Valid_food_intake_measures_of_adult_patients_for_use_within_the_GLIM.pdf
Meta-analysis
Valid food intake measures of adult patients for use within the GLIM
framework: A scoping review
R. Blaauw a,* , G. Bischoff b , M.A.E. de van der Schueren c,d , C. Compher e , K. Haines f,g ,
N. Kiss h , C.C.H. Lew i,j , A. Malone k , C. Maza l , N. Stoner m , H.H. Keller n,o
a Division Human Nutrition, Stellenbosch University, Cape Town, South Africa
b ZEP – Center for Clinical Nutrition and Prevention, Hospital Barmherzige Brueder Munich, Germany
c Department of Nutrition, Dietetics and Lifestyle, HAN University of Applied Sciences, Nijmegen, the Netherlands
d Division of Human Nutrition and Health, Wageningen University and Research, Wageningen, the Netherlands
e Department of Biobehavioral Health Science, University of Pennsylvania, School of Nursing, Philadelphia, PA, USA
f Division of Trauma, Critical Care, and Acute Care Surgery, Duke University, Durham, NC, USA
g Assistant Professor in Population Health Sciences, Duke University, Durham, NC, USA
h Institute for Physical Activity and Nutrition, Deakin University, Geelong, Australia
i Department of Dietetics and Nutrition, Ng Teng Fong General Hospital, Singapore
j Faculty of Health and Social Sciences, Singapore Institute of Technology, Singapore
k New Albany, Ohio, USA
l Division of Education and Research, Centro Medico Militar, Guatemala
m Clinical Nutrition Support, Penn Medicine at Home, University of Pennsylvania Health System, Philadelphia PA, USA
n Schlegel-UW Research Institute for Aging, Canada
o Department of Kinesiology and Health Sciences, University of Waterloo, Ontario, Canada
a r t i c l e i n f o
Article history:
Received 27 November 2025
Accepted 27 December 2025
Keywords:
Visual estimation methods
Food intake
GLIM
s u m m a r y
Background: Reduced food intake is one of the etiologic criteria proposed by the Global Leadership
Initiative on Malnutrition (GLIM). Various tools for the assessment of food intake exist, however, to
diagnose malnutrition, as in GLIM, they should be validated against established reference standards, and
quick and easy to complete by various members of the health care team.
Aim: This scoping review synthesizes the current evidence related to validated assessment measures of
food intake for adult patients in healthcare settings and reports on the application of these measures
within the GLIM diagnostic framework.
Methods: A comprehensive search strategy was performed using four bibliographic databases. To be
included, studies needed to be conducted on adults in a healthcare setting, have a dietary intake
assessment component (i.e. index method) that can be completed by various healthcare staff members,
with minimal training and in a short time period, and make comparison of this method to an estab-
lished dietary assessment reference. Studies needed to report energy and protein intake data and
provide appropriate validation statistics. Two reviewers independently reviewed all abstracts and
relevant full-text articles, and extracted data.
Results: After duplicate removal, 7866 abstracts were screened; 51 articles were eligible for full article
review, 13 articles fulfilled the inclusion criteria and one further article was obtained from grey liter-
ature, for a total of 14 articles included in the scoping review. Food weighing before and after con-
sumption was used as a reference method by most studies. For index methods, four different
measurement tools, with variations, were used. This included visual estimation methods (VEM) using a
1–10-point scale without any pictorials (2 studies); VEM based on plate-model pictures (8 studies); VEM
based on plate-models with associated defined nutritional values (4 studies) and digital technology (2
studies). Various levels of accuracy were found, with accuracy increasing when more options are pro-
vided and when employing digital technology. Index methods could be completed by participants
themselves, nurses, food service workers and dietitians. Adequate training on completion of the tools is
associated with improved results.
* Corresponding author.
E-mail address: rb@sun.ac.za (R. Blaauw).
