A local nutritional screening tool compared to malnutrition universal screening tool

Transcription

1 (2007) 61, & 2007 Nature Publishing Group All rights reserved /07 $ ORIGINAL ARTICLE A local nutritional screening tool compared to malnutrition universal screening tool K Gerasimidis 1, P Drongitis 1, L Murray 2, D Young 3 and RF McKee 4 1 Human Nutrition Section, Division of Developmental Medicine, University of Glasgow, Yorkhill Hospitals, Glasgow, UK; 2 Department of Nutrition and Dietetics, Glasgow Royal Infirmary, Glasgow, UK; 3 Department of Statistics and Modelling Science, University of Strathclyde, Livingstone Tower, Glasgow, UK and 4 Department of Surgery, Glasgow Royal Infirmary, Glasgow, UK Objective: The aim of the study was to compare the Glasgow Nutritional Screening Tool with the Malnutrition Universal Screening Tool (MUST) recently recommended for use by the British Association for Parenteral and Enteral Nutrition. Design: Comparison-validation study. Setting: Four adult acute hospitals in Glasgow, UK. Subjects: All 242 in-patients from a variety of specialties. Methods: Two investigators independently interviewed 202 in-patients for the comparison-validation study. Each used a single tool with each patient, using each tool in turn. Investigators were not aware of each other s assessments. Forty other patients were interviewed by both raters separately using the local tool to evaluate inter-rater reliability. Results: When compared with MUST as a gold standard, the local tool had a sensitivity of 95.3% and a specificity of 64.9%, with moderate agreement between the two tools using kappa test (k ¼ 0.57). Agreement between the raters was substantial (k ¼ 0.69) with 85% of patients classified the same by both raters. Conclusion: The Glasgow Nutritional Screening Tool is a valid and reliable tool that can be used on admission for nutritional screening. Sponsorship: University of Glasgow, Greek State Scholarship Foundation. (2007) 61, ; doi: /sj.ejcn ; published online 31 January 2007 Keywords: nutritional screening tool; malnutrition universal screening tool; glasgow nutritional screening tool; concurrent validity; reliability Introduction Studies over the past decades have shown a high prevalence of undernutrition in healthcare facilities, whereas at the same time other authors have shown that the nutritional status of the hospital patients is disregarded (McWhirter and Pennington, 1994; Edington et al., 2000; Kelly et al., 2000). Undernutrition impacts on morbidity, mortality, length of Correspondence: Dr R McKee, Department of Surgery, Glasgow Royal Infirmary, 16 Alexandra Parade, Glasgow G31 2ER, UK. ruth.mckee@northglasgow.scot.nhs.uk Guarantors: K Gerasimidis and R McKee. Contributors: KG contributed to the Recruitment, Research Activities, Data Analysis and Writing up of the manuscript. PD contributed to the Recruitment, Research Activities and Data Analysis of the manuscript. LM contributed to the Study Design of the manuscript. DY contributed to the Statistical Analysis of the manuscript. RM contributed to the Study Design, Recruitment, Data Analysis, Writing up and was the Project coordinator of the manuscript. Received 26 April 2006; revised 2 November 2006; accepted 2 November 2006; published online 31 January 2007 hospital stay and costs (Persson et al., 2002; Correia and Waitzberg, 2003) whereas nutritional support can improve patients clinical outcome (McWhirter and Pennington, 1996). Thus, it is important to identify patients at risk of undernutrition. There is no consensus on the best method for accurate assessment of nutritional status. In addition, it is impossible for dietitians to perform a full nutritional assessment of each patient. Therefore, nutritional screening, usually by the admitting nurse, has been recommended as a means of selecting at-risk patients who should then be referred to the dietitian (Laporte et al., 2001; Corish et al., 2004). National Health Service Quality Improvement Scotland recently published standards for food, fluid and nutritional care, which state that all patients should be screened for undernutrition on admission and periodically during their stay at hospital (NHSQIS, 2003). A plethora of nutrition screening tools (NSTs) have been developed, often related to a particular patient group

