Is subjective global assessment a reliable measure of nutritional status in hemodialysis?
Article Outline
Abstract
Objective
Subjective global assessment (SGA) is recommended in US and European guidelines for the nutritional assessment of patients with end-stage renal failure (ESRF). SGA identifies patient groups with abnormal nutritional parameters, but may fail to identify patients with malnutrition as identified by other techniques, such as total body nitrogen. We sought to compare SGA with a composite nutritional score.
Methods
HD patients were assessed by SGA, anthropometry, 3-day food diary, serum albumin, Kt/V urea, and normalized protein catabolic rate (nPCR). A composite nutritional score was derived from SGA, body mass index, percent of reference weight, triceps skinfold, midarm muscle circumference, and serum albumin.
Results
In 72 HD patients an abnormal SGA identified a patient group with reduced midarm circumference, midarm muscle circumference and serum creatinine and an increased composite nutritional score. However, overlap of nutritional scores was considerable between the normal and abnormal SGA groups, suggesting that SGA misclassified a large number of subjects. Serum albumin correlated with C-reactive protein (r = −0.473, P < .0001), not nutritional status. The composite nutritional score correlated with all of its components except for serum albumin.
Conclusions
SGA may not reliably identify hemodialysis patients with abnormal nutrition. Serum albumin is related to inflammation and not to nutrition status.
SUBJECTIVE GLOBAL ASSESSMENT (SGA) was originally developed to identify poor nutrition status in subjects undergoing gastrointestinal surgery,1 but has since been adapted for use in patients with chronic and end-stage renal failure2 and has been used to quantify the prevalence of malnutrition in hemodialysis and peritoneal dialysis patients.3, 4 An abnormal SGA score predicts increased mortality in peritoneal dialysis patients.5 SGA identifies a group of patients with reduced lean body mass, midarm muscle circumference, body mass index, and handgrip strength.6 However, there are concerns about using SGA in practice. Independent observers show only a moderate level of agreement when applying SGA scores to a population of patients.7 In the same study, SGA score was not a reliable predictor of nutrition state when compared with total body nitrogen as the gold-standard investigation.
We were concerned that SGA scoring used in our routine practice was not correctly identifying patients with clinically important malnutrition. We therefore compared SGA with a composite nutrition score to explore the value of SGA in this setting.
Methods
All patients receiving hemodialysis for end-stage renal failure in the York District Hospital renal unit were subject to nutrition and biochemical assessment as part of their routine care. Parameters measured included 3-day food diary, anthropometry (dry weight, height, and triceps skinfold thickness), and SGA score. Weight, triceps skinfold thickness, and midarm circumference were compared with standard charts to give percent of reference values. Dietary protein, calorie, sodium, and water intake were estimated from the food diary (Microdiet analysis). Serum albumin (bromocresol green) and CRP (Roche Tina quant immunoturbidometric assay) were measured. Kt/V urea and normalized protein catabolic rate (nPCR) were calculated by standard formulae. SGA was scored by an experienced renal dietician, both on the original 3-point scale (A to C) described by Detsky1 and on a 7-point scale recommended by the DOQI guidelines. A composite nutrition score was derived from a model described by Harty8 (Table 1).
Table 1. Composite Nutritional Score
| Score 0 | Score 1 | Score 2 | Score 3 | |
|---|---|---|---|---|
| SGA | A | B | C | - |
| Percent reference weight | >90 | 80–89 | 70–79 | <70 |
| BMI, male subjects (kg/m2) | >21 | 20–20.9 | 19–19.9 | <19 |
| BMI, female subjects (kg/m2) | >20 | 19–19.9 | 18–18.9 | <18 |
| Dry weight percentile | >15 | 10–15 | 5–10 | <5 |
| Triceps skinfold percentile | >15 | 10–15 | 5–10 | <5 |
| Midarm muscle circumference percentile | >15 | 10–15 | 5–10 | <5 |
| Albumin (g/L) | >35 | 30–34.9 | 25–29.9 | <25 |
The collected data were compared by SGA score using parametric or nonparametric statistics as appropriate. Correlations between nutrition parameters were sought by the Pearson test. The spread of composite nutrition scores in patients with normal versus abnormal SGA score were compared by nonparametric testing and by box and whisker plot. A P value of < .05 was regarded as significant.
