary intake, gastrointestinal symptoms, functional capacity, disease and its relation to nutritional requirements) and 3 components of a brief physical examination (signs of fat and muscle wasting, nutrition-associated alternations in ﬂuid balance) to assess nutritional status. SGA was originally used to predict outcomes in surgical patients; however, its use has gone beyond this function and population. In chronic kidney disease patients, SGA is incorporated into the complete nutritional assessment. Validation of SGA as a screening tool for surgical patients was done by Detsky et al in 1984. Since that time, SGA has been altered by different researchers and clinicians to better meet the needs of the patients they served. Validation of the altered SGA formats has not been thoroughly done. Further work in establishing validity and reliability of each version of SGA in different patient populations should be done to enable clinicians and researchers to properly use this nutritional assessment tool. © 2004 by the National Kidney Foundation, Inc.

To address these issues, a Subjective Global Assessment Consensus Conference was organized by the Department of Nutrition of the School of Medicine of Case Western Reserve University and held on November 7 and 8, 2003, in Cleveland, OH.The objectives of the conference were (1) to review the methods, techniques, and tools being used for SGA; (2) to examine the validity of SGA; and (3) to identify how and by whom SGA is being used in clinical practice and research.Attendance at this conference was by invitation only; announcements were placed in the Journal of Renal Nutrition, the American Journal of Kidney Disease, the Journal of the American Dietetic Association, and on an SGA website: http://www-.nephrology.rei.edu/sgahome.htm.The announcements requested applications from people interested in attending and/or presenting at the conference.Thirty individuals (physicians and dietitians) were invited to attend.During the day-and-a-half conference, presentations included original research results, experiences with SGA in clinical practice, and experiences with SGA in education programs for dietetics students.Throughout the conference, attendees participated in roundtable discussions to generate ideas for validating SGA within the renal population.The consensus of the group of professionals who attended this conference is that further study must be conducted to standardize and validate SGA for the CKD population.Figure 1 outlines the recommended plan for further scholarly work with SGA.This article is one component of that plan, and is intended to review current literature available on SGA and to make recommendations on work to be done.

History of SGA
Detsky et al 1,10 published the first reports of a nutritional assessment tool, entitled SGA, that used clinical judgment to assess nutritional status in preoperative surgical patients and to predict postoperative infections; SGA had the best sensitivity and specificity for predicting infection after surgery.SGA was quickly used in other populations such as elderly patients, 11-13 patients with cancer 14 or liver transplants, 15 and adult patients undergoing maintenance dialysis. 2,3,6,16The original SGA form (Fig 2) as reported by Detsky et al 1 had clinicians score 5 components of a medical history (ie, weight change, dietary intake, gastrointestinal symptoms, functional capacity, disease and its relation to nutritional requirements) and 3 components of a brief physical examination (ie, signs of fat and muscle wasting, nutrition-associated alternations in fluid balance).The patient is then assigned a rating of well nourished (A), moderately undernourished (B), or severely undernourished (C) by subjective consideration of the data collected in the 8 areas, without adhering to a rigid scoring system. Hirsch et al validated SGA in 175 gastroenterology patients in 1990.That study found significant differences between well-nourished and moderately or severely undernourished patients in serum albumin, weight, midarm muscle circumference (MAMC), and triceps skinfold measurements. 17he first validation study in CKD patients occurred in 1993 16 with continuous ambulatory peritoneal dialysis (CAPD) patients.SGA was performed on 23 CAPD and 36 hemodialysis patients, and significant correlations were seen between the subjects' SGA ratings and values for serum albumin, bioelectrical impedance,