2 (Wright, 1999; Thorsdottir et al., 2001; Bauer and Capra, 2003). The British Association for Parenteral and Enteral Nutrition (BAPEN) has developed an NST called the Malnutrition Universal Screening Tool (MUST) and has recommended that it is suitable for widespread use. It uses body mass index, recent unplanned weight loss and acute disease effect equally to produce an overall malnutrition risk score. Management guidelines and/or a care plan are followed based on the overall malnutrition risk score (BAPEN, 2003). The Glasgow Nutritional Screening Tool (Glasgow NST) is a locally developed screening tool, which has been used for several years in some wards in Glasgow Royal Infirmary (Figure 1). Although based on the validated tool, Birmingham Heartlands Tool (Reilly et al., 1995), Glasgow NST has not been formally validated. Each of its six questions, relating to strong predictors of undernutrition (Kondrup et al., 2003), are graded with a score from zero to five. Overall scores of more than six signify a risk of malnutrition and a care plan advises either re-scoring in a few days (score: 7 9) or referral to dietitian (score: X10), depending on the score. Although NHSQIS recommended the use of MUST for nutritional screening, there was a reluctance to abandon our local tool, which was popular with the users. The objective of this study was to compare the performance of the Glasgow NST with the authoritative MUST tool. Subjects and methods Study site and sample Two samples of patients were recruited, one for the evaluation of concurrent validity and another one for an inter-rater reliability study. Participants were in-patients from a variety of specialties within North Glasgow Trust. Each of six specialties was targeted for 1 week of a 6 week study period and as many adult admissions as possible, both elective and emergency, were included. Patients who were unable to give consent, had psychological problems, were pregnant or were unconscious were excluded. The required sample size for the concurrent validity study was calculated using the equation recommended by Jones (2004b) for an undernutrition prevalence of 40% (McWhirter and Pennington, 1994; Stratton et al., 2004), for 90% sensitivity and a maximum error of 5%. Additional patients were interviewed from Glasgow Royal Infirmary to evaluate the inter-rater reliability of the 917 Glasgow Royal Infirmary GLASGOW NUTRITIONAL SCREENING TOOL Date: Weight (kgs) PATIENT LABEL Height (metres) PLEASE CONTACT THE DIETITIAN IF A SPECIAL DIET IS REQUIRED e.g. LOW FAT, DIABETIC, REDUCTION, LIQUIDISED, GLUTEN FREE, ALLERGY RESCORE - DATE BODY WEIGHT Normal (no recent weight changes) Recent unintentional weight loss (<6 kgs) Underweight / weight loss > 6 kgs APPETITE LOW RISK 0-5 MEDIUM RISK 6-9 ACTION PLAN Encourage normal diet Check weight weekly Re-assess if condition changes Commence 3 Day Food Record Chart Check weight twice weekly Reassess after 3 days Good-finishing three meals per day Reduced-leaving quarter meals and fluids Poor-leaving half meals and fluids Little or no appetite, refusing or unable to eat / drink ABILITY TO EAT AND DRINK Intake / weight increasing Intake / weight not increasing No difficulties, eating and drinking independently Requires assistance with eating and drinking Difficulty swallowing and / or chewing SKIN CONDITION Continue to encourage oral diet Refer to dietitian Healthy Some red pressure areas Superficial breaks in pressure areas Multiple deep pressure sores GUT FUNCTION Normal Persistent Nausea Nausea + / or occasional vomiting + / or some diarrhoea / constipation Diarrhoea > 3 per day / unable to keep food or fluids down HIGH RISK 10+ Refer to Dietitian and Medical Staff MEDICAL CONDITION No impairment to food intake Minor surgery / mild infection Major surgery (esp. G.I. Tract) / CVA / Chronic illness Severe infection / Sepsis / cancer / Burns>15% / Multiple injuries Date Weight Score Date seen by Dietitian Refer to Dietetic Care Plan TOTAL SCORE (REFER TO ACTION PLAN) * SCORE 10+ REFER TO DIETITIAN IF YOU FEEL THAT YOUR PATIENT REQUIRES DIETARY ADVICE DESPITE THE SCORE, PLEASE CONTACT THE DIETITIAN Figure 1 Glasgow Nutritional Screening Tool.

