- Email: [email protected]
The two most popular malnutrition screening tools in the light of the new ESPEN consensus definition of the diagnostic criteria for malnutrition
Accepted Manuscript The two most popular malnutrition screening tools in the light of the new ESPEN consensus definition of the diagnostic criteria for malnutrition Kalliopi-Anna Poulia, Stanislaw Klek, Ioannis Doundoulakis, Emmanouil Bouras, Dimitrios Karayiannis, Aristea Baschali, Marili Passakiotou, Michael Chourdakis, MD RD MPH PhD PII:
S0261-5614(16)30180-7
DOI:
10.1016/j.clnu.2016.07.014
Reference:
YCLNU 2881
To appear in:
Clinical Nutrition
Received Date: 18 January 2016 Revised Date:
14 July 2016
Accepted Date: 25 July 2016
Please cite this article as: Poulia K-A, Klek S, Doundoulakis I, Bouras E, Karayiannis D, Baschali A, Passakiotou M, Chourdakis M, The two most popular malnutrition screening tools in the light of the new ESPEN consensus definition of the diagnostic criteria for malnutrition, Clinical Nutrition (2016), doi: 10.1016/j.clnu.2016.07.014. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT 1
1
Title: The two most popular malnutrition screening tools in the light of the new ESPEN
2
consensus definition of the diagnostic criteria for malnutrition
3
Authors: Kalliopi-Anna Poulia1, Stanislaw Klek2, Ioannis Doundoulakis3, Emmanouil Bouras3,
5
Dimitrios Karayiannis4, Aristea Baschali4, Marili Passakiotou5, Michael Chourdakis3 a
6
RI PT
4
1
Department of Nutrition and Dietetics, Laiko General Hospital, Athens, Greece
8
2
Stanley Dudrick's Memorial Hospital, Skawina, Poland
9
3
School of Medicine, Aristotle University of Thessaloniki, Thessaloniki, Greece
10
4
Department of Nutrition and Dietetics, Evangelismos Hospital, Athens, Greece
11
5
ICU, Ippokratio General Hospital, Thessaloniki, Greece
M AN U
SC
7
12 13
Names for Pubmed indexing: Poulia, Klek, Doundoulakis, Bouras, Karayiannis, Baschali,
14
Passakiotou, Chourdakis
15
Part of the data previously presented at the 36th ESPEN Congress (2015)
TE D
16 17 18
a
19
Michael Chourdakis, MD RD MPH PhD
20
Aristotle University of Thessaloniki, Greece
21
Department of Medicine
22
University Campus, 54124
23
Phone: +30 2310 999035, Fax: +30 2312 205270
24
Email: [email protected]
EP
AC C
25
Correspondence:
26
Abbreviations:
27
ASPEN: American Society for Enteral and Parenteral Nutrition; AUC: area under the (ROC)
28
curve;, BMI: body mass index; BAPEN: British Society of Enteral and Parenteral Nutrition;
29
DRM: disease related malnutrition; ESPEN: European Society for Clinical Nutrition and
ACCEPTED MANUSCRIPT 2
Metabolism; IQR: interquartile ranges; K: Cohen’s Kappa value; LOS: length of hospital stay;
31
LR(+): positive likelihood ratio; LR(-): negative likelihood ratio; MUST: Malnutrition
32
Universal Screening Tool; NPV: negative predicting value; NRS2002: Nutritional Risk
33
Screening 2002; PPV: positive predictive value; ROC: receiver operating characteristic; SD:
34
standard deviation;
RI PT
30
AC C
EP
TE D
M AN U
SC
35
ACCEPTED MANUSCRIPT 3
Abstract
37
Background: The new definition of malnutrition in adults proposed recently by The European
38
Society for Clinical Nutrition and Metabolism (ESPEN) changed the view on the issue and raised
39
the question of the reliability of available diagnostic tools. Therefore, the aim of this study was to
40
verify the accuracy of the two most commonly used screening tools by comparing their findings
41
with the new ESPEN criteria.
42
Methods: Nutritional screening was performed in 1146 (median age 60 years, interquartile
43
range: 44–73 years, 617 males, 529 females) patients on admission to hospitals with two
44
nutritional screening tools: Nutritional Risk Screening 2002 (NRS2002) and Malnutrition
45
Universal Screening Tool (MUST). The screening results were then compared to the ESPEN
46
new diagnostic criteria for malnutrition.
47
Results: According to the NRS2002 13.5% and 27.9% of the outpatients and hospitalized
48
patients respectively were found to be at moderate/high risk of malnutrition. With the use of
49
MUST 9.1% and 14.9% of the outpatients and hospitalized patients respectively were found to
50
be at moderate/high risk of malnutrition. According to the ESPEN diagnostic criteria 6.4% and
51
11.3% of outpatients and hospitalized patients respectively were classified as malnourished.
52
MUST was found to be better correlated to the latter for both outpatients (K=0.777, p<0.001)
53
and hospitalized patients (K=0.843, p<0.001) as compared to NRS2002 (k=0.256, p<0.001 and
54
k=0.228, p<0.001). ROC plots Area Under the Curve (AUC) was found to be higher for MUST
55
compared to NRS2002 (0.964 vs. 0.695 for outpatients and 0.980 vs 0.686 for hospitalized
56
patients respectively).
57
Conclusion: To our knowledge, this study is the first to analyze the clinical value of a
58
malnutrition screening tool in the light of the new ESPEN definition for malnutrition. According
AC C
EP
TE D
M AN U
SC
RI PT
36
ACCEPTED MANUSCRIPT 4 59
to our results, MUST was better correlated with ESPEN criteria for the definition of
60
malnutrition, leading us to the conclusion that it can more efficiently identify the malnourished
61
patients, during the screening process.
RI PT
62
Keywords
64
Disease related malnutrition, nutritional screening, screening tools, diagnostic criteria, ESPEN
65
malnutrition definition
AC C
EP
TE D
M AN U
SC
63
ACCEPTED MANUSCRIPT 5
66
Introduction Malnutrition, also called disease-related malnutrition (DRM), can be the result of poor
68
nutritional intake, impaired utilization or loss of macro- and/ or micronutrients, or may stem
69
from acute or chronic inflammation [1-3]. It is clear that nutritional status may deteriorate during
70
hospital stay, leading to malnutrition, which occurs in 20-60% of hospitalized patients, but also
71
affects outpatients in lower percentages (7-16%)[1-10]. This very often passes unnoticed,
72
ignored or untreated despite the fact that it has been recognized as a major public health concern
73
leading to a significant economic burden. As a consequence, increased morbidity, mortality,
74
hospital readmissions and length of hospital stay have been observed [3-5, 7].
M AN U
SC
RI PT
67
The first step to successfully treat malnutrition is the appropriate diagnosis. There are two
76
major difficulties linked to that issue: the first one is the definition of malnutrition and the second
77
is choosing an appropriate screening tool to assess the risk. Quite recently the European Society
78
for Clinical Nutrition and Metabolism (ESPEN) published consensus-based new criteria for
79
malnutrition, which were meant to be applied independently of clinical setting and aetiology [8,
80
11] when screening for malnutrition gives a positive result. Few years earlier, a similar task was
81
undertaken be the American Society for Enteral and Parenteral Nutrition (ASPEN) [3].
EP
TE D
75
As defining malnutrition according to the ESPEN criteria has as prerequisite a positive
83
screening result from a screening tool, we have to keep in mind that screening for malnutrition is
84
not a simple task either. Over the last decades several screening tools have been introduced and
85
evaluated. Most of them combine almost the same variables, such as percentage of weight loss
86
during a defined period of time, body mass index (BMI), reduction of food intake, and the
87
presence of disease and its severity [8]. The Nutritional Risk Screening 2002 (NRS-2002) is the
88
tool proposed by the ESPEN guidelines mainly for detection of indications for nutritional
AC C
82
ACCEPTED MANUSCRIPT 6 89
support [12]. Another popular screening tool is the Malnutrition Universal Screening Tool
90
(MUST), which was developed to detect the risk of malnutrition for all adult patients across all
91
health care settings [13, 14]. Despite the widespread use of these screening tools, the new ESPEN definition for
93
malnutrition changes the screening strategy as it gives a base upon which the results of the
94
screening process could be evaluated. At the same time, however, it raises a question of the
95
validity of the former screening modalities. As the two afore mentioned tools are designed to
96
screen malnutrition -anyhow examining it from another point of view- we decided to assess
97
whether “high risk for malnutrition” or “high nutritional risk” as identified by the two screening
98
tools actually corresponds to the prevalence of malnutrition according to the ESPEN criteria.
99
Therefore, the aim of current study was to assess the tools used in clinical practise to screen
SC
M AN U
100
RI PT
92
patient for malnutrition in the light of the new ESPEN consensus definition of malnutrition.
102
Materials and methods
103
Subjects
TE D
101
This multicenter study enrolled one thousand one hundred forty-six patients [median age
105
60 years (interquartile range-IQR: 44–73 years), 617 males, 529 females] on admission to 19
106
hospitals in 11 Greek cities, from November 2014 until April 2015. Adult patients (≥18 years
107
old) admitted to internal and surgery wards/outpatients units (50.9%, 583 internal medicine and
108
49.1%, 563 surgical patients) were eligible to participate. They were consecutively invited to
109
participate whenever data collection was possible within the first 24 hours after admission and/or
110
during their examination waiting time at the outpatient’s units or emergency unit. Demographic
111
and medical data along with the two questionnaires for nutritional risk were collected during a
AC C
EP
104
ACCEPTED MANUSCRIPT 7 112
structured interview. Patients unable to communicate with the study’s personnel were excluded.
