European Journal of Clinical Nutrition (2005) 59, 3540

& 2005 Nature Publishing Group All rights reserved 0954-3007/05 $30.00
www.nature.com/ejcn

ORIGINAL COMMUNICATION
Prognostic significance of Subjective Global Assessment (SGA) in advanced colorectal cancer
D Gupta1*, CA Lammersfeld1, PG Vashi1, J Burrows1, CG Lis1 and JF Grutsch1
1

Cancer Treatment Centers of Americas at Midwestern Regional Medical Center, Zion, IL, USA

Objective: To evaluate the prognostic significance of Subjective Global Assessment (SGA) in advanced colorectal cancer and create statistically distinct prognostic groups of colorectal cancer patients based on clinical and nutritional variables. Design: A retrospective clinical epidemiologic study. Setting: A private tertiary care American Cancer Center. Subjects: In total, 234 colorectal cancer patients aged 2982 y treated at Cancer Treatment Centers of America at Midwestern Regional Medical Center between January 1995 and March 2001. Intervention: SGA Questionnaire. SGA Awell nourished; SGA Bmoderately malnourished; and SGA Cseverely malnourished. Malnutrition was defined as either SGA B or SGA C. Results: The prevalence of malnutrition in this patient population, as determined by SGA, was 52% (113/217). The median survival of patients with SGA A was 12.8 months (95% CI; 9.116.5), those with SGA B was 8.8 months (95% CI; 6.710.9) and those with SGA C was 6 months (95% CI; 3.98.1); the difference being statistically significant at P 0.0013. Regression tree analysis identified prior treatment history, lactate dehydrogenase (LDH) and SGA to be important predictors of survival for our patient cohort. Patients with no prior treatment history (newly diagnosed disease), low LDH scores, and SGA A had the best overall survival of 40.4 months (95% CI; 30.4550.4), whereas patients with prior treatment history (progressive disease), high LDH scores, and SGA B/C had the worst overall survival of 4.5 months (95% CI; 2.226.76). Conclusion: The SGA provides useful prognostic information in patients with advanced colorectal cancer. Sponsors: Cancer Treatment Centers of America.

European Journal of Clinical Nutrition (2005) 59, 3540. doi:10.1038/sj.ejcn.1602029 Published online 14 July 2004
Keywords: advanced colorectal cancer; subjective global assessment; nutritional assessment; prognostic significance; survival

Introduction
Colorectal cancer is the second leading cause of cancer death in the United States (Hawk et al, 2002). Despite significant advances in early diagnosis and treatment, 5-y survival rates for stages III and IV colorectal cancer remain poor. Various clinical, biochemical and histological prognostic factors for
*Correspondence: D Gupta, Cancer Treatment Centers of America at Midwestern Regional Medical Center, 2610 N Sheridan Road, Zion, IL 60099, USA. E-mail: digant.gupta@mrmc-ctca.com Guarantor: D Gupta. Contributors: DG was the main author of the manuscript, initiated the study, collected, analyzed and interpreted data. PGV supervised the project and assisted in interpretation of results. CAL and JB assisted in writing the manuscript. CGL initiated the study and assisted in writing and interpretation. JFG supervised the project and assisted with the statistical analysis. Received 4 November 2003; revised 10 May 2004; accepted 11 June 2004; published online 14 July 2004

advanced colorectal cancer have been identified. Tumor stage has proved to be the most decisive prognostic factor in both univariate and multivariate analyses (Lindmark et al, 1994; DEredita et al, 1996; Heys et al, 1998). Other prognostic factors for colorectal cancer include age, tumor grade, tumor size and location, symptom duration, and vascular and neural invasion. More recently, other prognostic factors have been identified including carcinoembryonic antigen (CEA), carbohydrate antigen 19-9 (CA 19-9), serum albumin, alkaline phosphatase, lactate dehydrogenase (LDH), aspartate aminotransferase, gamma-glutamyl-transpeptidase, and performance status (Chang et al, 1989; Kouri et al, 1992; Steinberg et al, 1992; Wang et al, 2002; Yuste et al, 2003). Although such factors do provide useful prognostic information, many of these factors cannot favorably influence survival. As a result, an urgent need exists to identify those prognostic factors that are modifiable. Intervention