3 918 Glasgow NST but not that of the MUST. Patients from four different specialties (orthopaedics, plastic surgery, medicine and general surgery) were selected based on the protocol we followed for the comparison of the Glasgow NST with MUST. The research protocol was approved by the Ethics Committee of Glasgow Royal Infirmary. Written consent was obtained from each subject. Study design For the validation study, two graduates in dietetics studying for a postgraduate degree in Human Nutrition interviewed each patient independently, alternating between the two tools. For the reliability study, both assessors interviewed the same patients independently using the Glasgow NST. The results were recorded on two separate spreadsheets (MS Excel 2000), and the two spreadsheets were combined for analysis only at the end of the study. Each interviewer was blinded to assessments made by the other. For both NSTs, the same methods of assessment were applied. Specifically, for the measurement of height and weight, portable standardized electronic scales and a portable stadiometer (Seca, Leicester, UK) were used. Patients were weighed wearing light clothing. When patients were not able to stand or were bed-bound (e.g. orthopaedic wards), alternative measurements were used as recommended by BAPEN (BAPEN, 2003) for the assessment of body weight, height and BMI. Mid arm circumference was measured with a flexible, non-stretch measuring tape, midway between the acromion and olecranon on the left arm or on the right arm when the left was not available (e.g. plasters). In the absence of these measurements, recent height, weight and weight loss were extracted from the medical and nursing notes. When these were not available either, the self-reported weight and height of the patients were used. If these methods failed, the physical appearance of the patients and loose clothing, watches or jewellery were used to assess their weight status as recommended (BAPEN, 2003). For patients on chronic haemodialysis, their measured or self-reported dry weight after the end of dialysis was used. Demographic characteristics, diagnosis and pressure sore incidence were culled from the nursing and medical notes. Assessment of appetite, gut function, weight loss and ability to eat or drink was based on patient interview. Statistical methods In order to compare the two NSTs, and to gain statistical power, moderate-risk (score: 7 9 for Glasgow NST and 1 for MUST) and high-risk (score: X10 for Glasgow NST and X2 for MUST) groups were combined into one unified risk group as described by previous researchers (Ferguson et al., 1999; Jones, 2004a; Stratton et al., 2004). Contingency tables on the classification of the two screening tools were constructed to identify true positives, true negatives, false positives and false negatives and to calculate the sensitivity, specificity and predictive values (Deeks, 2001). There is no generally accepted clinical definition of undernutrition and no gold standard for assessing nutrition status. Thus, a study of diagnostic accuracy for a nutritional tool is particularly problematic. The data were analyzed by using Cohen s kappa test as a measure of agreement. Interpretation of the kappa value was according to tables from Landis and Koch (1977). Similarly, cross-classification of the data and Cohen s kappa test were used to check the inter-rater reliability of the Glasgow NST. The data was analysed using Minitab version 14 for Windows. Results Sample characteristics Two hundred and four patients were asked to participate in the comparison study. Two hundred and two (112 females) consented to participate and were interviewed. The median age of the participants was 64 years (range 18 95) with 49 and 75 years being the first and third inter-quartile values. Patients were recruited from the following specialties: medicine 24%, orthopaedics 20%, general surgical 19%, plastic surgery including head and neck and burns 15%, geriatrics 11%, oncology 11%. Of 202 patients, 180 had their weight and height measured, for 13 patients the data were culled from notes, five patients had mid-arm circumference measurements and, for four patients, other surrogates such as appearance and recalled data were used. Prevalence of patients at risk of undernutrition The risk of undernutrition using the two NST s is shown in Figure 2. For the combined nutritional risk categories, 60.4% of the participants were at risk of undernutrition according to the Glasgow NST and 42.1% according to MUST. Analysis of undernutrition risk by age group and NST showed that the prevalence of undernutrition risk was greater in older patients for both tools (risk of undernutrition: 470 years Glasgow NST 73%, MUST 56%; years Glasgow NST 57.90% MUST 42.10% 39.60% Glasgow NST Risk 60.40% Low Risk Figure 2 Charts showing incidence of undernutrition using the two tools divided into low-risk and risk (combined moderate-risk and high-risk) groups.