Results
A total of 72 patients (42 male) underwent routine nutrition assessment. Mean age was 63.8 ± 17.3 years. Using the 3-point scale, SGA score was A in 50 (69%) and B in 22. No patient had a score of C, and groups were therefore analyzed as A versus B. Midarm circumference, midarm muscle circumference, serum creatinine level, and daily sodium intake were significantly reduced in the group with an abnormal SGA score (Table 2). Composite nutrition score was significantly increased in this group. Other measured parameters were not different. On the 7-point scale, SGA score was 3 in 3 patients, 4 in 6, 5 in 12, 6 in 17, and 7 in 34. Midarm circumference, midarm muscle circumference, serum creatinine level, and composite nutrition score were significantly different by 7-point SGA score (ANOVA, Table 3). Other measured parameters were not different.
Table 2. Comparison of Nutritional and Biochemical Parameters in 2 Groups of Hemodialysis Patients Defined by SGA Score on a 3-Point Scale
| SGA A (n = 50) | SGA B (n = 22) | P | |
|---|---|---|---|
| Weight (kg) | 68.2 ± 16.2 | 62.2 ± 18.2 | |
| Percent reference weight | 103.0 ± 24.7 | 100.5 ± 20.5 | |
| BMI (kg/m2) | 24.3 ± 4.5 | 23.1 ± 4.7 | |
| TSF (mm) | 14.5 ± 7.0 | 14.4 ± 5.0 | |
| MAC (cm) | 30.3 ± 4.1 | 28.1 ± 3.7 | .03 |
| MAMC (cm) | 25.8 ± 3.5 | 23.6 ± 2.9 | .007 |
| nPCR | 0.993 ± 0.21 | 0.988 ± 0.29 | |
| DPI (g/day) | 70.6 ± 18.3 | 63.1 ± 16.8 | |
| DPI/kg (g/kg ideal weight/day) | 1.13 ± 0.30 | 1.01 ± 0.31 | |
| Energy intake (kcal/day) | 1,860 ± 331 | 1,690 ± 369 | |
| Sodium intake (mmol/day) | 113.7 ± 27.9 | 94.0 ± 19.9 | .009 |
| Albumin (g/L) | 37.5 ± 2.8 | 36.2 ± 3.8 | |
| Creatinine (μmol/L) | 800 ± 238 | 676 ± 220 | .04 |
| CRP (mg/L) | 17.0 ± 23.6 | 36.8 ± 52.2 | .03 |
| Hemoglobin (g/L) | 12.2 ± 1.6 | 11.7 ± 1.6 | |
| Composite nutritional score | 2.11 ± 3.0 | 5.02 ± 3.6 | .003 |
Table 3. Comparison of Nutritional and Biochemical Parameters According to Groups of Hemodialysis Patients Defined by SGA Score on a Seven-Point Scale
| SGA (7-Point) | 3 (n = 3) | 4 (n = 6) | 5 (n = 12) | 6 (n = 17) | 7 (n = 34) | P |
|---|---|---|---|---|---|---|
| Weight (kg) | 48.3 ± 7.3 | 65.9 ± 11.9 | 63.5 ± 14.1 | 62.0 ± 18.4 | 71.1 ± 2.9 | |
| Percent reference weight | 85.9 ± 15.9 | 104 ± 13.69 | 99.6 ± 16.4 | 97.5 ± 19.4 | 106.6 ± 28.2 | |
| BMI (kg/m2) | 19.7 ± 3.7 | 24.0 ± 3.1 | 22.9 ± 3.8 | 22.5 ± 4.4 | 25.3 ± 4.9 | |
| TSF (mm) | 12.4 ± 2.7 | 16.3 ± 6.3 | 14.2 ± 4.9 | 13.2 ± 6.0 | 15.1 ± 7.5 | |
| MAC (cm) | 26.7 ± 2.6 | 28.5 ± 4.2 | 28.7 ± 4.1 | 27.5 ± 3.2 | 31.4 ± 4.0 | .009 |
| MAMC (cm) | 22.8 ± 2.0 | 23.4 ± 3.6 | 24.2 ± 3.1 | 23.6 ± 3.3 | 26.7 ± 3.2 | .005 |
| NPCR | 0.88 ± 0.22 | 0.99 ± 0.38 | 0.91 ± 0.21 | 1.07 ± 0.24 | 0.99 ± 0.22 | |
| DPI (g/day) | 54.4 ± 22.2 | 55.0 ± 21.6 | 61.3 ± 13.8 | 71.4 ± 16.9 | 73.1 ± 17.9 | |
| DPI/kg (g/kg ideal weight/day) | 0.98 ± 0.46 | 0.84 ± 0.29 | 0.96 ± 0.28 | 1.12 ± 0.26 | 1.19 ± 0.30 | |
| Energy intake (kcal/day) | 1,513 ± 118 | 1,585 ± 337 | 1,779 ± 324 | 1,775 ± 357 | 1,905 ± 357 | |
| Sodium intake (mmol/day) | 76 ± 10 | 104 ± 27 | 91 ± 15 | 114 ± 29 | 114 ± 27 | |