113
The population of the study group was homogenous, Caucasians, Greek or originated for Balkan
114
and other European countries. Participants were informed about the aim of the study and written consent was obtained
116
by all participants. The study was approved by the Medical Research Ethics Committee of the
117
Medical School of the Aristotle University of Thessaloniki.
119
SC
118
RI PT
115
Anthropometric measurements
Anthropometric measurements were performed with the subjects wearing light clothing,
121
without shoes. Body weight and height were measured at the time of recruitment, i.e. on
122
admission, using a calibrated weighting scale and a wall-mounted stadiometer to the nearest 0.5
123
kg and 0.5 cm, respectively. If height measurement was not feasible (e.g. in the case of an unable
124
to stand patient), self-reported height and weight were used (if reliable and realistic), from the
125
patient or the patient’s caregiver. BMI was computed as weight (in kilograms) divided by the
126
square of height (in meters squared). Percentage of unintentional weight loss over the last three
127
months was calculated from patients’ reports.
130
TE D
EP
129
Assessment of the nutritional risk
AC C
128
M AN U
120
For each patient, the steps of each tool were completed by the same investigator in the
131
same order. The total score for each screening tool was computed during the analysis of the data.
132
The assessors were encouraged not to add the scores for each tool during data collection in order
133
to avoid bias by the knowledge of the categorization in a screening tool.
ACCEPTED MANUSCRIPT 8
MUST uses current BMI, unintentional weight loss and the presence of any acute disease
135
effect that could compromise nutritional intake for >5 days [13]. It includes three parameters
136
rating them as 0, 1 or 2 as follows: BMI>20 kg/m2=0; 18.5–20.0 kg/m2=1; <18.5 kg/m2=2;
137
unintentional weight loss in the past 3-6 months <5%=0; 5–10%=1; >10%=2; acute disease:
138
absent=0; present=2. Overall risk of malnutrition is established after addition of all points
139
allocated, as follows: 0=low risk; 1=medium risk; 2=high risk [14]. It was primarily designed for
140
the use in the community but it has been validated in all settings, including hospital.
SC
RI PT
134
NRS2002, the nutritional screening tool proposed by the ESPEN guidelines for the
142
nutritional screening of patients [12, 15] includes an initial screening including four questions
143
about the following parameters: BMI (if it is <20.5), presence of weight loss in the past three
144
months, presence of low dietary intake in the past week and the severity of disease. In the case of
145
any positive response a final screening is required [12]. The final screening evaluates impaired
146
nutritional status (Score=0 for normal nutritional status, Score=1 for weight loss >5% in the last
147
3 months or food intake below 50-75% of the normal requirements, Score=2 for weight loss >5%
148
in the last two months or BMI 18.5-20.5 + impaired general condition or food intake 25-60% of
149
normal requirements and Score=3 for weight loss >5% in the last month or BMI <18.5+impaired
150
general condition or Food intake 0-25% of normal requirement) and the “severity of disease
151
score” of 0-3 plus an additional point if the patient is older than 70 years. Nutritional risk is
152
established after addition of all points allocated, as follows: <3 at no or low risk, ≥3= at high
153
risk.
TE D
EP
AC C
154
M AN U
141
We retrospectively used the new European Society for Clinical Nutrition and Metabolism
155
(ESPEN) criteria for the definition of malnutrition [11], namely BMI≤18.5 kg/m2, or weight loss
156
>10% (indefinite of time) or 5% (in 3 months) and BMI <20 kg/m2 for patients <70 years or <22
ACCEPTED MANUSCRIPT 9
kg/m2 for patients ≥70 years, but not Fat Free Mass Index-FFMI (below <15 and <17 kg/m²) to
158
identify malnourished patients in our study’s population and evaluate the correlation of these
159
criteria with the prognostic value of the two screening tools. As the analysis was performed
160
retrospectively, no data regarding FFMI in our population were available; therefore, the
161
categorization of malnutrition according to the ESPEN criteria, was based only in the BMI and
162
weight loss. As both options are given in the consensus it can be assumed that by choosing one
163
versus the other, the results of our evaluation would not be biased. The aforementioned analysis
164
was performed by an investigator from our group who was neither involved in the initial
165
screening tools analysis nor aware of its results, for blinding purposes.
M AN U
SC
RI PT
157
The total score and classification of malnutrition risk (low, medium or high) was
167
determined for each study participant and screening tool. The scores obtained by the two
168
screening tools were then related to the ESPEN newly established consensus definition of
169
malnutrition.
TE D
166
For the cross-tabulation of risk classification between the tools we decided to group the
171
classification of malnutrition risk into two categories (i.e. “high/moderate” vs. “low/no risk”) as
172
patients allocated in the high group category are the ones that need to be further referred for
173
assessment to the dietetic and medical team.
175 176
AC C
174
EP
170
Statistical analysis
Normally distributed continuous variables were presented as mean values and standard
177
deviations (SD), skewed variables were presented as medians and IQR, whereas categorical
178
variables as absolute and relative (%) frequencies. The Kolmogorov-Smirnov test was applied to
179
evaluate normality of the distributions. All reported P-values are based on two-sided tests and
ACCEPTED MANUSCRIPT 10
compared to a significance level of 5%. Cohen’s kappa (Κ) statistic was calculated to determine
181
diagnostic concordance between the assessment tools (i.e. MUST, NRS2002) and ESPEN’s
182
diagnostic criteria for malnutrition. K coefficient is a statistical measure of inter–annotator
183
agreement for qualitative variables. In case of complete agreement between the variables K=1,
184
whereas if there is no agreement among the variables measured (other than what would be
185
expected by chance) then K≤0. Positive and negative likelihood ratios (LR+ and LR-
186
respectively) were calculated for both tools. For diagnostic procedures similar to nutritional
187
screening with the tools that were used in the study, it is desired LR+ to be >10 and LR-<0.1.
SC
RI PT
180
Sensitivity, specificity, predictive values for the two nutrition screening tools were
189
calculated with the use of diagnostic criteria as the way to identify the malnourished patients, as
190
can be seen in Supplementary Table 1.
M AN U
188
Receiver operating characteristic (ROC) curves were also used to evaluate the ability of
192
NRS2002 and MUST to correctly distinguish between the malnourished and the non-
193
malnourished patients (according to the ESPEN consensus definition of malnutrition). Area
194
under the ROC curve equal to 0.5 indicates that a tool cannot distinguish between the two
195
groups, whereas area under the ROC curve equal to 1 indicates perfect separation of the values
196
of the two groups.
198 199
EP
Statistical analysis was performed using SPSS for Windows, version 16.0 (SPSS Inc,
AC C
197
TE D
191
Chicago, IL).
ACCEPTED MANUSCRIPT 11
200
Results The baseline characteristics of the sample are presented in Table 1. Median BMI was
202
26.5 kg/m2 (IQR: 24.1–29.4 kg/m2) and median hospital stay among the hospitalized patients
203
was 3 days (IQR: 2–7 days).
RI PT
201
204 205
Table 1: Characteristics of the sample of patients
(n=784)
185 (51.1%)
617 (53.8%)
177 (48.9%)
529 (46.2%)
Age§ (years)
57 (36–70)
62 (46–75)
60 (44–73)
BMI§ (kg/m2)
26.4 (24.9–29.8)
26.8 (23.9–30.5)
26.5 ( 24.1–29.4)
BMI (<18,5 kg/m2)*
12 (1.5%)
10 (2.8%)
22 (1.9%)
Recent weight loss (>5% in 3 months)*
71 (9.1%)
54 (14.9%)
125 (10.9%)
–
3 (2–7)
–
M AN U
352 (44.9%)
TE D
207
§
Values expressed as medians and interquartile range (IQR)
EP
208
The classification of the risk of malnutrition of the patients by the two screening tools showed a variation among the two tools and the ESPEN criteria as can be seen in Table 2.
AC C
212
(n=1146)
Female*
*Values expressed as frequencies and percentages
211
(n=362)
432 (55.1%)
206
210
Overall
Male*
Length of stay in hospital§ (days)
209
Hospitalized
SC
Outpatients
Table 2: Classification of the risk of malnutrition by the two screening tools and the ESPEN criteria
Risk of malnutrition Moderate/high Low
Outpatients (n=784)
MUST
NRS2002
9.1% (71/784) 90.9%
13.5% (106/784) 86.5%
Hospitalized (n=362) ESPEN Criteria* 6.4% (50/784) 93.6%
MUST
NRS2002
14.9% (54/362) 85.1%
27.9% (101/362) 72.1%
ESPEN Criteria* 11.3% (41/362) 88.7%
ACCEPTED MANUSCRIPT 12 (713/784) 213
(678/784)
(734/784)
(308/362) (261/362)
(321/362)
* Classification of malnutrition according to the ESPEN consensus definition of malnutrition Cross tabulation of the results of screening of nutritional risk with MUST and NRS2002
215
and the classification of malnutrition according to the ESPEN diagnostic criteria can be found in
216
Table 3.