SGA and survival in colorectal cancer D Gupta et al

36
directed at such factors could possibly lead to improved survival. Nutritional status is one such factor. However, the evidence documenting its prognostic significance in colorectal cancer remains sparse. In part, this is due to the difficulty of measuring nutritional status in a valid and reliable manner. Historically, nutritional status has been evaluated by various objective measures including anthropometric and laboratory measurements (eg weight change, arm muscle circumference, triceps skinfold thickness, serum albumin, transferrin assays and nitrogen balance studies). Some of these objective measures such as serum albumin are likely to be influenced by many non-nutritional factors (Bauer et al, 2002; Carney & Meguid, 2002; Waitzberg & Correia, 2003). The interpretation of these measures is often difficult because non-nutritional factors, such as hydration state and disease process, can obscure the effects of actual nutrient deprivation (Detsky et al, 1984). Furthermore, some objective indicators such as serum albumin have long half-lives, thus, assessing changes in the nutritional status over a short period of time is challenging. Moreover, in the clinical setting, these methods are not ideal because they are time consuming and require well-trained staff. In an effort to overcome the problems of traditional nutritional assessment, an easy-to-use, inexpensive, and noninvasive clinical instrument has been developedthe subjective global assessment (SGA). The SGA is a clinical technique that combines data from subjective and objective aspects of medical history (weight change, dietary intake change, gastrointestinal symptoms, and changes in functional capacity) and physical examination (loss of subcutaneous fat, muscle wasting, ankle or sacral edema and ascites) (Detsky et al, 1987). After evaluation, patients are categorized into three distinct classes of nutritional status; well nourished (SGA A), moderately malnourished (SGA B) and severely malnourished (SGA C). The SGA has been validated in a number of diverse patient populations including cancer patients (Enia et al, 1993; Ek et al, 1996; Jones et al, 1997; Duerksen et al, 2000; Sacks et al, 2000; Thoresen et al, 2002). It has also been correlated with a number of objective nutritional assessment indicators, morbidity, mortality, and quality of life (QoL) measures (Detsky et al, 1987; Hirsch et al, 1991; Hasse et al, 1993; Ferguson et al, 1999a, b; Persson et al, 1999; Bauer et al, 2002). The scored patient-generated subjective global assessment (PG-SGA) is a further modification of the SGA and has been specifically developed for patients with cancer. (Ottery, 1994, 1996). Unlike the SGA, which is categorical in nature, the PG-SGA measures nutritional status on a continuous scale, thus allowing for the detection of subtle changes in nutritional status over a short period of time. The higher the PG-SGA score, the greater the risk of malnutrition. Also, unlike the SGA, which must be completed entirely by the health professional, much of the PG-SGA can be completed by the patient. European Journal of Clinical Nutrition Although extensive research has been conducted on the validation of the SGA, a relatively small body of evidence exists regarding its use as a prognostic indicator in patients with cancer. To the best of our knowledge, no studies conducted to date have evaluated the prognostic significance of the SGA in advanced colorectal cancer. The primary objective of this study is to evaluate the prognostic significance of the SGA in patients with advanced (stages III and IV) colorectal cancer. The secondary objective is to create statistically distinct prognostic groups of colorectal cancer patients based on clinical and nutritional variables.

Patients and methods

Patients A retrospective chart review was performed on a consecutive series of 234 stages III and IV colorectal cancer patients treated at Cancer Treatment Centers of America (CTCA) at Midwestern Regional Medical Center (MRMC) between January 1995 and March 2001. The patients were identified from the MRMC tumor registry. Data were collected on a wide range of variables including demographic (age, sex); tumor-related (grade, stage, size, site); clinical (duration of symptoms, metastatic sites, lymph node involvement, comorbidities); prior treatment history (newly diagnosed vs progressive disease); laboratory (blood counts, tumor markers, serum enzymes) and nutritional (SGA, serum albumin, body mass index). The SGA and the scored PG-SGA were used to assess nutritional status; however, for the purpose of this study only the SGA ratings were used. All patients in this study were scheduled for a consultation with a dietitian. Prior to each consultation, a dietitian reviewed the patients history from the medical record and verified the patients current weight. During the consultation, the dietitians reviewed the SGA instrument with the patient to obtain answers to all the questions. The dietitians also completed a physical exam paying particular attention to loss of subcutaneous fat, muscle wasting, presence of ankle and sacral edema and ascites. After the consultation, the dietitians ranked the patients nutritional status as well nourished (SGA A), moderately malnourished (SGA B) or severely malnourished (SGA C) as described by Detsky et al (1987). For the purpose of this analysis, malnutrition was defined as either SGA B or SGA C. This study was approved by the Institutional Review Board at the Midwestern Regional Medical Center.