Contents lists available at ScienceDirect
Clinical Nutrition ESPEN
journal homepage: http://www.clinicalnutritionespen.com
https://doi.org/10.1016/j.clnesp.2025.102899
2405-4577/© 2026 The Author(s). Published by Elsevier Ltd on behalf of European Society for Clinical Nutrition and Metabolism. This is an open access article under the CC
BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
Clinical Nutrition ESPEN 72 (2026) 102899
Conclusion: Index methods to assess food intake accurately for energy and protein intake can determine
inadequate intake as compared to the reference method. Yet, a further step is required to interpret food
intake relative to the patient's energy requirement to determine sufficiency of that intake for deter-
mination of the GLIM criterion. However, visual estimation methods identified in this review can be
used by diverse clinicians with confidence to determine patient food intake.
© 2026 The Author(s). Published by Elsevier Ltd on behalf of European Society for Clinical Nutrition and
Metabolism. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/
licenses/by-nc-nd/4.0/).
1. Introduction
The Global Leadership Initiative on Malnutrition (GLIM) pro-
posed a three-step approach for the diagnosis of malnutrition in
adult populations. This approach includes a screening for risk of
malnutrition component, followed by the diagnosis of malnutri-
tion by evaluating three phenotypic criteria (weight loss, low body
mass index and reduced muscle mass) and two etiologic criteria
(reduced food intake/or assimilation and the presence of inflam-
mation/disease burden). The presence of at least one phenotypic
and one etiologic criterion constitutes a malnutrition diagnosis
[1,2]. The last step involves quantifying disease severity. This
process will be imperative to feed into the new ICD-11 code for
malnutrition in adults, now called “undernutrition in adults”. This
etiology-based diagnosis classification makes allowance for
defining three subcategories, namely “Undernutrition in adults
related to disease with moderate to severe inflammation”; “Un-
dernutrition in adults related to disease with non-discernible
inflammation” and “Undernutrition in adults related to starva-
tion” [3].
A key benefit of the GLIM approach is that diagnosis of
malnutrition is streamlined so that any clinician world-wide can
determine if their patient has malnutrition. Since the original
GLIM criteria were proposed, papers were developed on the spe-
cific indicators and how to perform validity assessment [4–7].
Reduced food intake or assimilation is the only remaining criteria
that remains to be well-defined for clinical care.
In preparation for a guidance document on the diet/assimila-
tion criteria, we decided that a scoping review of available litera-
ture focusing on validated measures of assessing food intake in any
clinical setting was needed. A narrative review on visual estima-
tion methods (VEM) for assessment of food intake was published
in 2022 and provides a detailed explanation of the use of various
methods, including the general validity and applicability of these
tools in the hospital setting [8]. A scoping review on dietary
assessment methods for oral intake was published in 2023 [9]. This
review also focused only on hospitalized patients and included
assessment methods that require time and specialized skills. As
neither of these reviews included articles for clinical practice as
whole, regardless of specific setting, further identification of po-
tential brief measures that can be completed by any clinician is
needed for use with the GLIM framework. Further, recommended
methods to support guidance for the GLIM framework need to be
quick and easy to complete by various members of the healthcare
team in various healthcare systems.
To address this gap in the literature and need for GLIM guid-
ance, we planned a scoping review to answer the following
research question: “What are valid, brief measures for determi-
nation of food intake of adult patients in healthcare settings?” We
aimed to critically appraise the validated measures for food intake
of adult patients in various settings, with the objective to syn-
thesize the current evidence related to brief validated assessment
measures of food intake of adult patients in those settings and
report on the potential application of these methods within the
GLIM diagnostic framework.
2. Methods
2.1. Protocol and registration
The study protocol was approved by all authors and registered
on Open Science Framework in August 2024 (https://doi.org/10.
17605/OSF.IO/KNQB6).
2.2. Eligibility criteria
Studies were included based on the following inclusion and
exclusion criteria (Table 1). To be included, these studies needed to
be conducted on adults in a healthcare setting. They needed to
have a dietary intake assessment component consisting of less
than 20 items and one that could be completed in about 5 min by
various healthcare staff members with minimal training. The
method needed to translate the findings to energy (and protein)
content to answer the original dietary intake criteria of GLIM
at ≤ 50 % of energy requirement. For the reference test, an
established and accurate assessment of dietary intake was
required (e.g. food record), and appropriate validation statistics
needed to be reported.
2.3. Information sources
A comprehensive search strategy was used, including the
following
electronic
bibliographic
databases:
EBSCOhost
(CINAHL); Scopus; Web of Science and PubMed and covering all
publications to date. Only English language studies were included.