SUBJECTIVE GLOBAL ASSESSMENT IN CHRONIC KIDNEY DISEASE
MAMC, percent body fat, and normalized protein catabolic rate.This study's SGA methodology was used in the next major study in Canada and the United States (CANUSA) in the CKD population.CANUSA was a multicenter study conducted in Canada and the United States that investigated mortality and nutritional status in 680 patients on peritoneal dialysis. 12This study changed Detsky's A, B, C method of rating SGA to a 7-point scale (Fig 3).The components assessed remained the same, but the rating scale was expanded.Using survival analysis, the relative risk of death was increased with worsening nutritional status as defined by SGA and loss of lean body mass. 2 A major outcome of the CANUSA study was that a 1-unit decrease in SGA equaled a 25% increase in mortality for CAPD patients.The 7-point rating scale has been pilot tested by Visser et al 3 and Jones et al. 4 The cross-sectional study by Visser et al 3 on 13 hemodialysis and 9 peritoneal dialysis patients showed that SGA was positively correlated with body mass index (BMI), percent body fat, and MAMC.In a recently published article by Jones et al, 4 both the A, B, C (3-point) scale and the 7-point scale SGA forms were conducted with 72 hemodialysis patients.Statistical differences were found between SGA scores (both A, B, C and 7-point scales) for MAMC and serum creatinine. 4The A, B, C scale was also statistically different between A and B groups with the serum C-reactive protein concentration. 3alantar-Zadeh et al, 7 Stenvinkel et al, 9 and Pifer et al 8 have each studied different modified versions of SGA in samples ranging from 41 to 7,719 patients.Modifications in the rating scale (ie, from 7 points to 4 9 or 5 7 points) and the direction of data collection (ie, from prospective to retrospective 8 ) have been made.
In 1999, Kalanter-Zadeh et al 7 presented another version of the SGA that was originally referred to as modified quantitative SGA and in subsequent publications as the Dialysis Malnutrition Score (DMS). 18,19This fully quantitative version of SGA used the 7 original SGA components and created a quantitative scoring system.The scoring was a 5-point scale with 1 as normal and 5 as very severe malnutrition (Fig 4).The final score was the total sum of all 7 components.Each component was rated on a scale of 1 to 5 with a possible total range from 7 to 35.This method of SGA scoring produced high correlations with objective nutritional indicators such as total iron-binding capacity (TIBC) (r ϭ 0 to 0.77) and MAMC (r ϭ 0 to 0.66) and moderate correlations with serum albumin, BMI, bicep skinfold, age, and years on dialysis. 7he Malnutrition-Inflammation Score (MIS), developed by Kalantar-Zadeh, is a recently introduced, fully quantitative tool that is based on the 7 original SGA components and also includes 3 additional items (BMI and serum concentrations of albumin and serum TIBC). 18,20Each MIS component has 4 levels of severity from 0 (normal) to 3 (very severe).The sum of all 10 MIS components ranges from 0 to 30, denoting the increasing degree of severity (Fig 5).In a 2001 prospective study on 83 hemodialysis patients, MIS was compared with conventional SGA, its fully quantitative version (DMS), anthropometry, near-infrared measured body fat percentage, laboratory measures including serum C-reactive protein (CRP), and 12-month prospective hospitalization and mortality rates. 18MIS had significant correlations with prospective hospitalization and mortality as well as measures of nutrition, inflammation, and anemia in dialysis patients.The correlations were higher for MIS than either the conventional SGA or DMS with individual laboratory values as a predictor of outcome.In a 2004 recent multicenter study by the same group of investigators, the mortality and hospitalization predictability of the MIS was assessed in 378 hemodialysis patients; MIS was found to be comparable with serum CRP and serum interleukin-6. 20The MIS is currently being used in the multicenter Nutritional and Inflammatory Evaluation in Dialysis study (www.NIEDstudy.org). 21,22n 1999, Stenvinkel et al 9 published another version of the SGA.Although these researchers cited Detsky et al and Baker et al in their methods sections, Stenvinkel et al changed the scoring from the original A, B, C scale to a 4-point scale using 1 as normal nutritional status and 4 as severe malnutrition. 9Data on 109 adults with chronic kidney failure were analyzed by creating a bivariate variable with SGA scores 2 to 4 as one group and an SGA score of 1 as another group.In this manner they found those with scores between 2 and 4 were older, more frequently had a history of tobacco use, and had significantly lower BMI, serum creatinine, serum albumin, urine urea, and lean body mass (measured by dual-energy x-ray absorptiometry). 9e Dialysis Outcomes and Practice Patterns Study (DOPPS) study created m-SGA that was graded retrospectively using a patient interview.The score was based on the caregiver's ratings relative to weight loss, visual somatic store loss, appetite, nausea and vomiting, energy level, and disease burden.The rating for m-SGA is normal, moderate (any 3 areas rated as a moderate or severe level), or severe (at least 3 areas at severe level).Those patients who rated a severe m-SGA level had a relative risk of 1.33 for mortality compared with those with a moderate or normal rating, 8 which was statistically significant.
Although each of the versions has strengths, their lack of uniformity makes it difficult both to compare research results on nutritional status from one study to the next and to provide consistent methodology guidance for clinicians wishing to use this