Statistical methods All data were analyzed using SPSS 11.0 (SPSS Inc., Chicago, IL, USA). Patient survival was defined as the time interval between the first date of patient visit to the hospital and date of death from any cause or date of last contact/last known to be alive. The KaplanMeier or product-limit method was

SGA and survival in colorectal cancer D Gupta et al

37
used to calculate survival. The log-rank test statistic was used to evaluate the equality of survival distributions across different strata. A wide range of demographic, clinical, pathological, and serological factors were evaluated for patient survival in a univariate analysis. KaplanMeier survival curves were constructed for each variable and the corresponding log-rank scores and P-values were also determined. The method of recursive partitioning or regression trees was used to identify distinct prognostic categories of the patient population. The variable with the highest log-rank score was used to sequentially bifurcate the patient population into two mutually exclusive groups. This process was repeated until no variable achieved statistical significance at Pr0.05. This analysis strategy allows for identification of a large number of prognostic groups, each containing only a small number of patients. There are several advantages to this approach. First, the scale of measurement is arbitrary, except that a monotonic relationship must be assumed between covariates and survival. Second, this technique allows us to classify patients into mutually exclusive groups, that can be easily described and understood (Green et al, 2003).
Table 1 patients Baseline characteristics of 234 stage III/IV colorectal cancer Categories Male Female Expired censored (lost to follow-up) Progressive disease Newly diagnosed Unknown Stage III Stage IV Well Moderate Poor Unknown Right sided Left sided A B C Unknown Mean 58.4 3.5 6.6 11.8 26.4 1222.8 1342.5 72.4 629.9 Number 133 101 200 34 170 63 1 95 139 17 164 43 10 89 145 104 75 38 17 s.d. 10.6 0.62 1.4 1.9 5.7 1565.9 3977.7 329.9 1619.6 Percent (%) 56.8 43.2 85.5 14.5 72.6 26.9 40.6 59.4 7.3 70.1 18.4 4.2 38 62 44.4 32.1 16.2 7.2 Range 2982 1.84.9 0.28.8 516.1 4.946.4 12910 212 025 704 13450 0.513 191

(A) Characteristic Sex Vital status

Prior treatment history Tumor stage at diagnosis Tumor grade at diagnosis

Tumor site Subjective global assessment

(B) Characteristic Age (y) Albumin (g/dl) Total protein (g/dl) Hemoglobin (g/dl) Body mass index (kg/ m2) Lactate dehydrogenase (U/l) Carbohydrate antigen 19-9 (IU/ml) Carbohydrate antigen 72-4 (ng/ml) Carcinoembryonic antigen (ng/ml)

Results
At the time of this analysis, 200 patients had expired and 34 were lost to follow-up. In all, 133 patients (56.8%) were male and 101 (43.2%) were female. The mean age was 58.4 y (s.d. 10.6; range 2982 y). A total of 170 (72.6%) patients had been diagnosed and treated outside of our hospital and 63 (26.9%) were newly diagnosed. Moderately differentiated tumors were the most common type (70.1%) followed by poorly differentiated (18.4%) and well-differentiated (7.3%) tumours. All patients had a histologically confirmed diagnosis of stages III or IV colorectal cancer. A total of 95 (40.6%) had stage III disease and 139 (59.4%) had stage IV. All tumors were adenocarcinomas. The prevalence of malnutrition in this patient population, as determined by SGA, was 52% (113/217). Table 1 shows the characteristics of our patient cohort in greater detail. Table 2 shows the univariate survival analysis of different prognostic factors. The variables have been rank-ordered based on their statistical strength of association with survival. Prior treatment history was most strongly associated with survival (log-rank 53.1, P-valueo0.0001) and was therefore the first variable to split our patient population into two mutually exclusive groups: patients with progressive disease and those with newly diagnosed disease. The SGA was also found to be significantly associated with survival on univariate analysis (Figure 1). The median survival of patients with SGA A was 12.8 months (95% CI; 9.116.5), those with SGA B was 8.8 months (95% CI; 6.7 10.9) and those with SGA C was 6 months (95% CI; 3.98.1), the difference being statistically significant at P 0.0013.