| Albumin (g/L) | 38.3 ± 5.5 | 34.7 ± 4.3 | 36.1 ± 2.6 | 36.5 ± 2.6 | 38.0 ± 3.0 | |
| Creatinine (μmol/L) | 409 ± 148 | 680 ± 102 | 670 ± 221 | 717 ± 190 | 861 ± 242 | .002 |
| Composite nutritional score | 8.7 ± 4.5 | 3.5 ± 3.1 | 4.4 ± 3.4 | 4.1 ± 3.7 | 1.4 ± 2.2 | <.001 |
Daily dietary protein, calorie, sodium, and fluid intake were all highly correlated. There was a trend toward decreasing nutrition intake with decreasing SGA score on the 7-point scale, but values were not significant by ANOVA (Table 3).
Serum creatinine was positively correlated with dry weight (r = 0.425, P < .0001), body mass index (r = 0.344, P = .003), and midarm muscle circumference (r = 0.436, P < .0001) and negatively correlated with composite nutrition score (r = −0.245, P = .04). On the 3-point SGA score, composite nutrition score was significantly lower in the group with a normal SGA score, but the overlap in values was considerable (Fig 1). On the 7-point SGA score, there was good discrimination of composite nutrition score between patients with the most abnormal (SGA = 3) and normal (SGA = 7) SGA (Fig 2). However, 3 patients with an SGA score of 7 had composite scores in the range seen in the most abnormal group. Patients with SGA scores in the midrange had complete overlap of composite nutrition scores (Fig 3).
Figure 1.
Box and whisker plot comparing composite nutritional score and SGA based on a 3-point scale (A to C) as described by Detsky.1 Values were significantly different (P < .0001; Mann-Whitney U test), but the overlap in scores was considerable. Bold line, median value; box, 25th and 75th percentiles; thin lines, 95% confidence intervals; circles, outliers.
Figure 2.
Box and whisker plot comparing composite nutritional score and SGA based on the 7-point scale recommended in the Disease Outcomes and Quality Initiative (DOQI) guidelines. Values were significantly different (P = .001; Kruskal Wallis test), but the overlap in scores was considerable, particularly for midrange SGA scores. Bold line, median value; box, 25th and 75th percentiles; thin lines, 95% confidence intervals; circles, outliers.
Figure 3.
The relationship between serum albumin and CRP (plotted on a log scale).
The composite nutrition score correlated with all of its components (3-point SGA score, r = 0.394, P = .001; 7-point SGA score, r = −0.472, P < .0001; percent of reference weight, r = −0.487, P < .0001; body mass index, r = −0.646, P < .0001; dry weight percentile, r = −0.581, P < .0001; triceps skinfold, r = −0.360, P = .002; midarm circumference, r = −0.719, P < .0001), except for serum albumin. Serum albumin correlated with CRP (r = −0.473, P < .0001) but not with any measure of food intake or nutrition status (Fig. 2).
Discussion
The technique of subjective global assessment allows a rapid, equipment-free scoring of nutrition status in patients with renal failure. This has led to the widespread use of SGA in renal patients, both in research studies and in clinical practice. US,9 UK,10 and European11 guidelines all recommend the use of SGA in screening renal patients for malnutrition.