2.4. Search
A trained research librarian assisted with developing a search
string, based on the scoping review objectives. The search string
was adapted according to the requirements of all the databases
and based on the following key concepts: terms related to eating
or intake; intake assessment tools; and validation terminology. We
searched for relevant studies using various databases from
inception to 29 May 2024. An updated search was performed one
year later on 13 May 2025. Refer to Supplementary Table 1 for the
detailed search string used for PubMed.
2.5. Selection of sources of evidence
Search results were exported to Covidence systematic review
software [10] to assist with the process of screening and selecting
studies. Inclusion and exclusion criteria were applied to determine
the final selection of relevant articles for data extraction. The
research team was trained on Covidence and practiced identifi-
cation of abstracts (n = 25) that met inclusion criteria. All of the
R. Blaauw, G. Bischoff, M.A.E. de van der Schueren et al.
Clinical Nutrition ESPEN 72 (2026) 102899
2
The local Leadership Initiative on Malnutrition (GLIM) pro-
Conclusion: Index methods to assess food intake accurately for energy and protein intake will determine
research team was involved in screening abstracts with two re-
viewers independently reviewing the abstracts and determining if
the paper met inclusion criteria. Full-text article reviews was also
completed in duplicate by a subset of authors. Disagreements were
resolved by team discussion.
2.6. Data charting process and data items
A data-extraction form was developed and used to extract data
from each relevant study. The research team determined which
variables to extract (based on the research questions), including
author(s), year of publication, country of origin, study setting and
design, study population, methodology, index and reference
methods, details of how these methods were employed, how
outcomes were measured, relevant validity data and key findings
that relate to the review question. The first and senior author
independently extracted data and consulted in cases of
disagreement.
2.7. Synthesis of results
PRISMA Extension for Scoping Reviews (PRISMA-ScR) guide-
lines are used as reporting guidelines for this scoping review and a
PRISMA flow diagram (Fig. 1) is used to graphically depict the
process [11].
Data are summarized in table format, including general study
features (authors, publication year, country, study setting, study
design, study population and sample size (Table 2). Table 3 depicts
the information on index and reference methods, including
implementation information, with more detailed information in
Supplementary Table 2. Validity data (sensitivity and specificity,
interclass correlation and Bland Altman results) are provided in
Table 4. Lastly, all excluded studies are listed in Supplementary
Table 3.
3. Results
3.1. Selection of sources of evidence
In total 12225 articles were sourced. After duplicate removal,
7866 abstracts were screened and 51 articles were eligible for full
article review, of which 13 articles fulfilled the selection criteria.
One article was obtained from grey literature and in total 14 arti-
cles were included (Fig. 1).
3.2. Characteristics of sources of evidence
As can be seen in Table 2, the studies were published between
2013 and 2025. Investigators in France [12–14] and the
Netherlands [15–17] produced three studies each, two were from
each of Australia [18,19], and Denmark [20,21]. One study was
included from Canada [22,23], Iceland [24], Indonesia [25] and
Japan [26]. All studies were conducted on hospitalized in-patients,
with one also including outpatients [14]. The ages ranged from 15
to 96 years, with three studies focused on older adults (>65 years
of age) [12,20–22].
For the reference method, the majority of studies used food
weighing before and after consumption [12,16,18–21,23–27]. Other
reference methods included 3-day reported food record [14] and
24-hour recall [13]. One study used visual estimation of food waste
as the reference method [17,22] (Table 3 and Supplementary
Table 2).
A variety of index methods were evaluated (Table 3 and
Supplementary Table 2). Visual plate pictorials depicting four op-
tions [18,20,24] and five options [20,22,23] were most common.