Table 2 Univariate analysis of prognostic factors of 234 stage III/IV colorectal cancer patients Variable Prior treatment history Lactate dehydrogenase Carbohydrate antigen 19-9 Number of prior chemo regimens Hemoglobin Carbohydrate antigen 72-4 Subjective global assessment Stage at diagnosis Carcinoembryonic antigen Albumin Total protein Tumor grade Tumor location Number of positive lymph nodes Body mass index Age Sex Log-rank score 53.1 44.5 41.47 31.82 20.59 19.31 13.36 13.02 12.95 11.61 6.88 3.3 3.2 2.1 0.38 0.18 0.01 P-value o0.0001 o0.0001 o0.0001 o0.0001 o0.0001 o0.0001 0.0013 0.0003 0.0003 0.0007 0.0087 0.192 0.0742 0.3431 0.539 0.675 0.942

European Journal of Clinical Nutrition

SGA and survival in colorectal cancer D Gupta et al

38

Discussion
According to the National Cancer Institutes Nutrition in Cancer Care guidelines, timely identification and treatment of nutrition problems may improve cancer patients prognosis by helping the patient gain or maintain weight, improving the patients response to therapy, and reducing the complications of treatment (http://www.cancer.gov/ cancerinfo/pdq/supportivecare/nutrition). However, the published literature documenting the prognostic significance of nutritional status in patients with advanced cancer remains sparse. Moreover, nutritional status has traditionally been assessed by anthropometric and laboratory measures, which are costly, time consuming, and difficult to perform. Therefore, we examined the prognostic significance of an easy-to-use and inexpensive clinical technique called the SGA in patients with advanced colorectal cancer. In this study, we found that SGA A (well-nourished status) vs SGA B/C (moderate-to-severe malnourished status) identified patients with better survival outcomes. We found that the SGA provides useful prognostic information in patients with advanced colorectal cancer and thereby, deserves serious attention in cancer treatment settings. In a clinical setting, the SGA is invaluable in identifying malnourished patients in a quick and noninvasive manner. Moreover, the simplicity of use of the SGA also enables health professionals other than oncologists to accurately assess the patients nutritional status. We also identified and described the distinct prognostic categories of patients with advanced colorectal cancer. This information could be of great value to clinicians in predicting the prognosis of a patient at admission and deciding upon subsequent treatment intervention. It also highlights the importance of assessing nutritional status of cancer patients early on in their treatment cycle. Another finding of this study was the identification of LDH as a useful prognostic indicator. Several studies have reported the prognostic significance of LDH in patients with metastatic colorectal cancer (Chang et al, 1989; Kemeny et al, 1989; Steinberg et al, 1992; Fountzilas et al, 1996). Indeed, one study (Kemeny & Braun, 1983) described the importance of LDH as an independent predictor of survival in advanced colorectal carcinoma. Our findings are consistent with these studies. In our regression tree analysis, LDH was the second variable to influence survival for both previously treated and newly diagnosed patients. The strength of this study lies in its adequate statistical power. A large sample size of 234 patients with survival information available for 85.4% (200) patients provides adequate power to effectively identify the true prognostic effect of the variables studied. Only 14.6% (34) patients were lost to follow-up in our study. The recursive partitioning analytic methodology used in this study has some drawbacks. For example, the precise split-points identified for our data set may not be duplicated in other similar data sets. As a result, this exploratory data analysis approach needs to be validated in other

Figure 1

Survival stratified by SGA categories (N 234).

Some other variables that were found to have a statistical association with survival were tumor stage, serum LDH, CA 19-9 and CA 72-4, serum albumin, total protein and CEA. Age, tumor grade, and tumor location were not found to have a statistically significant association with survival. Figure 2 shows the recursive-partitioning model that was constructed based on the strength of association of different prognostic variables with survival. It shows the variable splitpoints that were used to divide our patient cohort into statistically distinct and mutually exclusive groups. We found that for both previously treated and newly diagnosed patients, LDH had the highest statistical strength of association with survival (P-valueo0.0001 for both previously treated and newly diagnosed patients). Further analysis showed the SGA to be most strongly associated with survival (P-value 0.01 for previously treated patients with high LDH levels and 0.0003 for newly diagnosed patients with low LDH levels). The analysis stopped at the SGA because no other variable was found to be statistically significantly associated with survival beyond that point. Table 3 shows six (6) statistically distinct prognostic classes of our patient cohort. This classification scheme was derived from the final recursive-partitioning tree. Patients with no prior treatment history, low LDH scores, and SGA A had the best overall survival of 40.4 months (95% CI; 30.4550.4), whereas patients with prior treatment history (progressive disease), high LDH scores, and SGA B/C had the worst overall survival of 4.5 months (95% CI; 2.226.76). European Journal of Clinical Nutrition

SGA and survival in colorectal cancer D Gupta et al

39

Figure 2 Recursive partitioning tree. Variable split-points that were used to divide the patient cohort into statistically distinct and mutually exclusive groups.