RI PT
214
217
Table 3: Cross tabulation of the results of screening of nutritional risk with MUST (≥1) and NRS2002
219
(≥3) and the classification of malnutrition according to the ESPEN consensus definition of malnutrition
SC
218
M AN U
Outpatients (n=784)
Hospitalized (n=362)
Not at moderate/high nutritional risk
Malnourished
Not malnourished
Malnourished
Not malnourished
(N=50)
(N=734)
(N=41)
(N=321)
48
23
41
13
2
711
0
308
TE D
220
At moderate/high nutritional risk
ESPEN Criteria
At moderate/high nutritional risk
25
81
25
76
Not at moderate/high nutritional risk
25
653
16
245
EP
NRS2002
MUST
ESPEN Criteria
The analysis of the agreement between the diagnostic criteria of malnutrition and the
222
results of nutritional screening was found to be different, according to the tool used for the
223
screening. More specifically, better agreement with the diagnostic criteria of malnutrition was
224
found in MUST for hospitalized [96.4% (349/362) of the cases (K=0.843, p<0.001)] and
225
outpatients [96.8% (759/784) of the cases (K=0.777, p<0.001)]. NRS2002 was found to be in
226
lower agreement in both outpatients [86.5% (678/784) of the cases (K=0.256, p<0.001)] and
227
hospitalized patients [74.6% (270/362) of the cases (K=0.228, p<0.001)]. MUST was also found
228
to have a very high sensitivity and specificity for both inpatients and outpatients (MUST
AC C
221
ACCEPTED MANUSCRIPT 13
Sensitivity=96.0% for outpatients and 100% for hospitalized respectively and MUST
230
Specificity=96.0% for both subgroups vs NRS-2002 Sensitivity = 50.0% in the outpatients and
231
61.0% in the hospitalized respectively.), showing that the likelihood that a patient screened
232
positive for malnutrition risk with MUST to be actually malnourished according to the ESPEN
233
diagnostic criteria was higher, compared to NRS 2002, with which according to our results, 50%
234
and 39% respectively of malnourished patients according to ESPEN criteria were not identified
235
as at risk of malnutrition.
SC
RI PT
229
On the other hand, NRS2002 was found to have a lower sensitivity (50% for outpatients
237
and 61% for hospitalized) and specificity (89.0% for outpatients vs. 76.3% for hospitalized), and
238
in combination with the low positive predictive value (23.6% for outpatients and 24.8% for
239
inpatients) it was found that the likelihood of a false positive malnutrition screening result was
240
higher for patients screened with NRS2002.
M AN U
236
Furthermore, MUST was found to have higher LR+ and lower LR- compared to
242
NRS2002 for both groups confirming the better performance of MUST regarding the agreement
243
with the diagnostic criteria of ESPEN for malnutrition. More specifically, NRS2002 did not
244
perform well with only a fair agreement with the ESPEN criteria and low positive likelihood
245
ratio (LR+).
EP
Finally, the area under the curve as it was calculated by the ROC curves was also higher
AC C
246
TE D
241
247
in MUST compared to NRS2002 (0.964 vs. 0.695 for outpatients and 0.980 vs 0.686 for
248
hospitalized patients respectively), confirming the better ability of MUST to distinguish a
249
malnourished patient compared to NRS2002. Results are presented in details in Table 4.
250
ACCEPTED MANUSCRIPT 14 251
Table 4: Statistical evaluation of the nutritional screening tools compared to the diagnostic criteria of
252
malnutrition (MUST (≥1) and NRS2002 (≥3). Outpatients
Hospitalized
(n=784)
(n=362) MUST
NRS2002
50.0
96.0
Specificity (%)
89.0
96.9
Positive predictive value (%)
23.6
67.6
Negative predictive value (%)
96.3
99.7
Positive Likelihood ratio (LR+)
4.53
30.64
2.58
24.69
Negative likelihood ratio (LR-)
0.56
0.04
0.51
-
0.695
AUC
100.0
76.3
96.0
24.8
75.9
93.9
100.0
0.777 (<0.001) 0.228 (<0.001) 0.843 (<0.001)
M AN U
0.256 (<0.001)
61.0
SC
Sensitivity (%)
Κ value (p)
0.964
0.686
253
Κ value derived from Cohen kappa statistics.
254
MUST: Malnutrition Universal Screening Tool. NRS2002: Nutritional Risk Screening 2002,
255
AUC=Area under the curve from ROC plots
0.980
TE D
256 257
MUST
RI PT
NRS2002
Discussion
The purpose of using a screening tools is to identify patients at risk of malnutrition and to
259
select those individuals, who are in need for further evaluation and potential intervention.
260
According to the NRS2002 13.5% and 27.9% of the outpatients and hospitalized patients
261
respectively were found to be at moderate/high risk of malnutrition. With the use of MUST 9.1%
262
and 14.9% of the outpatients and hospitalized patients respectively were found to be at
263
moderate/high risk of malnutrition. Other studies evaluating the risk of malnutrition, showed
264
results ranging from 20-60% or even higher, reaching 80% when the population of patients is
265
elderly [2, 8, 9, 16-18].
AC C
EP
258
ACCEPTED MANUSCRIPT 15
Finally, 6.4% and 11.3% of outpatients and hospitalized patients among our study
267
population were classified as malnourished according to the ESPEN diagnostic criteria. This is
268
the first study to show the prevalence of malnutrition in a Greek outpatients and hospitalized
269
patients in a multicenter study, from several areas in Greece, and at the same time it is also
270
interesting to see the difference of the prevalence of malnutrition in outpatients and hospitalized
271
patients, putting emphasis on the need of more actions needed for the early identification of
272
malnutrition during hospital stay.
SC
RI PT
266
The wide range of the assessed prevalence of malnutrition or the risk of malnutrition can
274
be attributed to the differences in the nutritional screening tools and/or differences in the
275
population included. The outpatients’ group included free living subjects, either outpatients or
276
those visiting an emergency unit. Our study showed that MUST was found to have a greater K
277
value than NSR2002, which shows that patients found at high risk with MUST were in greater
278
proportion identified as malnourished, according to the diagnostic criteria of ESPEN, when
279
compared to NRS2002. An alternative reading of this is that those are malnourished according to
280
the ESPEN criteria are better detected with the use of MUST. Moreover, MUST was also found
281
to have greater sensitivity and specificity compared to NRS2002. MUST has been proven to be a
282
valid tool for the identification of nutritional risk in specific patients’ population. According to a
283
systematic review for the identification of malnourishment in colorectal cancer patients, MUST
284
was found to have a high sensitivity and specificity and excellent diagnostic accuracy [19].
TE D
EP
AC C
285
M AN U
273
Available published data regarding the performance of the two aforementioned screening
286
tools show that NRS2002 was found to be better in predicting clinical outcome in hospitalized
287
patients compared to MUST [20] and the performance of the tools was increased when it was
288
combined with Subjective Global Assessment (SGA), a more complex tool that includes data
ACCEPTED MANUSCRIPT 16
from nutritional assessment.[21]. In another study by Velasco et al, a comparison of four tools,
290
namely NRS2002, SGA, MUST and Mini Nutritional Assessment (MNA), was performed using
291
SGA as the gold standard, in an era that diagnostic criteria for malnutrition did not exist. [22]. In
292
that study, NRS2002 and MUST were found to perform equally well, suggesting the use of those
293
tools for the screening of patients on admission to the hospital. The only study that did not use a
294
screening tool as the gold standard -since a gold standard for nutritional screening does not
295
actually exist- was a study by Poulia et al [23] in an elderly hospitalized population, confirming
296
our results for the better performance of MUST in comparison to NRS2002. In the
297
aforementioned study, the criterion of a patient being recognized as malnourished was to fulfil
298
the criteria of malnutrition in 4 out of 6 screening tools used, a method used in other studies as
299
well. Moreover, the statistical analysis and comparison of the tools was based on the extended
300
methods of triads, a method used for the analysis of measurements for which no gold standard is
301
set.
TE D
M AN U
SC
RI PT
289
The varying results of this comparison can be attributed to differences in the original
303
design of the two screening tools. In particular, NRS2002 was originally designed to screen
304
patients who would benefit from receiving nutritional support [12] and therefore its use might be
305
linked with an increased number of patients classified to be at moderate/high risk of
306
malnutrition, as a larger number of patients could indeed benefit from receiving nutritional
307
support. On the other hand, MUST is the only tool that was specifically developed for screening
308
of malnutrition [13, 14]. Therefore, its higher level of agreement may imply that those identified
309
at high risk of malnutrition are actually those who are more possibly indeed malnourished
310
according to the ESPEN criteria.
AC C
EP
302
ACCEPTED MANUSCRIPT 17
As nutritional screening tools are the first step to identify patients either as malnourished
312
or as at greater risk to become malnourished [24] and therefore to identify this problem before
313
the healthcare team schedules a treating plan, the importance of using a reliable and easy to
314
perform screening tool should be emphasized. Given the fact that there is no gold standard for
315
nutritional screening, a more specific definition of malnutrition could better serve to test whether
316
the identification at moderate/high risk of malnutrition is somehow correlated with the actual
317
nutritional status (i.e. if the patients screened as “high risk” are actually among those who fulfill
318
the newly established criteria of malnutrition by ESPEN) [11].
SC
RI PT
311
Due to the fact the both tools were found to have a very high negative predictive value in
320
both groups (96.3% for NRS2002 and 99.7% for MUST for outpatients and 93.9% for NRS2002
321
and 100% for MUST for hospitalized patients) one could safely assume that the probability for a
322
patient who was found not at risk in both tools to be malnourished is practically minimal.