One study used 11 options to depict food consumed based on the
whole tray [26]. Four studies used a visual rating of food consumed
and allocated nutritional value accordingly. These included the
Rate-a-plate concept calculating energy and protein content based
on food consumption [15]; Calorie Intake Tool (CIT) calculating
only energy value based on consumption [12]; Meal Intake Points
(MIP), which is similar to CIT, but calculating energy and protein
content [19] and Pictorial Dietary Assessment Tool (PDAT) calcu-
lating energy and protein content based on six consumption op-
tions [25]. The Self-Evaluation of Food Intake (SEFI) Tool was used
by two studies [14,17], with an earlier version of SEFI, the Ingesta-
VVAS (Visual/verbal analogue scale of food ingested) was used by
another [13]. Lastly, digital technology was used in two studies
with a digital camara image comparing before and after con-
sumption pictures with either a trained rater [16,21] or weighing
of the food [16,21].
Table 1
Scoping review inclusion and exclusion criteria.
Component
Include
Exclude
Population
Adults
Hospital/clinics
Primary care
Long-term care
Older adult community (at risk of poor food intake)
Paediatrics
Pregnancy
Prisons
Animal studies
Athletes
Healthy individuals, non-older adult
Overweight, noncommunicable
diseases e.g. diabetes
Index measurement
Variables or brief measures assessing food intake e.g. self-report of decreased intake; takes less
than 5 min to complete (e.g. < 20 items)
Minimal burden to the patient
Focused on energy and protein consumed
Assessed at first/only visit
Risk factors for low food intake
Micronutrients
Diet quality scores e.g. HEI
Food groups
Reference standard
Assessment of food intake for comparison to variable/brief measure e.g. observation, weighed food
record, photographs, food diary
Administered by any relevant staff
Lack of comparator
Outcome
Validity of variable/brief measure for assessing food intake e.g. correlation, association, agreement,
sensitivity, specificity, area under the curve
No statistics on comparison
Study type
Primary studies
Pilot studies
Letter to the editor
Case reports
Reviews
Protocols
Abstracts
R. Blaauw, G. Bischoff, M.A.E. de van der Schueren et al.
Clinical Nutrition ESPEN 72 (2026) 102899
3
3.3. Synthesis of results based on three main outcomes
3.3.1. Description of index methods used to assess dietary intake
Four types of measurement tools, with variations, were used. These
included visual estimation methods (VEM) using a 1–10-point scale,
without any pictorials; VEM based on plate-model pictures; VEM
based on plate-models with associated defined nutritional values; and
digital technology (Table 3, Fig. 2 and Supplementary Table 2).
Three studies used a VEM based on a 10-point scale concept
[13,14]. Participants had to indicate on a scale of 0 (eat nothing) to 10
(eat everything) the proportion of food they consume. Two studies
were conducted in France with the Ingesta-VVAS [13] regarded as an
earlier version of the SEFI tool [14]. In both studies, a verbal and
visual format of the tools was used. Another study combined the 10-
point scale with a 5-point plate pictorial for different meals [17]. The
VEM were completed by patients [17] and dietitians [13] in a hos-
pital setting and by patients in both an inpatient and out-patient
setting [14]. Different reference tools were used, namely 24-hour
recall [13] and 3-day food record [14], as well as visual assess-
ment of food intake based on 5-options [17].
VEM based on visual plate pictorials, depicting four options
[18,20,24]); 5 options and eleven options [26] were used. The four-
Fig. 1. PRISMA flow diagram. Graphic depiction of the process followed to select the included studies.
R. Blaauw, G. Bischoff, M.A.E. de van der Schueren et al.
Clinical Nutrition ESPEN 72 (2026) 102899
4
seven types of measurement tools, with variations, were used. These point scale with a 5-point plate pictorial for different meals [17]. The
option studies [18,20,24] asked participants to indicate intake
based on 0 %, 25 %, 50 % and 100 % of intake. All determined energy
and protein intake, and used the weighed plate waste as the
reference test. To transfer estimated intakes to nutritional value,
Palmer et al. [18] determined nutritional estimations in two ways:
by using supplier information or using a computer programmed to
provide energy and protein content for each grouped menu item,
e.g. the average for all desserts on the menu. Bjornsdottir et al. [24]
used the proportion of known energy and protein content of the
meal, and Husted et al. [20] used proportions of calculated refer-
ence food.