Table 3 Prognostic classes of 234 stage III/IV colorectal cancer patients No. 1 2 3 4 5 6 Strata No treatment history, low LDH, and SGA A No treatment history, low LDH, and SGA B/C No treatment history and high LDH Prior treatment history and low LDH Prior treatment history, high LDH, and SGA A Prior treatment history, high LDH, and SGA B/C N 16 20 22 75 40 46 Median survival (in months) 40.4 32.8 10.8 13.7 7.4 4.5 95% CI 30.4550.4 5.959.61 022.5 9.9417.5 5.219.72 2.226.76

LDHlactate dehydrogenase; SGASubjective Global Assessment.

independent but similar data sets before it can be considered valid and reliable. Despite these limitations, recursive partitioning provides an effective way to classify patients into distinct and easily understood prognostic groupings. We also think that restricting the analysis to newly diagnosed patients (patients with no prior treatment history) would have been more accurate, since it would have allowed for

evaluation of true overall survival time, that is, time from the date of diagnosis to the date of death. However, doing so would have caused a significant reduction in the sample size. In our study, the survival time was calculated from the day of first visit at our hospital because the SGA information was not available at the time of diagnosis for previously treated patients. This drawback emphasizes the need for conducting European Journal of Clinical Nutrition

SGA and survival in colorectal cancer D Gupta et al

40
prospective studies having nutritional information available since the date of diagnosis. This study, because of its retrospective nature, relies on data not primarily meant for research. The SGA, being a subjective method, relies on the observers ability to collect and interpret information, and as a result, is likely to suffer from observer bias. No assessment of inter-rater reliability of the users of the SGA was made in this study. This bias, however, was minimized by restricting the use of the SGA to well-trained dietitians with an expertise in the use of this clinical instrument. This study did not evaluate the effectiveness of nutritional intervention on survival and future prospective studies should attempt to address this important research question. In summary, this study has identified the SGA as a useful prognostic indicator in patients with advanced colorectal cancer. However, the importance of the SGA as an independent prognostic indicator needs to be established in subsequent studies. This study provides a good starting point for future research in that direction.
Fountzilas G, Gossios K, Zisiadis A, Svarna E, Skarlos D & Pavlidis N (1996): Prognostic variable in patients with advanced colorectal cancer treated with fluorouracil and leucovorin-based chemotherapy. Med. Pediatr. Oncol. 26, 305317. Green S, Benedetti J & Crowley J (2003): Interdisciplinary Statistics: Clinical Trials in Oncology pp 219224. London/Boca Raton, FL: Chapman & Hall/CRC. Hasse J, Strong S, Gorman MA & Liepa G (1993): Subjective global assessment: alternative nutrition-assessment technique for livertransplant candidates. Nutrition 9, 339343. Hawk ET, Limburg PJ & Viner JL (2002): Epidemiology and prevention of colorectal cancer. Surg. Clin. North Am. 82, 905941. Heys SD, Walker LG, Deehan DJ & Eremin OE (1998): Serum albumin: a prognostic indicator in patients with colorectal cancer. J. R. Coll. Surg. Edinb. 43, 163168. Hirsch S, de Obaldia N, Petermann M, Rojo P, Barrientos C, Iturriaga H & Bunout D (1991): Subjective global assessment of nutritional status: further validation. Nutrition 7, 3537. discussion 3738. Jones CH, Newstead CG, Will EJ, Smye SW & Davison AM (1997): Assessment of nutritional status in CAPD patients: serum albumin is not a useful measure. Nephrol. Dial. Transplant. 12, 14061413. Kemeny N & Braun Jr DW (1983): Prognostic factors in advanced colorectal carcinoma. Importance of lactic dehydrogenase level, performance status, and white blood cell count. Am. J. Med. 74, 786794. Kemeny N, Niedzwiecki D, Shurgot B & Oderman P (1989): Prognostic variables in patients with hepatic metastases from colorectal cancer. Importance of medical assessment of liver involvement. Cancer 63, 742747. Kouri M, Pyrhonen S & Kuusela P (1992): Elevated CA19-9 as the most significant prognostic factor in advanced colorectal carcinoma. J. Surg. Oncol. 49, 7885. Lindmark G, Gerdin B, Pahlman L, Bergstrom R & Glimelius B (1994): Prognostic predictors in colorectal cancer. Dis. Colon. Rectum 37, 12191227. Ottery FD (1994): Rethinking nutritional support of the cancer patient: the new field of nutritional oncology. Semin. Oncol. 21, 770778. Ottery FD (1996): Definition of standardized nutritional assessment and interventional pathways in oncology. Nutrition 12, S15S19. Persson C, Sjoden PO & Glimelius B (1999): The Swedish version of the patient-generated subjective global assessment of nutritional status: gastrointestinal vs urological cancers. Clin. Nutr. 18, 7177. Sacks GS, Dearman K, Replogle WH, Cora VL, Meeks M & Canada T (2000): Use of subjective global assessment to identify nutritionassociated complications and death in geriatric long-term care facility residents. J. Am. Coll. Nutr. 19, 570577. Steinberg J, Erlichman C, Gadalla T, Fine S & Wong A (1992): Prognostic factors in patients with metastatic colorectal cancer receiving 5-fluorouracil and folinic acid. Eur. J. Cancer 28A, 18171820. Thoresen L, Fjeldstad I, Krogstad K, Kaasa S & Falkmer UG (2002): Nutritional status of patients with advanced cancer: the value of using the subjective global assessment of nutritional status as a screening tool. Palliat. Med. 16, 3342. Waitzberg DL & Correia MI (2003): Nutritional assessment in the hospitalized patient. Curr. Opin. Clin. Nutr. Metab. Care 6, 531538. Wang WS, Lin JK, Chiou TJ, Liu JH, Fan FS, Yen CC, Lin TC, Jiang JK, Yang SH, Wang HS & Chen PM (2002): CA19-9 as the most significant prognostic indicator of metastatic colorectal cancer. Hepatogastroenterology 49, 160164. Yuste AL, Aparicio J, Segura A, Lopez-Tendero P, Girones R, PerezFidalgo JA, Diaz R & Calderero V (2003): Analysis of clinical prognostic factors for survival and time to progression in patients with metastatic colorectal cancer treated with 5-fluorouracil-based chemotherapy. Clin. Colorectal Cancer 2, 231234.