Current Literature With SGA as a Nutritional Assessment Tool
Table 1 includes studies that used SGA as a method of nutritional status determination for further comparisons against a dependent variable (eg, mortality).From this table it is clear that SGA, using either the A, B, C or the 7-point scale, detects the presence of malnutrition; however, the controversy appears when SGA is correlated with serum albumin.In some studies serum albumin was significantly lower in the SGA malnourished group, 9,[23][24][25] whereas in others, serum albumin was not significantly different between the normal and the malnourished groups. 4,26Serum albumin is one of the most commonly used indicators for malnutrition in the CKD population, and although it is affected by several other factors including inflammation, this inconsistency has raised questions about the validity of SGA.To that end, incorporating serum laboratory markers for malnutrition may be a solution, as done in the MIS.
Interventional trials using kilocalorie and protein supplements, such as in the studies by Caglar et al 32 and Steiber et al, 33 have shown varying effects on changes in an individual's pre-SGA and post-SGA rating, depending on the intervention duration.The trial by Caglar et al 32 was 6 months, included 85 patients, and used the 7-point scale.They were able to show an improvement in the 7-point SGA over time; however, Steiber et al 33 did not see a significant change in pre-SGA and post-SGA scores over a 3-month period when the A, B, C rating system was used in 22 patients.

Recommendations
A review of the literature indicates that use of SGA as a nutrition assessment tool for CKD patients is growing, in both the clinical and research settings.However, given the variability of published results, SGA cannot be considered a gold standard in nutrition assessment for CKD patients.The validity and reliability of SGA must be proven in a large, multicenter trial with sufficient power to be able to prevent type I and II errors.Additionally, the study's sample must represent the current CKD population.One of the difficulties associated with conducting a study such as this is choosing which version of SGA to test.It may be that different SGA versions are appropriate for different patient disease states, different age stages, or different clinical purposes (eg, screening preoperatively versus full assessment of maintenance hemodialysis patients).Another difficulty is data collection.To get a representative sample, data would need to be collected from all areas of the country in a random manner.This could be done in a way similar to that of Beto et al 34 in a nationally collaborative research project through the National Kidney Foundation's Council on Renal Nutrition (CRN).Using this model, registered dietitians from local CRN groups throughout the United States could randomly collect data on patients in their dialysis centers.
Many of the studies reviewed collapsed the SGA scores into 2 groups (normal and malnourished) for analyses.For instance, Julien et al, 30 Lawson et al, 27 Abdullah et al, 35 and Jones et al, 26 used the A, B, C rating system and all dichotomized the final results by merging the B and C groups together for comparison against the A-rated group.Davies et al 28 used the 7-point scale and collapsed it into 6 to 7, 3 to 5, and 1 to 2 for analysis, and then grouped those with a 5 or less into a "malnourished group" and compared those patients with the 6 to 7 group.This method of analysis substantiates the conclusion of Cooper et al, 6 who found that SGA detects the presence of malnutrition but not the degree.It is possible that the need for the collapsed groups in such studies has more to do with inadequately powered studies or analytical tools (eg, logistic regression) than the lack of detectable precision of SGA.When presenting results of SGA in aggregate, it may be useful to show them in both a full and a collapsed or aggregated format.This would highlight any linear relationships as well as show differences between those with and without malnutrition.
In a large study with sufficient power, SGA may be able to detect differences between all 7/5 points or A, B, and C. Similarly, a continuous score may resolve the issue independent of sample size.Theoretically, with careful methodology and statistical analysis, a large, nationally representative study could be designed to determine the validity and reliability of SGA within the diverse United States CKD population.
Until the issue of which form of SGA is best suited to the hemodialysis population is determined, clinicians who are currently using one of the forms of SGA should continue to perform SGA.SGA is without a doubt a useful tool for nutritional assessment.However, as with all of the available tools, it should be used in conjunction with anthropometric, laboratory, and dietary intake measures to form a comprehensive nutritional assessment.

Figure 1 .
Figure 1.Plan for scholarly work by the SGA Consensus Conference Group.

Figure 2 .
Figure 2. A, B, and C original SGA.

Figure 4 .
Figure 4.The fully quantitative version of the SGA, also known as modified SGA or DMS.Five scale parameters are used, and the values are summed.A value of 7 is normal, and 35 is the most severe malnutrition.
tool.Currently the NKF regularly offers training sessions at its Clinical Nephrology meetings to train renal dietitians in the use of the 7-point SGA.No other formal training forum currently exists.Therefore, it is assumed that the majority of renal dietitians currently conducting SGA are using the version recommended by K/DOQI and studied by Visser et al3 and Jones et al.4