M AN U
319
The strengths of our study are its multicentre setting and the large number of participants. To
324
our best knowledge, this is the first study that assesses different screening tools in a large
325
population in the light of the newly established ESPEN consensus. We acknowledge that our
326
study may have suffered from a sample selection bias as some patients who were severely sick
327
may have not joined the study as was also the case for a substantial number of patients that were
328
on nutritional support at study entry. A further potential limitation of this study is the fact that we
329
did not perform full nutritional assessment as a reference for the comparison of the screening
330
scores. Furthermore, our study evaluated the screening tools in the specific study population
331
enrolled, and extrapolation of results to other populations may be done cautiously. Finally, the
332
lack of FFMI is a limitation affecting the results of the study. According to Rojer et al [25], the
333
combination of FFMI and WL compared to BMI and WL, but yet it remains unclear if the
AC C
EP
TE D
323
ACCEPTED MANUSCRIPT 18 334
selection of the one criterion versus the other could affect the results of the classification of the
335
prevalence of malnutrition. Nutritional indices, due to their objectivity, could also be rather helpful in estimating
337
malnutrition risk on admission to the hospital [26]. MUST is an easy-to-use tool with more
338
straight forward, objective questions [14]. NRS2002, although containing similar questions with
339
MUST, relies on the decision of the medical professional regarding the severity of disease. Also,
340
it may impose a subjective bias or an added difficulty, especially for conditions not included in
341
the list of the referred clinical states and especially when the person performing the screening is
342
not a medical doctor. Therefore, MUST seems to be an extremely useful tool in settings where
343
time is limited and personnel dedicated in the nutritional screening is also rather compromised.
M AN U
SC
RI PT
336
344 345
Conclusion
To our knowledge, this study is the first to analyze the clinical value of a malnutrition
347
screening tool in the light of the new ESPEN definition for malnutrition. According to our
348
results, MUST was better correlated with ESPEN criteria for the definition of malnutrition,
349
leading us to the conclusion that it can more efficiently identify the malnourished patients,
350
during the screening process. More studies, however, are needed to fully assess all diagnostic
351
modalities.
353 354
EP
AC C
352
TE D
346
Acknowledgments
We would like to thank those who helped on site and all members of the participating wards.
355
356
Statement of authorship
ACCEPTED MANUSCRIPT 19
KAP conceived the study, wrote the manuscript, and helped with the statistical analyses.
358
SK contributed writing the manuscript and coordinated intragroup discussions. ID and EB
359
contributed to the sample collection and statistical data interpretation and helped with the
360
statistical analysis. DK and AB participated in the initial part of study design and contributed in
361
the sample collection and the data acquisition. MP participated in study’s design and contributed
362
in writing the manuscript. MC coordinated the study, the intragroup reviews and communication,
363
participated in its design, wrote the study protocol, and drafted the manuscript. KAP, SK, ID, EB
364
and MC commented on the first and subsequent drafts. All authors read and approved the final
365
manuscript.
366
367
CONFLICT OF INTEREST
M AN U
SC
RI PT
357
The authors hereby declare that the article is original, is not under consideration for
369
publication anywhere else and has not been previously published. Authors declare no potential or
370
actual personal, political or financial interest in the material, information or techniques described
371
in the paper.
AC C
EP
TE D
368
REFERENCES 1. 2. 3.
Bistrian, B., et al., Protein status of general surgical patients. J Am Med Assoc, 1974. 230: p. 85860. Hill, G., et al., Malnutrition in surgical patients. An unrecognised problem. Lancet, 1977. 1(8013): p. 689-92. White, J., et al., Academy Malnutrition Work Group; A.S.P.E.N. Malnutrition Task Force; A.S.P.E.N. Board of Directors. Consensus statement: Academy of Nutrition and Dietetics and American Society for Parenteral and Enteral Nutrition: characteristics recommended for the identification and documentation of adult malnutrition (undernutrition). JPEN J Parenter Enteral Nutr, 2012. 36(3): p. 275-83.
ACCEPTED MANUSCRIPT 20
10. 11. 12. 13.
14.
15. 16. 17. 18. 19.
20. 21.
22. 23. 24.
RI PT
9.
SC
8.
M AN U
7.
TE D
6.
EP
5.
Strong, J.P., The natural history of atherosclerosis in childhood. Ann N Y Acad Sci, 1991. 623: p. 915. Jensen, G., et al., Malnutrition syndromes: a conundrum vs continuum. . JPEN J Parenter Enteral Nutr, 2009. 33: p. 710-716. Jensen, G.L., et al., Adult starvation and disease-related malnutrition: A proposal for etiologybased diagnosis in the clinical practice setting from the International Consensus Guideline Committee. Clin Nutr, 2010. 29: p. 151-153. McWhirter, J.P. and C.R. Pennington, Incidence and recognition of malnutrition in hospital. BMJ, 1994. 308(6934): p. 945-8. Kondrup, J., et al., Incidence of nutritional risk and causes of inadequate nutritional care in hospitals. Clin Nutr, 2002. 21(6): p. 461-8. Leistra, E., et al., Prevalence of undernutrition in Dutch hospital outpatients. Eur J Intern Med, 2009. 20(5): p. 509-13. Neelemaat, F., et al., Screening malnutrition in hospital outpatients. Can the SNAQ malnutrition screening tool also be applied to this population? . Clin Nutr, 2008. 27(3): p. 439-446. Cederholm, T., et al., Diagnostic criteria for malnutrition - An ESPEN Consensus Statement. Clin Nutr, 2015. 34(3): p. 335-40. Kondrup, J., et al., ESPEN guidelines for nutrition screening 2002. Clin Nutr, 2003. 22(4): p. 41521. Stratton, R.J., et al., Malnutrition in hospital outpatients and inpatients: prevalence, concurrent validity and ease of use of the "malnutrition universal screening tool" ("MUST") for adults. British Journal of Nutrition, 2004. 92(05): p. 799-808. Elia, M., Screening for Malnutrition: A Multidisciplinary Responsibility. Development and Use of the ‘Malnutrition Universal Screening Tool’ (‘MUST’) for Adults. Malnutrition Advisory Group (MAG), a Standing Committee of BAPEN. Redditch, . Worcs.: BAPEN, 2003. Kondrup, J., et al., Nutritional risk screening (NRS 2002): a new method based on an analysis of controlled clinical trials. Clin Nutr, 2003. 22(3): p. 321-36. Guigoz, Y., S. Lauque, and B.J. Vellas, Identifying the elderly at risk for malnutrition. The Mini Nutritional Assessment. Clin Geriatr Med, 2002. 18(4): p. 737 - 757. Omran, M.L. and J.E. Morley, Assessment of protein energy malnutrition in older persons, part I: History, examination, body composition, and screening tools. Nutrition, 2000. 16(1): p. 50-63. Christensson, L., M. Unosson, and A.C. Ek, Evaluation of nutritional assessment techniques in elderly people newly admitted to municipal care. Eur J Clin Nutr, 2002. 56(9): p. 810-8. Håkonsen, S., et al., Diagnostic test accuracy of nutritional tools used to identify undernutrition in patients with colorectal cancer: a systematic review. JBI Database System Rev Implement Rep, 2015. 13(4): p. 141-87. Raslan, M., et al., Comparison of nutritional risk screening tools for predicting clinical outcomes in hospitalized patients. Nutrition, 2010. 26(7-8): p. 721-6. Raslan, M., et al., Complementarity of Subjective Global Assessment (SGA) and Nutritional Risk Screening 2002 (NRS 2002) for predicting poor clinical outcomes in hospitalized patients. Clin Nutr, 2011. 30(1): p. 49-53. Velasco, C., et al., Comparison of four nutritional screening tools to detect nutritional risk in hospitalized patients: a multicentre study. Eur J Clin Nutr, 2011. 65(2): p. 269-74. Poulia, K., et al., Evaluation of the efficacy of six nutritional screening tools to predict malnutrition in the elderly. Clin Nutr, 2012 31(3): p. 378-85. Blackburn, G. and P. Thornton, Nutritional assessment of the hospitalized patient. Med. Clin. North Am, 1979. 63: p. 1103-1115.
AC C
4.
ACCEPTED MANUSCRIPT 21
RI PT SC M AN U TE D EP
26.
Rojer, A.G., et al., The prevalence of malnutrition according to the new ESPEN definition in four diverse populations. Clin Nutr, 2016. 35(3): p. 758-62. Schneider, S. and X. Hebuterne, Use of nutritional scores to predict clinical outcomes in chronic diseases. . Nutrition reviews 2000. 58: p. 31-38.
AC C
25.