VEM based on visual plate pictorials, depicting five options was
used in one study from Canada [22,23] and one from Denmark
[20]. Different staff members were asked to complete the index
tests, including nursing staff [20] and food service workers [23]
after receiving training on the tools. Kawasaki et al. [26] assessed
food intake of hospitalized patients using a 11-point scale for vi-
sual estimation of whole tray consumption by nursing assistants
and dietitians, and for consumption of individual food items,
grouped by food category, by dietitians only [26].
VEM, based on visual plate pictorials with associated pre-
defined nutritional value, was used by four studies [12,15,19,25].
Dekker et al. [15] used researchers and nutrition assistants to
complete a visual estimation of foods and fluids consumed after
receiving training on the tools. Points are awarded based on pre-
determined energy and protein content of the menu items, with 1
Table 2
Included articles – General features.
Authors
Year
Country
Study setting
Study design
Population
Sample size enrolled
1
Bjornsdottir R,
Oskarsdottir ES,
Thordardottir FR,
Ramel A, Thorsdottir I,
Gunnarsdottir I [24]
2013
Iceland
Hospital
Cross-sectional
Adults, 19–94 years;
mean age 63 ± 17 years
73 patients
2
Budiningsari D, Shahar
S, Manaf ZA,
Susetyowati S. [25]
2016
Indonesia
Hospital
Cross-sectional,
Adults; mean age of
patients 44 ± 15.4
years and staff 40 ± 6.6
years
67 patients,
37 staff including
dietitians, nurses and
serving assistants
3
Dekker IM, Langius JAE,
Stelten S, de Vet HCW,
Kruizenga HM, de van
der Schueren MAE [15]
2020
Netherlands
Hospital
Validation study
Adults >55 years; mean
age 80.5 ± 10.3 years
(phase 1) and
83.2 ± 10,1 years
(phase 2)
Phase 1 = 24 - 1 patient
for 1 day, 3 patients for
2 days and 20 patients
for 3 days.
Phase 2 = 14 patients
for 2 days
4
Ghisolfi A, Dupuy C,
Gines-Farano A, Lepage
B, Vellas B, Ritz P [12]
2014
France
Hospital
Validation study
Older adults >65 years;
mean age 83.1 ± 7.2
years
100 patients
5
Guerdoux-Ninot E,
Flori N, Janiszewski C,
Vaille A, de Forges H,
Raynard B, Baracos VE,
Thezenas S, Senesse P
[13]
2018
France
Hospital
Cross-sectional
Adults 15–96 years;
Mean age 61.7 ± 12.9
years.
1762 patients
6
Husted MM, Fournaise
A, Matzen L, Scheller
RA [20]
2017
Denmark
Hospital
Validation study
Older adults
103 meals
7
Kawasaki Y, Sakai M,
Nishimura K, Fujiwara
K, Fujisaki K, SM,
Akamatsu R [26]
2016
Japan
Hospital
Validation study
Adults
450 trays assessed and
412 trays used in
analysis
8
Ofei KT, Mikkelsen BE,
Scheller RA [21]
2018
Denmark
Hospital
Validation study
Older adults
17 meals
9
Palmer M, Miller K,
Noble S [18]
2015
Australia
Hospital
Observational study
Adults, mean age
77 ± 8 years
15 patients, 43 intake
days
10
Schumacker CSM,
Paulus MC, Boelens
YFN, van Zanten ARH &
Kouw IWK [16]
2025
Netherlands
Hospital
Prospective study
Hospital meals
27 food trays,
comprising 108 food
items
11
Tan J, Lau KM, Ross L,
Kinneally J, Banks M,
Pelecanos A, Young A
[19]
2021
Australia
Hospital
Cross-sectional
Adults
90 trays retrieved from
trolleys from 24 wards
12
Thibault R, Goujon N,
Le Gallic E, Clairand R,
Sebille V, Vibert J,
Schneider SM,
Darmaun D [14]
2009
France
Hospital and at-home
Cross-sectional;
Adults 18–75 years,
mean age 56 ± 15 years
114 patients; visual
analogue only in 48
undernourished in
patients
13
Tulloch H, Cook S,
Nasser R, Guo G, Clay A
[23]
2019
Canada
Hospital
Validation study
Adults
401 meals from 33
patients
14
Van den Berg G, Van
Noort H, Borkent J,
Vermeulen H,
Huisman-de Waal G, de
Van der Schueren
M(17)
2025
Netherlands
Hospital
Cross-sectional
Adults
15 patients, 60 patient
days
R. Blaauw, G. Bischoff, M.A.E. de van der Schueren et al.
Clinical Nutrition ESPEN 72 (2026) 102899
5
Table 3
Included articles – Information about index and reference methods.