SGA identifies patient groups with significantly lower body weight, body mass index, midarm muscle circumference, and handgrip strength.6 Lean body mass by anthropometry,6 bioimpedance,6 and DEXA12 are also reduced in patients with an abnormal SGA scores. In addition, a study of 76 hemodialysis patients found that the nitrogen index (measured total body nitrogen indexed to a reference population) was significantly lower in patients with an SGA score of B or C as compared with A.7 However, up to one quarter of patients with an SGA score of A had a nitrogen index of less than 85% predicted, and up to 80% of those with a score of C had an index greater than 85%. This led the authors of the paper to conclude that SGA was a poor predictor of nutrition status in hemodialysis. An earlier study of 29 patients on continuous ambulatory peritoneal dialysis (CAPD) compared SGA with a Marckmann score (based on relative body weight, triceps skinfold thickness, midarm muscle circumference, serum transferrin)13 and reported that the 2 scores only agreed in 50% of patients.
This study confirms that SGA identifies a patient group with abnormal nutrition as determined by a reduced midarm circumference and midarm muscle circumference. Dry weight, percent of reference weight, body mass index, and daily protein and daily calorie intake were lower but did not reach statistical significance. A composite nutrition score was also significantly greater, remembering that SGA is 1 of the parameters contributing to the score. Of concern, however, there was a considerable overlap in composite score between the groups. In particular, individual patients with an SGA of A showed high composite scores, suggesting significant abnormalities of nutrition. Using a modified SGA score based on a 7-point scale gave better discrimination of the best- versus worst-nourished patients as judged by composite nutrition score. However a number of subjects with apparently normal SGA scores had evidence of significant malnutrition on the composite score.
Serum albumin has frequently been used as a marker of nutrition status. This study further supports the growing body of evidence showing that serum albumin is related to inflammation14, 15 and not to nutrition state or protein intake.6 There was no relationship between serum albumin and any of the nutrition parameters measured, and serum albumin did not differ significantly by SGA score on either the 3- or the 7-point scale. The finding that serum albumin was the only parameter that did not correlate with the composite nutrition score lends further support to the view that it is not a marker of nutrition in renal failure.
Conclusion
The results of this study suggest caution over the use of SGA as a stand-alone assessment of nutrition status. No single measure of nutrition status is likely to be reliable in renal failure, and a composite score that includes both subjective and objective measures may represent the best method of cross-sectional and longitudinal assessment of dialysis patients.
References
- Predicting nutrition-associated complications for patients undergoing gastrointestinal surgery. JPEN J Parenter Enteral Nutr. 1987;11:440–446
- Subjective global assessment of nutrition in dialysis patients. Nephrol Dial Transplant. 1993;8:1094–1098
- Nutritional assessment of patients on continuous ambulatory peritoneal dialysis. Jpn J Nephrol. 1993;35:843–851
- Nutritional assessment of continuous ambulatory peritoneal dialysis patients (an international study). Am J Kidney Dis. 1991;17:462–471
- . Adequacy of dialysis and nutrition in continuous peritoneal dialysis (Association with clinical outcomes). J Am Soc Nephrol. 1996;7:198–207
- Assessment of nutritional status in CAPD patients (Serum albumin is not a useful measure). Nephrol Dial Transplant. 1997;12:1406–1413
- Validity of subjective global assessment as a nutritional marker in end-stage renal disease. Am J Kidney Dis. 2002;40:126–132
- The normalized protein catabolic rate is a flawed marker of nutrition in CAPD patients. Kidney Int. 1994;45:103–109
- . K/DOQI Clinical practice guidelines for nutrition in chronic renal failure. Am J Kidney Dis. 2000;35(suppl 2):S1–S140
- Treatment of Adults and Children With Renal Failure—Standards Document. (ed 3). London, England: Royal College of Physicians of London/Renal Association; 2002;
- Nutritional status in dialysis patients (A European consensus). Nephrol Dial Transplant. 2002;17:563–572
- Strong association between malnutrition, inflammation, and atherosclerosis in chronic renal failure. Kidney Int. 1999;55:1899–1911
- Nutritional status of CAPD patients in Lazio. Rivista Europea Per Le Scienze Mediche e Farmacologiche. 1993;15:149–153
- Factors affecting low values of serum albumin in CAPD patients. Adv Perit Dial. 1996;12:288–292
- . Factors influencing serum albumin in dialysis patients. Am J Kidney Dis. 1998;32:S118–S125
- . New standards of weight and body composition by frame size and height for assessment of adults and the elderly. Am J Clin Nutr. 1984;40:808–819
PII: S1051-2276(03)00139-0
doi:10.1053/j.jrn.2003.09.006
© 2004 National Kidney Foundation, Inc. Published by Elsevier Inc All rights reserved.