4 MUST Glasgow NGT a kappa value of 0.69, which is interpreted as substantial agreement. 919 Percent (%) Oncology Geriatric Medicine Orthopaedics Plastics Surgery Speciality Figure 3 Distribution of undernutrition risk between specialties using the two screening tools. Table 1 Cross-classification of patients undernutrition risk using the Glasgow NST and MUST (two risk categories) MUST Low Risk Risk Total Glasgow NST Low risk Risk Total Abbreviations: NST, Nutrition Screening Tool; MUST, Malnutrition Universal Screening Tool. 58%, MUST 33%; years Glasgow NST 44%, MUST 37%). The risk of undernutrition varied between specialties (Figure 3), although the numbers involved preclude statistical analysis. Comparison of Glasgow NST and MUST Four of the 85 (4.7%) patients who were screened being at risk of undernutrition by MUST were not identified by the Glasgow NST (Table 1). Forty-one of the 117 (35%) patients screened as low risk by MUST were assessed as at risk by the Glasgow NST (Table 1). Using MUST as the gold standard for the assessment of undernutrition, the Glasgow NST achieved high sensitivity (95.3%) and fair specificity (64.9%). The positive and negative predictive values of the Glasgow NST were 95 and 66.4%, respectively. Comparison of the two tools using Cohen s kappa test revealed moderate to substantial agreement (k ¼ 0.57) between the two methods using the interpretation values of Landis and Koch (1977). Inter-rater reliability of the Glasgow NST Forty additional patients were interviewed for the inter-rater reliability study. The two interviewers rated 34 out of 40 (85%) of the subjects at the same degree of risk. Evaluation of the inter-rater agreement with Cohen s kappa test resulted in Discussion In order for an NST to be useful, it should be simple, quick, non-invasive and cost effective, making it practical to use at ward level whereas enabling the routine assessment of patients by ward staff with minimal or no training on how to complete the tool (BDA, 1999; Burden et al., 2001). Although many NSTs have been described, most previous studies assessing these tools have been of poor quality. Jones (Jones, 2002) reviewed 44 studies of NSTs and found that 28 assessed validity, 20 reliability and 17 both of them. The majority of these studies used an appropriate method of analysis to estimate a tool s effectiveness but very little attention was given to the sample size required to achieve this. Only three studies attempted to address this issue, and only one referenced appropriate formulae. Our study deals with most of these issues. It includes a comparison with a well-validated and reliable tool, although not an actual gold standard, the sample size was based on published equations and its reliability was assessed despite the small sample we used. Comparison of NSTs In our study, the lack of an actual gold standard did not permit the direct validation of the Glasgow NST and its validity depends on the sensitivity and specificity of the MUST tool to which it is compared. A formal validation study would require a comparison with a global nutritional assessment using dietary assessment, anthropometry, functional, clinical and biochemical indices. Such validation is time consuming and requires a large number of patients. On the other hand, MUST has been validated against nutritional assessment methods and compared with other NSTs (BAPEN, 2003; Stratton et al., 2004). These previous studies permitted us to use it to define risk of undernutrition. The results of the comparison study revealed that the Glasgow NST is at least as sensitive in identifying undernourished patients as MUST. As far as specificity is concerned, the Glasgow NST either correctly assessed more patients as undernourished than MUST or the Glasgow NST lacks specificity. The present study did not have statistical power to distinguish between these two explanations. The kappa test showed moderate agreement between the Glasgow NST and MUST using Landis and Koch tables as the ranking scale. Other studies comparing NSTs have also found reasonable agreement between tools. Corish et al. (2004) compared the Nutrition Risk Index (NRI) and the original form of the tool we used (Reilly et al., 1995) against the same anthropometric gold standard and found moderate agreement. Burden et al. (2001) validated an NST, similar to ours, with anthropometrics and weight loss and found its sensitivity and