Acknowledgements This study was funded by Cancer Treatment Centers of America.

References
Bauer J, Capra S & Ferguson M (2002): Use of the scored PatientGenerated Subjective Global Assessment (PG-SGA) as a nutrition assessment tool in patients with cancer. Eur. J. Clin. Nutr. 56, 779785. Carney DE & Meguid MM (2002): Current concepts in nutritional assessment. Arch. Surg. 137, 4245. Chang AE, Steinberg SM, Culnane M & White DE (1989): Determinants of survival in patients with unresectable colorectal liver metastases. J. Surg. Oncol. 40, 245251. DEredita G, Serio G, Neri V, Polizzi RA, Barberio G & Losacco T (1996): A survival regression analysis of prognostic factors in colorectal cancer. Aust. N. Z. J. Surg. 66, 445451. Detsky AS, Baker JP, Mendelson RA, Wolman SL, Wesson DE & Jeejeebhoy KN (1984): Evaluating the accuracy of nutritional assessment techniques applied to hospitalized patients: methodology and comparisons. J. Parenter. Enteral. Nutr. 8, 153159. Detsky AS, McLaughlin JR, Baker JP, Johnston N, Whittaker S, Mendelson RA & Jeejeebhoy KN (1987): What is subjective global assessment of nutritional status? J. Parenter. Enteral. Nutr. 11, 813. Duerksen DR, Yeo TA, Siemens JL & OConnor MP (2000): The validity and reproducibility of clinical assessment of nutritional status in the elderly. Nutrition 16, 740744. Ek AC, Unosson M, Larsson J, Ganowiak W & Bjurulf P (1996): Interrater variability and validity in subjective nutritional assessment of elderly patients. Scand. J. Caring. Sci. 10, 163168. Enia G, Sicuso C, Alati G & Zoccali C (1993): Subjective Global Assessment of nutrition in dialysis patients. Nephrol. Dial. Transplant. 8, 10941098. Ferguson ML, Bauer J, Gallagher B, Capra S, Christie DR & Mason BR (1999a): Validation of a malnutrition screening tool for patients receiving radiotherapy. Australas. Radiol. 43, 325327. Ferguson M, Capra S, Bauer J & Banks M (1999b): Development of a valid and reliable malnutrition screening tool for adult acute hospital patients. Nutrition 15, 458464.

European Journal of Clinical Nutrition