ACCEPTED MANUSCRIPT 22
Supplemental Files
RI PT
Supplementary Table 1: Calculation model for sensitivity, specificity and predictive values
Malnourished (ESPEN Criteria)
Not malnourished (ESPEN Criteria)
A
B
SC
Moderate/ High nutritional risk
C
D
M AN U
Not at moderate/ high nutritional risk
Sensitivity, specificity, predictive values for the two nutrition screening tools were calculated by the use of diagnostic criteria as the way to identify the malnourished patients, as follows: Sensitivity= A/A+C; Specificity= D/B+D; Positive predictive value= A/A+B; Negative predictive value = D/C+D,
AC C
EP
TE D
Where, A: high risk of malnutrition as resulted from the nutritional tool and malnourished according to the diagnostic criteria, B: high risk of malnutrition as resulted from the nutritional screening tool but not malnourished according to the diagnostic criteria, C: malnourished only from the diagnostic criteria but not screened at high risk of malnutrition from the nutritional screening tool and D: not malnourished according to the diagnostic criteria and not screened at high risk of malnutrition Positive likelihood ratio was calculated from sensitivity and specificity as follows: Positive Likelihood ratio (LR+) = Sensitivity/1-specificity and Negative Likelihood ratio (LR-) as follows: Negative Likelihood ratio (LR -) = 1-sensitivity/specificity
S0261-5614(16)30180-7
DOI:
10.1016/j.clnu.2016.07.014
Reference:
YCLNU 2881
To appear in:
Clinical Nutrition
Received Date: 18 January 2016 Revised Date:
14 July 2016
Accepted Date: 25 July 2016
Please cite this article as: Poulia K-A, Klek S, Doundoulakis I, Bouras E, Karayiannis D, Baschali A, Passakiotou M, Chourdakis M, The two most popular malnutrition screening tools in the light of the new ESPEN consensus definition of the diagnostic criteria for malnutrition, Clinical Nutrition (2016), doi: 10.1016/j.clnu.2016.07.014. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT 1
1
Title: The two most popular malnutrition screening tools in the light of the new ESPEN
2
consensus definition of the diagnostic criteria for malnutrition
3
Authors: Kalliopi-Anna Poulia1, Stanislaw Klek2, Ioannis Doundoulakis3, Emmanouil Bouras3,
5
Dimitrios Karayiannis4, Aristea Baschali4, Marili Passakiotou5, Michael Chourdakis3 a
6
RI PT
4
1
Department of Nutrition and Dietetics, Laiko General Hospital, Athens, Greece
8
2
Stanley Dudrick's Memorial Hospital, Skawina, Poland
9
3
School of Medicine, Aristotle University of Thessaloniki, Thessaloniki, Greece
10
4
Department of Nutrition and Dietetics, Evangelismos Hospital, Athens, Greece
11
5
ICU, Ippokratio General Hospital, Thessaloniki, Greece
M AN U
SC
7
12 13
Names for Pubmed indexing: Poulia, Klek, Doundoulakis, Bouras, Karayiannis, Baschali,
14
Passakiotou, Chourdakis
15
Part of the data previously presented at the 36th ESPEN Congress (2015)
TE D
16 17 18
a
19
Michael Chourdakis, MD RD MPH PhD
20
Aristotle University of Thessaloniki, Greece
21
Department of Medicine
22
University Campus, 54124
23
Phone: +30 2310 999035, Fax: +30 2312 205270
24
Email: [email protected]
EP
AC C
25
Correspondence:
26
Abbreviations:
27
ASPEN: American Society for Enteral and Parenteral Nutrition; AUC: area under the (ROC)
28
curve;, BMI: body mass index; BAPEN: British Society of Enteral and Parenteral Nutrition;
29
DRM: disease related malnutrition; ESPEN: European Society for Clinical Nutrition and
ACCEPTED MANUSCRIPT 2
Metabolism; IQR: interquartile ranges; K: Cohen’s Kappa value; LOS: length of hospital stay;
31
LR(+): positive likelihood ratio; LR(-): negative likelihood ratio; MUST: Malnutrition
32
Universal Screening Tool; NPV: negative predicting value; NRS2002: Nutritional Risk
33
Screening 2002; PPV: positive predictive value; ROC: receiver operating characteristic; SD:
34
standard deviation;
RI PT
30
AC C
EP
TE D
M AN U
SC
35
ACCEPTED MANUSCRIPT 3
Abstract
37
Background: The new definition of malnutrition in adults proposed recently by The European
38
Society for Clinical Nutrition and Metabolism (ESPEN) changed the view on the issue and raised
39
the question of the reliability of available diagnostic tools. Therefore, the aim of this study was to
40
verify the accuracy of the two most commonly used screening tools by comparing their findings
41
with the new ESPEN criteria.
42
Methods: Nutritional screening was performed in 1146 (median age 60 years, interquartile
43
range: 44–73 years, 617 males, 529 females) patients on admission to hospitals with two
44
nutritional screening tools: Nutritional Risk Screening 2002 (NRS2002) and Malnutrition
45
Universal Screening Tool (MUST). The screening results were then compared to the ESPEN
46
new diagnostic criteria for malnutrition.
47
Results: According to the NRS2002 13.5% and 27.9% of the outpatients and hospitalized
48
patients respectively were found to be at moderate/high risk of malnutrition. With the use of
49
MUST 9.1% and 14.9% of the outpatients and hospitalized patients respectively were found to
50
be at moderate/high risk of malnutrition. According to the ESPEN diagnostic criteria 6.4% and
51
11.3% of outpatients and hospitalized patients respectively were classified as malnourished.
52
MUST was found to be better correlated to the latter for both outpatients (K=0.777, p<0.001)
53
and hospitalized patients (K=0.843, p<0.001) as compared to NRS2002 (k=0.256, p<0.001 and
54
k=0.228, p<0.001). ROC plots Area Under the Curve (AUC) was found to be higher for MUST
55
compared to NRS2002 (0.964 vs. 0.695 for outpatients and 0.980 vs 0.686 for hospitalized
56
patients respectively).
57
Conclusion: To our knowledge, this study is the first to analyze the clinical value of a
58
malnutrition screening tool in the light of the new ESPEN definition for malnutrition. According
AC C
EP
TE D
M AN U
SC
RI PT
36
ACCEPTED MANUSCRIPT 4 59
to our results, MUST was better correlated with ESPEN criteria for the definition of
60
malnutrition, leading us to the conclusion that it can more efficiently identify the malnourished
61
patients, during the screening process.
RI PT
62
Keywords
64
Disease related malnutrition, nutritional screening, screening tools, diagnostic criteria, ESPEN
65
malnutrition definition
AC C
EP
TE D
M AN U
SC
63
ACCEPTED MANUSCRIPT 5
66
Introduction Malnutrition, also called disease-related malnutrition (DRM), can be the result of poor
68
nutritional intake, impaired utilization or loss of macro- and/ or micronutrients, or may stem
69
from acute or chronic inflammation [1-3]. It is clear that nutritional status may deteriorate during
70
hospital stay, leading to malnutrition, which occurs in 20-60% of hospitalized patients, but also
71
affects outpatients in lower percentages (7-16%)[1-10]. This very often passes unnoticed,
72
ignored or untreated despite the fact that it has been recognized as a major public health concern
73
leading to a significant economic burden. As a consequence, increased morbidity, mortality,
74
hospital readmissions and length of hospital stay have been observed [3-5, 7].
M AN U
SC
RI PT
67
The first step to successfully treat malnutrition is the appropriate diagnosis. There are two
76
major difficulties linked to that issue: the first one is the definition of malnutrition and the second
77
is choosing an appropriate screening tool to assess the risk. Quite recently the European Society
78
for Clinical Nutrition and Metabolism (ESPEN) published consensus-based new criteria for
79
malnutrition, which were meant to be applied independently of clinical setting and aetiology [8,
80
11] when screening for malnutrition gives a positive result. Few years earlier, a similar task was
81
undertaken be the American Society for Enteral and Parenteral Nutrition (ASPEN) [3].
EP
TE D
75
As defining malnutrition according to the ESPEN criteria has as prerequisite a positive
83
screening result from a screening tool, we have to keep in mind that screening for malnutrition is
84
not a simple task either. Over the last decades several screening tools have been introduced and
85
evaluated. Most of them combine almost the same variables, such as percentage of weight loss
86
during a defined period of time, body mass index (BMI), reduction of food intake, and the
87
presence of disease and its severity [8]. The Nutritional Risk Screening 2002 (NRS-2002) is the
88
tool proposed by the ESPEN guidelines mainly for detection of indications for nutritional
AC C
82
ACCEPTED MANUSCRIPT 6 89
support [12]. Another popular screening tool is the Malnutrition Universal Screening Tool
90
(MUST), which was developed to detect the risk of malnutrition for all adult patients across all
91
health care settings [13, 14]. Despite the widespread use of these screening tools, the new ESPEN definition for
93
malnutrition changes the screening strategy as it gives a base upon which the results of the
94
screening process could be evaluated. At the same time, however, it raises a question of the
95
validity of the former screening modalities. As the two afore mentioned tools are designed to
96
screen malnutrition -anyhow examining it from another point of view- we decided to assess
97
whether “high risk for malnutrition” or “high nutritional risk” as identified by the two screening
98
tools actually corresponds to the prevalence of malnutrition according to the ESPEN criteria.
99
Therefore, the aim of current study was to assess the tools used in clinical practise to screen
SC
M AN U
100
RI PT
92
patient for malnutrition in the light of the new ESPEN consensus definition of malnutrition.
102
Materials and methods
103
Subjects
TE D
101
This multicenter study enrolled one thousand one hundred forty-six patients [median age
105
60 years (interquartile range-IQR: 44–73 years), 617 males, 529 females] on admission to 19
106
hospitals in 11 Greek cities, from November 2014 until April 2015. Adult patients (≥18 years
107
old) admitted to internal and surgery wards/outpatients units (50.9%, 583 internal medicine and
108
49.1%, 563 surgical patients) were eligible to participate. They were consecutively invited to
109
participate whenever data collection was possible within the first 24 hours after admission and/or
110
during their examination waiting time at the outpatient’s units or emergency unit. Demographic
111
and medical data along with the two questionnaires for nutritional risk were collected during a
AC C
EP
104
ACCEPTED MANUSCRIPT 7 112
structured interview. Patients unable to communicate with the study’s personnel were excluded.