5 920 specificity to be 78 and 52%, respectively. Finally, Stratton et al. (2004) found that the MUST tool gave good to excellent agreement when compared with the original Birmingham Heartlands NST (Reilly et al., 1995) and other NSTs (kappa values from 0.55 to 0.89 depending on the NST) (Stratton et al., 2004). Inter-rater reliability The results from the inter-rater reliability study showed that the Glasgow NST is reliable when it was implemented by two different dietitians. The kappa value of 0.69 denotes substantial agreement between the assessors and is similar to that for other NSTs. Ferguson et al. (1999) estimated the reliability of the Malnutrition Screening Tool using at least two different dietitians or nutritionists as assessors and found good agreement for more than 93% of the cases with a kappa value higher than Prevalence of patients at risk of undernutrtion The prevalence of patients at risk of undernutrition in our sample is high, although comparable to some reports in the literature (Edington et al., 2000; Stratton et al, 2004). This may be owing to different definitions of undernutrition, use of different NSTs, changes in the prevalence of undernutrition (Elia and Stratton, 2000; Kelly et al., 2000), different study design or different case mix. The use of dietitians to screen patients may have increased the numbers judged to be at risk (Pattison et al., 1999). Our study included only in-patients in four large city hospitals. No day patients and few patients undergoing minor surgery were included. The Glasgow NST screened more people as at risk of undernutrition than MUST did. Whether this is attributed to the fact that MUST is not an actual gold standard, which infallibly assesses the actual nutritional status of the patients, or that Glasgow NST lacks specificity, is not clear. A comparison with a global nutritional assessment is needed to distinguish whether the Glasgow NST is accurately screening more patients at risk than MUST or is less specific. Moreover, the Glasgow NST is a screening tool, that identifies risk of undernutrition. The aim is to identify all those at risk rather than only those already undernourished, and therefore the criteria must be looser. Although some well-nourished patients may be screened as at risk of undernutrition, the ward dietitian could easily and quickly identify that this was inappropriate. An alternative explanation might be a different case mix. Bauer and Capra, (2003) have found that the Malnutrition Screening Tool (Ferguson et al., 1999), a tool quite similar to MUST, fails to identify some cancer patients at risk of malnutrition and it may be that our case mix with a large number of cancer patients even in general and plastic surgery helps to explain these results. Limitations Although our sample size was sufficient to be 95% confident that the sensitivity of the Glasgow NST was 95.3%, our study was underpowered to prove a specificity of 64.9%. Our estimate of sample size was based on an expected prevalence of undernutrition of 40%. Although this is supported by our results using MUST and other recent reports in the literature (Stratton et al., 2004), if the actual prevalence of malnutrition was lower, a bigger sample size would be required to support the validity of Glasgow NST. A weakness of our study is that the assessors did not evaluate the pressure sores personally because they were dietitians rather than nurses. The assessment was based on the nursing notes. However, this assessment is likely to be very accurate. Pressure sore assessment is included in the routine hospital admission procedure and nurses are keen to make every effort to prevent them (Brem and Lyder, 2004). We found comment on pressure sores was widely recorded and the incidence on admission was low. Another inevitable drawback was that the assessors were not blinded to the use of the NSTs, although they were blinded to the results of the other assessor s screening result for each patient and we were careful to ensure that the two sets of results were not combined until data collection was complete. We cannot exclude recall and report bias as a possible source of error for both screening tools. These tools rely on the patients ability to recall their body weight 6 months ago, and on how accurately they can report their appetite or gut function. Although we excluded patients who were overtly confused, inaccurate recall may well lead to inappropriate screening results. Finally, the use of researchers who were dietitians to screen patients may produce results that are different from those of the average busy ward nurse. Conclusion Although there are problems with the precise definition of undernutrition, a significant proportion of hospital patients do show evidence of undernutrition. Nutritional screening on admission by nursing staff should be included in hospital admission procedures. The Glasgow NST appears to be a useful tool with high sensitivity and inter-rater reliability. It is cheap, easy to use, quick and non-invasive, complying with the standards set by the various national dietetic associations for a functional NST. Acknowledgements We wish to thank the 242 participants as well as the nursing and medical staff of the wards who were involved in this study. Konstantinos Gerasimidis was granted a postgraduate scholarship from the Greek State Scholarships Foundation.

6 References Bauer J, Capra S (2003). Comparison of a malnutrition screening tool with subjective global assessment in hospitalised patients with cancer-sensitivity and specificity. Asia Pac J Clin Nutr 12, Brem H, Lyder C (2004). Protocol for the successful treatment of pressure ulcers Am. J Surg 188 (1A Suppl), British Association for Parenteral and Enteral Nutrition (2003). The MUST Explanatory Booklet. A guide to the MUST for adults. ISBN: Bapen, pp British Dietetic Association (1999). Nutrition Screening Tools. Professional Development Committee Briefing Paper no. 9. Burden ST, Bodey S, Bradburn YJ, Murdoch S, Thompson AL, Sim JM et al. (2001). Validation of a nutrition screening tool: testing the reliability and validity. J Hum Nutr Diet 14, Correia MI, Waitzberg DL (2003). The impact of malnutrition on morbidity, mortality, length of hospital stay and costs valuated through a multivariate model analysis. Clin Nutr 22, Corish CA, Flood P, Kennedy NP (2004). Comparison of nutritional risk screening tools in patients on admission to hospital. J Hum Nutr Diet 17, Deeks JJ (2001). Systematic reviews in health care: systematic reviews of evaluations of diagnostic and screening tests. BMJ 323, Edington J, Boorman J, Durrant ER, Perkins A, Giffin CV, James R et al. (2000). Prevalence of malnutrition on admission to four hospitals in England. The Malnutrition Prevalence Group.. Clin Nutr 19, Elia M, Stratton RJ (2000). How much undernutrition is there in hospitals? Br J Nutr 84, Ferguson M, Capra S, Bauer J, Banks M (1999). Development of a valid and reliable malnutrition screening tool for adult acute hospital patients. Nutrition 15, Jones JM (2002). The methodology of nutritional screening and assessment tools. J Hum Nutr Diet 15, Jones JM (2004a). Reliability of nutritional screening and assessment tools. Nutrition 20, Jones JM (2004b). Validity of nutritional screening and assessment tools. Nutrition 20, Kelly IE, Tessier S, Cahill A, Morris SE, Crumley A, McLaughlin D et al. (2000). Still hungry in hospital: identifying malnutrition in acute hospital admissions. QJMed93, Kondrup J, Allison SP, Elia M, Vellas B, Plauth M (2003). ESPEN guidelines for nutrition screening Clin Nutr 22, Landis JR, Koch GG (1977). The measurement of observer agreement for categorical data. Biometrics 33, Laporte M, Villalon L, Payette H (2001). Simple nutrition screening tools for healthcare facilities: development and validity assessment. Can J Diet Pract Res 62, McWhirter JP, Pennington CR (1994). Incidence and recognition of malnutrition in hospital. BMJ 308, Mcwhirter JP, Pennington CR (1996). A comparison between oral and nasogastric nutritional supplements in malnourished patients. Nutrition 12, NHS Quality Improvement Scotland (2003). Clinical Standards: Food, Fluid and Nutritional Care in Hospitals ISBN NHS Quality Improvement: Scotland, pp Pattison R, Corr J, Ogilvie M, Farquhar D, Sutherland D, Davidson HI et al. (1999). Reliability of a qualitative screening tool versus physical measurements in identifying undernutrition in an elderly population. J Hum Nutr Diet 12, Persson MD, Brismar KE, Katzarski KS, Nordenstrom J, Cederholm TE (2002). Nutritional status using mini nutritional assessment and subjective global assessment predict mortality in geriatric patients. J Am Geriatr Soc 50, Reilly HM, Martineau JK, Moran A, Kennedy H (1995). Nutritional screening evaluation and implementation of a simple nutrition risk score. Clin Nutr 14, Stratton RJ, Hackston A, Longmore D, Dixon R, Price S, Stroud M et al. (2004). Malnutrition in hospital outpatients and in-patients: prevalence, concurrent validity and ease of use of the malnutrition universal screening tool ( MUST ) for adults. Br J Nutr 92, Thorsdottir I, Gunnarsdottir I, Eriksen B (2001). Screening method evaluated by nutritional status measurements can be used to detect malnourishment in chronic obstructive pulmonary disease. J Am Diet Assoc 101, Wright L (1999). A nutritional screening tool for use by nurses in residential and nursing homes for elderly people: development and pilot study results. J Hum Nutr Diet 12,