113
The population of the study group was homogenous, Caucasians, Greek or originated for Balkan
114
and other European countries. Participants were informed about the aim of the study and written consent was obtained
116
by all participants. The study was approved by the Medical Research Ethics Committee of the
117
Medical School of the Aristotle University of Thessaloniki.
119
SC
118
RI PT
115
Anthropometric measurements
Anthropometric measurements were performed with the subjects wearing light clothing,
121
without shoes. Body weight and height were measured at the time of recruitment, i.e. on
122
admission, using a calibrated weighting scale and a wall-mounted stadiometer to the nearest 0.5
123
kg and 0.5 cm, respectively. If height measurement was not feasible (e.g. in the case of an unable
124
to stand patient), self-reported height and weight were used (if reliable and realistic), from the
125
patient or the patient’s caregiver. BMI was computed as weight (in kilograms) divided by the
126
square of height (in meters squared). Percentage of unintentional weight loss over the last three
127
months was calculated from patients’ reports.
130
TE D
EP
129
Assessment of the nutritional risk
AC C
128
M AN U
120
For each patient, the steps of each tool were completed by the same investigator in the
131
same order. The total score for each screening tool was computed during the analysis of the data.
132
The assessors were encouraged not to add the scores for each tool during data collection in order
133
to avoid bias by the knowledge of the categorization in a screening tool.
ACCEPTED MANUSCRIPT 8
MUST uses current BMI, unintentional weight loss and the presence of any acute disease
135
effect that could compromise nutritional intake for >5 days [13]. It includes three parameters
136
rating them as 0, 1 or 2 as follows: BMI>20 kg/m2=0; 18.5–20.0 kg/m2=1; <18.5 kg/m2=2;
137
unintentional weight loss in the past 3-6 months <5%=0; 5–10%=1; >10%=2; acute disease:
138
absent=0; present=2. Overall risk of malnutrition is established after addition of all points
139
allocated, as follows: 0=low risk; 1=medium risk; 2=high risk [14]. It was primarily designed for
140
the use in the community but it has been validated in all settings, including hospital.
SC
RI PT
134
NRS2002, the nutritional screening tool proposed by the ESPEN guidelines for the
142
nutritional screening of patients [12, 15] includes an initial screening including four questions
143
about the following parameters: BMI (if it is <20.5), presence of weight loss in the past three
144
months, presence of low dietary intake in the past week and the severity of disease. In the case of
145
any positive response a final screening is required [12]. The final screening evaluates impaired
146
nutritional status (Score=0 for normal nutritional status, Score=1 for weight loss >5% in the last
147
3 months or food intake below 50-75% of the normal requirements, Score=2 for weight loss >5%
148
in the last two months or BMI 18.5-20.5 + impaired general condition or food intake 25-60% of
149
normal requirements and Score=3 for weight loss >5% in the last month or BMI <18.5+impaired
150
general condition or Food intake 0-25% of normal requirement) and the “severity of disease
151
score” of 0-3 plus an additional point if the patient is older than 70 years. Nutritional risk is
152
established after addition of all points allocated, as follows: <3 at no or low risk, ≥3= at high
153
risk.
TE D
EP
AC C
154
M AN U
141
We retrospectively used the new European Society for Clinical Nutrition and Metabolism
155
(ESPEN) criteria for the definition of malnutrition [11], namely BMI≤18.5 kg/m2, or weight loss
156
>10% (indefinite of time) or 5% (in 3 months) and BMI <20 kg/m2 for patients <70 years or <22
ACCEPTED MANUSCRIPT 9
kg/m2 for patients ≥70 years, but not Fat Free Mass Index-FFMI (below <15 and <17 kg/m²) to
158
identify malnourished patients in our study’s population and evaluate the correlation of these
159
criteria with the prognostic value of the two screening tools. As the analysis was performed
160
retrospectively, no data regarding FFMI in our population were available; therefore, the
161
categorization of malnutrition according to the ESPEN criteria, was based only in the BMI and
162
weight loss. As both options are given in the consensus it can be assumed that by choosing one
163
versus the other, the results of our evaluation would not be biased. The aforementioned analysis
164
was performed by an investigator from our group who was neither involved in the initial
165
screening tools analysis nor aware of its results, for blinding purposes.
M AN U
SC
RI PT
157
The total score and classification of malnutrition risk (low, medium or high) was
167
determined for each study participant and screening tool. The scores obtained by the two
168
screening tools were then related to the ESPEN newly established consensus definition of
169
malnutrition.
TE D
166
For the cross-tabulation of risk classification between the tools we decided to group the
171
classification of malnutrition risk into two categories (i.e. “high/moderate” vs. “low/no risk”) as
172
patients allocated in the high group category are the ones that need to be further referred for
173
assessment to the dietetic and medical team.
175 176
AC C
174
EP
170
Statistical analysis
Normally distributed continuous variables were presented as mean values and standard
177
deviations (SD), skewed variables were presented as medians and IQR, whereas categorical
178
variables as absolute and relative (%) frequencies. The Kolmogorov-Smirnov test was applied to
179
evaluate normality of the distributions. All reported P-values are based on two-sided tests and
ACCEPTED MANUSCRIPT 10
compared to a significance level of 5%. Cohen’s kappa (Κ) statistic was calculated to determine
181
diagnostic concordance between the assessment tools (i.e. MUST, NRS2002) and ESPEN’s
182
diagnostic criteria for malnutrition. K coefficient is a statistical measure of inter–annotator
183
agreement for qualitative variables. In case of complete agreement between the variables K=1,
184
whereas if there is no agreement among the variables measured (other than what would be
185
expected by chance) then K≤0. Positive and negative likelihood ratios (LR+ and LR-
186
respectively) were calculated for both tools. For diagnostic procedures similar to nutritional
187
screening with the tools that were used in the study, it is desired LR+ to be >10 and LR-<0.1.
SC
RI PT
180
Sensitivity, specificity, predictive values for the two nutrition screening tools were
189
calculated with the use of diagnostic criteria as the way to identify the malnourished patients, as
190
can be seen in Supplementary Table 1.
M AN U
188
Receiver operating characteristic (ROC) curves were also used to evaluate the ability of
192
NRS2002 and MUST to correctly distinguish between the malnourished and the non-
193
malnourished patients (according to the ESPEN consensus definition of malnutrition). Area
194
under the ROC curve equal to 0.5 indicates that a tool cannot distinguish between the two
195
groups, whereas area under the ROC curve equal to 1 indicates perfect separation of the values
196
of the two groups.
198 199
EP
Statistical analysis was performed using SPSS for Windows, version 16.0 (SPSS Inc,
AC C
197
TE D
191
Chicago, IL).
ACCEPTED MANUSCRIPT 11
200
Results The baseline characteristics of the sample are presented in Table 1. Median BMI was
202
26.5 kg/m2 (IQR: 24.1–29.4 kg/m2) and median hospital stay among the hospitalized patients
203
was 3 days (IQR: 2–7 days).
RI PT
201
204 205
Table 1: Characteristics of the sample of patients
(n=784)
185 (51.1%)
617 (53.8%)
177 (48.9%)
529 (46.2%)
Age§ (years)
57 (36–70)
62 (46–75)
60 (44–73)
BMI§ (kg/m2)
26.4 (24.9–29.8)
26.8 (23.9–30.5)
26.5 ( 24.1–29.4)
BMI (<18,5 kg/m2)*
12 (1.5%)
10 (2.8%)
22 (1.9%)
Recent weight loss (>5% in 3 months)*
71 (9.1%)
54 (14.9%)
125 (10.9%)
–
3 (2–7)
–
M AN U
352 (44.9%)
TE D
207
§
Values expressed as medians and interquartile range (IQR)
EP
208
The classification of the risk of malnutrition of the patients by the two screening tools showed a variation among the two tools and the ESPEN criteria as can be seen in Table 2.
AC C
212
(n=1146)
Female*
*Values expressed as frequencies and percentages
211
(n=362)
432 (55.1%)
206
210
Overall
Male*
Length of stay in hospital§ (days)
209
Hospitalized
SC
Outpatients
Table 2: Classification of the risk of malnutrition by the two screening tools and the ESPEN criteria
Risk of malnutrition Moderate/high Low
Outpatients (n=784)
MUST
NRS2002
9.1% (71/784) 90.9%
13.5% (106/784) 86.5%
Hospitalized (n=362) ESPEN Criteria* 6.4% (50/784) 93.6%
MUST
NRS2002
14.9% (54/362) 85.1%
27.9% (101/362) 72.1%
ESPEN Criteria* 11.3% (41/362) 88.7%
ACCEPTED MANUSCRIPT 12 (713/784) 213
(678/784)
(734/784)
(308/362) (261/362)
(321/362)
* Classification of malnutrition according to the ESPEN consensus definition of malnutrition Cross tabulation of the results of screening of nutritional risk with MUST and NRS2002
215
and the classification of malnutrition according to the ESPEN diagnostic criteria can be found in
216
Table 3.
RI PT
214
217
Table 3: Cross tabulation of the results of screening of nutritional risk with MUST (≥1) and NRS2002
219
(≥3) and the classification of malnutrition according to the ESPEN consensus definition of malnutrition
SC
218
M AN U
Outpatients (n=784)
Hospitalized (n=362)
Not at moderate/high nutritional risk
Malnourished
Not malnourished
Malnourished
Not malnourished
(N=50)
(N=734)
(N=41)
(N=321)
48
23
41
13
2
711
0
308
TE D
220
At moderate/high nutritional risk
ESPEN Criteria
At moderate/high nutritional risk
25
81
25
76
Not at moderate/high nutritional risk
25
653
16
245
EP
NRS2002
MUST
ESPEN Criteria
The analysis of the agreement between the diagnostic criteria of malnutrition and the
222
results of nutritional screening was found to be different, according to the tool used for the
223
screening. More specifically, better agreement with the diagnostic criteria of malnutrition was
224
found in MUST for hospitalized [96.4% (349/362) of the cases (K=0.843, p<0.001)] and
225
outpatients [96.8% (759/784) of the cases (K=0.777, p<0.001)]. NRS2002 was found to be in
226
lower agreement in both outpatients [86.5% (678/784) of the cases (K=0.256, p<0.001)] and
227
hospitalized patients [74.6% (270/362) of the cases (K=0.228, p<0.001)]. MUST was also found
228
to have a very high sensitivity and specificity for both inpatients and outpatients (MUST
AC C
221
ACCEPTED MANUSCRIPT 13
Sensitivity=96.0% for outpatients and 100% for hospitalized respectively and MUST
230
Specificity=96.0% for both subgroups vs NRS-2002 Sensitivity = 50.0% in the outpatients and
231
61.0% in the hospitalized respectively.), showing that the likelihood that a patient screened
232
positive for malnutrition risk with MUST to be actually malnourished according to the ESPEN
233
diagnostic criteria was higher, compared to NRS 2002, with which according to our results, 50%
234
and 39% respectively of malnourished patients according to ESPEN criteria were not identified
235
as at risk of malnutrition.
SC
RI PT
229
On the other hand, NRS2002 was found to have a lower sensitivity (50% for outpatients
237
and 61% for hospitalized) and specificity (89.0% for outpatients vs. 76.3% for hospitalized), and
238
in combination with the low positive predictive value (23.6% for outpatients and 24.8% for
239
inpatients) it was found that the likelihood of a false positive malnutrition screening result was
240
higher for patients screened with NRS2002.
M AN U
236
Furthermore, MUST was found to have higher LR+ and lower LR- compared to
242
NRS2002 for both groups confirming the better performance of MUST regarding the agreement
243
with the diagnostic criteria of ESPEN for malnutrition. More specifically, NRS2002 did not
244
perform well with only a fair agreement with the ESPEN criteria and low positive likelihood
245
ratio (LR+).
EP
Finally, the area under the curve as it was calculated by the ROC curves was also higher
AC C
246
TE D
241
247
in MUST compared to NRS2002 (0.964 vs. 0.695 for outpatients and 0.980 vs 0.686 for
248
hospitalized patients respectively), confirming the better ability of MUST to distinguish a
249
malnourished patient compared to NRS2002. Results are presented in details in Table 4.
250
ACCEPTED MANUSCRIPT 14 251
Table 4: Statistical evaluation of the nutritional screening tools compared to the diagnostic criteria of
252
malnutrition (MUST (≥1) and NRS2002 (≥3). Outpatients
Hospitalized
(n=784)
(n=362) MUST
NRS2002
50.0
96.0
Specificity (%)
89.0
96.9
Positive predictive value (%)
23.6
67.6
Negative predictive value (%)
96.3
99.7
Positive Likelihood ratio (LR+)
4.53
30.64
2.58
24.69
Negative likelihood ratio (LR-)
0.56
0.04
0.51
-
0.695
AUC
100.0
76.3
96.0
24.8
75.9
93.9
100.0
0.777 (<0.001) 0.228 (<0.001) 0.843 (<0.001)
M AN U
0.256 (<0.001)
61.0
SC
Sensitivity (%)
Κ value (p)
0.964
0.686
253
Κ value derived from Cohen kappa statistics.
254
MUST: Malnutrition Universal Screening Tool. NRS2002: Nutritional Risk Screening 2002,
255
AUC=Area under the curve from ROC plots
0.980
TE D
256 257
MUST
RI PT
NRS2002
Discussion
The purpose of using a screening tools is to identify patients at risk of malnutrition and to
259
select those individuals, who are in need for further evaluation and potential intervention.
260
According to the NRS2002 13.5% and 27.9% of the outpatients and hospitalized patients
261
respectively were found to be at moderate/high risk of malnutrition. With the use of MUST 9.1%
262
and 14.9% of the outpatients and hospitalized patients respectively were found to be at
263
moderate/high risk of malnutrition. Other studies evaluating the risk of malnutrition, showed
264
results ranging from 20-60% or even higher, reaching 80% when the population of patients is
265
elderly [2, 8, 9, 16-18].
AC C
EP
258
ACCEPTED MANUSCRIPT 15
Finally, 6.4% and 11.3% of outpatients and hospitalized patients among our study
267
population were classified as malnourished according to the ESPEN diagnostic criteria. This is
268
the first study to show the prevalence of malnutrition in a Greek outpatients and hospitalized
269
patients in a multicenter study, from several areas in Greece, and at the same time it is also
270
interesting to see the difference of the prevalence of malnutrition in outpatients and hospitalized
271
patients, putting emphasis on the need of more actions needed for the early identification of
272
malnutrition during hospital stay.
SC
RI PT
266
The wide range of the assessed prevalence of malnutrition or the risk of malnutrition can
274
be attributed to the differences in the nutritional screening tools and/or differences in the
275
population included. The outpatients’ group included free living subjects, either outpatients or
276
those visiting an emergency unit. Our study showed that MUST was found to have a greater K
277
value than NSR2002, which shows that patients found at high risk with MUST were in greater
278
proportion identified as malnourished, according to the diagnostic criteria of ESPEN, when
279
compared to NRS2002. An alternative reading of this is that those are malnourished according to
280
the ESPEN criteria are better detected with the use of MUST. Moreover, MUST was also found
281
to have greater sensitivity and specificity compared to NRS2002. MUST has been proven to be a
282
valid tool for the identification of nutritional risk in specific patients’ population. According to a
283
systematic review for the identification of malnourishment in colorectal cancer patients, MUST
284
was found to have a high sensitivity and specificity and excellent diagnostic accuracy [19].
TE D
EP
AC C
285
M AN U
273
Available published data regarding the performance of the two aforementioned screening
286
tools show that NRS2002 was found to be better in predicting clinical outcome in hospitalized
287
patients compared to MUST [20] and the performance of the tools was increased when it was
288
combined with Subjective Global Assessment (SGA), a more complex tool that includes data
ACCEPTED MANUSCRIPT 16
from nutritional assessment.[21]. In another study by Velasco et al, a comparison of four tools,
290
namely NRS2002, SGA, MUST and Mini Nutritional Assessment (MNA), was performed using
291
SGA as the gold standard, in an era that diagnostic criteria for malnutrition did not exist. [22]. In
292
that study, NRS2002 and MUST were found to perform equally well, suggesting the use of those
293
tools for the screening of patients on admission to the hospital. The only study that did not use a
294
screening tool as the gold standard -since a gold standard for nutritional screening does not
295
actually exist- was a study by Poulia et al [23] in an elderly hospitalized population, confirming
296
our results for the better performance of MUST in comparison to NRS2002. In the
297
aforementioned study, the criterion of a patient being recognized as malnourished was to fulfil
298
the criteria of malnutrition in 4 out of 6 screening tools used, a method used in other studies as
299
well. Moreover, the statistical analysis and comparison of the tools was based on the extended
300
methods of triads, a method used for the analysis of measurements for which no gold standard is
301
set.
TE D
M AN U
SC
RI PT
289
The varying results of this comparison can be attributed to differences in the original
303
design of the two screening tools. In particular, NRS2002 was originally designed to screen
304
patients who would benefit from receiving nutritional support [12] and therefore its use might be
305
linked with an increased number of patients classified to be at moderate/high risk of
306
malnutrition, as a larger number of patients could indeed benefit from receiving nutritional
307
support. On the other hand, MUST is the only tool that was specifically developed for screening
308
of malnutrition [13, 14]. Therefore, its higher level of agreement may imply that those identified
309
at high risk of malnutrition are actually those who are more possibly indeed malnourished
310
according to the ESPEN criteria.
AC C
EP
302
ACCEPTED MANUSCRIPT 17
As nutritional screening tools are the first step to identify patients either as malnourished
312
or as at greater risk to become malnourished [24] and therefore to identify this problem before
313
the healthcare team schedules a treating plan, the importance of using a reliable and easy to
314
perform screening tool should be emphasized. Given the fact that there is no gold standard for
315
nutritional screening, a more specific definition of malnutrition could better serve to test whether
316
the identification at moderate/high risk of malnutrition is somehow correlated with the actual
317
nutritional status (i.e. if the patients screened as “high risk” are actually among those who fulfill
318
the newly established criteria of malnutrition by ESPEN) [11].
SC
RI PT
311
Due to the fact the both tools were found to have a very high negative predictive value in
320
both groups (96.3% for NRS2002 and 99.7% for MUST for outpatients and 93.9% for NRS2002
321
and 100% for MUST for hospitalized patients) one could safely assume that the probability for a
322
patient who was found not at risk in both tools to be malnourished is practically minimal.
M AN U
319
The strengths of our study are its multicentre setting and the large number of participants. To
324
our best knowledge, this is the first study that assesses different screening tools in a large
325
population in the light of the newly established ESPEN consensus. We acknowledge that our
326
study may have suffered from a sample selection bias as some patients who were severely sick
327
may have not joined the study as was also the case for a substantial number of patients that were
328
on nutritional support at study entry. A further potential limitation of this study is the fact that we
329
did not perform full nutritional assessment as a reference for the comparison of the screening
330
scores. Furthermore, our study evaluated the screening tools in the specific study population
331
enrolled, and extrapolation of results to other populations may be done cautiously. Finally, the
332
lack of FFMI is a limitation affecting the results of the study. According to Rojer et al [25], the
333
combination of FFMI and WL compared to BMI and WL, but yet it remains unclear if the
AC C
EP
TE D
323
ACCEPTED MANUSCRIPT 18 334
selection of the one criterion versus the other could affect the results of the classification of the
335
prevalence of malnutrition. Nutritional indices, due to their objectivity, could also be rather helpful in estimating
337
malnutrition risk on admission to the hospital [26]. MUST is an easy-to-use tool with more
338
straight forward, objective questions [14]. NRS2002, although containing similar questions with
339
MUST, relies on the decision of the medical professional regarding the severity of disease. Also,
340
it may impose a subjective bias or an added difficulty, especially for conditions not included in
341
the list of the referred clinical states and especially when the person performing the screening is
342
not a medical doctor. Therefore, MUST seems to be an extremely useful tool in settings where
343
time is limited and personnel dedicated in the nutritional screening is also rather compromised.
M AN U
SC
RI PT
336
344 345
Conclusion
To our knowledge, this study is the first to analyze the clinical value of a malnutrition
347
screening tool in the light of the new ESPEN definition for malnutrition. According to our
348
results, MUST was better correlated with ESPEN criteria for the definition of malnutrition,
349
leading us to the conclusion that it can more efficiently identify the malnourished patients,
350
during the screening process. More studies, however, are needed to fully assess all diagnostic
351
modalities.
353 354
EP
AC C
352
TE D
346
Acknowledgments
We would like to thank those who helped on site and all members of the participating wards.
355
356
Statement of authorship
ACCEPTED MANUSCRIPT 19
KAP conceived the study, wrote the manuscript, and helped with the statistical analyses.
358
SK contributed writing the manuscript and coordinated intragroup discussions. ID and EB
359
contributed to the sample collection and statistical data interpretation and helped with the
360
statistical analysis. DK and AB participated in the initial part of study design and contributed in
361
the sample collection and the data acquisition. MP participated in study’s design and contributed
362
in writing the manuscript. MC coordinated the study, the intragroup reviews and communication,
363
participated in its design, wrote the study protocol, and drafted the manuscript. KAP, SK, ID, EB
364
and MC commented on the first and subsequent drafts. All authors read and approved the final
365
manuscript.
366
367
CONFLICT OF INTEREST
M AN U
SC
RI PT
357
The authors hereby declare that the article is original, is not under consideration for
369
publication anywhere else and has not been previously published. Authors declare no potential or
370
actual personal, political or financial interest in the material, information or techniques described
371
in the paper.
AC C
EP
TE D
368
REFERENCES 1. 2. 3.
Bistrian, B., et al., Protein status of general surgical patients. J Am Med Assoc, 1974. 230: p. 85860. Hill, G., et al., Malnutrition in surgical patients. An unrecognised problem. Lancet, 1977. 1(8013): p. 689-92. White, J., et al., Academy Malnutrition Work Group; A.S.P.E.N. Malnutrition Task Force; A.S.P.E.N. Board of Directors. Consensus statement: Academy of Nutrition and Dietetics and American Society for Parenteral and Enteral Nutrition: characteristics recommended for the identification and documentation of adult malnutrition (undernutrition). JPEN J Parenter Enteral Nutr, 2012. 36(3): p. 275-83.
ACCEPTED MANUSCRIPT 20
10. 11. 12. 13.
14.
15. 16. 17. 18. 19.
20. 21.
22. 23. 24.
RI PT
9.
SC
8.
M AN U
7.
TE D
6.
EP
5.
Strong, J.P., The natural history of atherosclerosis in childhood. Ann N Y Acad Sci, 1991. 623: p. 915. Jensen, G., et al., Malnutrition syndromes: a conundrum vs continuum. . JPEN J Parenter Enteral Nutr, 2009. 33: p. 710-716. Jensen, G.L., et al., Adult starvation and disease-related malnutrition: A proposal for etiologybased diagnosis in the clinical practice setting from the International Consensus Guideline Committee. Clin Nutr, 2010. 29: p. 151-153. McWhirter, J.P. and C.R. Pennington, Incidence and recognition of malnutrition in hospital. BMJ, 1994. 308(6934): p. 945-8. Kondrup, J., et al., Incidence of nutritional risk and causes of inadequate nutritional care in hospitals. Clin Nutr, 2002. 21(6): p. 461-8. Leistra, E., et al., Prevalence of undernutrition in Dutch hospital outpatients. Eur J Intern Med, 2009. 20(5): p. 509-13. Neelemaat, F., et al., Screening malnutrition in hospital outpatients. Can the SNAQ malnutrition screening tool also be applied to this population? . Clin Nutr, 2008. 27(3): p. 439-446. Cederholm, T., et al., Diagnostic criteria for malnutrition - An ESPEN Consensus Statement. Clin Nutr, 2015. 34(3): p. 335-40. Kondrup, J., et al., ESPEN guidelines for nutrition screening 2002. Clin Nutr, 2003. 22(4): p. 41521. Stratton, R.J., et al., Malnutrition in hospital outpatients and inpatients: prevalence, concurrent validity and ease of use of the "malnutrition universal screening tool" ("MUST") for adults. British Journal of Nutrition, 2004. 92(05): p. 799-808. Elia, M., Screening for Malnutrition: A Multidisciplinary Responsibility. Development and Use of the ‘Malnutrition Universal Screening Tool’ (‘MUST’) for Adults. Malnutrition Advisory Group (MAG), a Standing Committee of BAPEN. Redditch, . Worcs.: BAPEN, 2003. Kondrup, J., et al., Nutritional risk screening (NRS 2002): a new method based on an analysis of controlled clinical trials. Clin Nutr, 2003. 22(3): p. 321-36. Guigoz, Y., S. Lauque, and B.J. Vellas, Identifying the elderly at risk for malnutrition. The Mini Nutritional Assessment. Clin Geriatr Med, 2002. 18(4): p. 737 - 757. Omran, M.L. and J.E. Morley, Assessment of protein energy malnutrition in older persons, part I: History, examination, body composition, and screening tools. Nutrition, 2000. 16(1): p. 50-63. Christensson, L., M. Unosson, and A.C. Ek, Evaluation of nutritional assessment techniques in elderly people newly admitted to municipal care. Eur J Clin Nutr, 2002. 56(9): p. 810-8. Håkonsen, S., et al., Diagnostic test accuracy of nutritional tools used to identify undernutrition in patients with colorectal cancer: a systematic review. JBI Database System Rev Implement Rep, 2015. 13(4): p. 141-87. Raslan, M., et al., Comparison of nutritional risk screening tools for predicting clinical outcomes in hospitalized patients. Nutrition, 2010. 26(7-8): p. 721-6. Raslan, M., et al., Complementarity of Subjective Global Assessment (SGA) and Nutritional Risk Screening 2002 (NRS 2002) for predicting poor clinical outcomes in hospitalized patients. Clin Nutr, 2011. 30(1): p. 49-53. Velasco, C., et al., Comparison of four nutritional screening tools to detect nutritional risk in hospitalized patients: a multicentre study. Eur J Clin Nutr, 2011. 65(2): p. 269-74. Poulia, K., et al., Evaluation of the efficacy of six nutritional screening tools to predict malnutrition in the elderly. Clin Nutr, 2012 31(3): p. 378-85. Blackburn, G. and P. Thornton, Nutritional assessment of the hospitalized patient. Med. Clin. North Am, 1979. 63: p. 1103-1115.
AC C
4.
ACCEPTED MANUSCRIPT 21
RI PT SC M AN U TE D EP
26.
Rojer, A.G., et al., The prevalence of malnutrition according to the new ESPEN definition in four diverse populations. Clin Nutr, 2016. 35(3): p. 758-62. Schneider, S. and X. Hebuterne, Use of nutritional scores to predict clinical outcomes in chronic diseases. . Nutrition reviews 2000. 58: p. 31-38.
AC C
25.
ACCEPTED MANUSCRIPT 22
Supplemental Files
RI PT
Supplementary Table 1: Calculation model for sensitivity, specificity and predictive values
Malnourished (ESPEN Criteria)
Not malnourished (ESPEN Criteria)
A
B
SC
Moderate/ High nutritional risk
C
D
M AN U
Not at moderate/ high nutritional risk
Sensitivity, specificity, predictive values for the two nutrition screening tools were calculated by the use of diagnostic criteria as the way to identify the malnourished patients, as follows: Sensitivity= A/A+C; Specificity= D/B+D; Positive predictive value= A/A+B; Negative predictive value = D/C+D,
AC C
EP
TE D
Where, A: high risk of malnutrition as resulted from the nutritional tool and malnourished according to the diagnostic criteria, B: high risk of malnutrition as resulted from the nutritional screening tool but not malnourished according to the diagnostic criteria, C: malnourished only from the diagnostic criteria but not screened at high risk of malnutrition from the nutritional screening tool and D: not malnourished according to the diagnostic criteria and not screened at high risk of malnutrition Positive likelihood ratio was calculated from sensitivity and specificity as follows: Positive Likelihood ratio (LR+) = Sensitivity/1-specificity and Negative Likelihood ratio (LR-) as follows: Negative Likelihood ratio (LR -) = 1-sensitivity/specificity