Authors
Index method
Person responsible for
completion of index
method a
Reference method
Person responsible for
completion of
reference method
1
Bjornsdottir R,
Oskarsdottir ES,
Thordardottir FR,
Ramel A, Thorsdottir I,
Gunnarsdottir I [24]
Plate diagram sheet
with four options
Trained nursing staff
Weighed plate wastage
Trained research staff
2
Budiningsari D, Shahar
S, Manaf ZA,
Susetyowati S. [25]
Pictorial Dietary
Assessment Tool
(PDAT) using six
options
Staff consisting of
dietitians, nurses and
serving assistants
Weighed plate wastage
Researcher
3
Dekker IM, Langius JAE,
Stelten S, de Vet HCW,
Kruizenga HM, de van
der Schueren MAE [15]
Rate a plate where
points scored are
linked to
nutritional value
Trained Researcher and
nutrition assistants
Weighed plate wastage
Researcher and
Research assistant
4
Ghisolfi A, Dupuy C,
Gines-Farano A, Lepage
B, Vellas B, Ritz P [12]
Calorie Intake Tool
(CIT) where menu
items are assigned
points based on
nutritional value.
Trained nursing staff
Weighed plate waste.
Dietitian
5
Guerdoux-Ninot E,
Flori N, Janiszewski C,
Vaille A, de Forges H,
Raynard B, Baracos VE,
Thezenas S, Senesse P
[13]
Visual/verbal
analogue scale of
food ingested
(Ingesta-VVAS) on
a scale of 0–10.
Dietitians
24-h recall
Dietitian
6
Husted MM, Fournaise
A, Matzen L, Scheller
RA [20]
Plate diagram
methods with four
or five options.
Trained nursing staff
Weighed plate wastage
Dietitian
7
Kawasaki Y, Sakai M,
Nishimura K, Fujiwara
K, Fujisaki K, SM,
Akamatsu R [26]
Visual estimation
of whole tray and
food items using a
11-point scale.
Non-trained nursing
assistants and
dietitians
Weighed plate wastage
Research staff
8
Ofei KT, Mikkelsen BE,
Scheller RA [21]
Dietary Intake
Monitoring System
(DIMS) using a
digital camera.
Non-trained assessors
Weighed plate wastage
Not stated
9
Palmer M, Miller K,
Noble S [18]
Food intake charts
using four options.
Non-trained nursing
staff
Weighed plate wastage
Dietetic staff and
dietetic students
10
Schumacker CSM,
Paulus MC, Boelens
YFN, van Zanten ARH &
Kouw IWK [16]
Food record charts
using 6 options and
digital
photography.
Non-trained healthcare
professionals (16
nurses, 10 healthcare
assistants, 4
researchers)
Weighed plate waste
Researcher
11
Tan J, Lau KM, Ross L,
Kinneally J, Banks M,
Pelecanos A, Young A
[19]
Meal Intake Points
(MIP) where menu
items are assigned
points based on
nutritional value.
Trained researcher
Weighed plate wastage
Researcher
12
Thibault R, Goujon N,
Le Gallic E, Clairand R,
Sebille V, Vibert J,
Schneider SM,
Darmaun D [14]
Self-Evaluation of
Food Intake (SEFI)
tool estimating
food intake on a
1–10 scale
Patients provided
answers based on
instructions from
interviewer.
Visual scales were only
administered on a sub-
group of malnourished
in-patients.
3-day diet record
Nursing staff for in-
patients, patients
themselves for out-
patients following
training by a dietitian
13
Tulloch H, Cook S,
Nasser R, Guo G, Clay A
[23]
Visual assessment
of whole tray using
meal plate pictorial
rating scale based
on the My Meal
Intake Tool
Trained food service
workers
Weighed plate wastage
Research assistant
14
Van den Berg G, Van
Noort H, Borkent J,
Vermeulen H,
Huisman-de Waal G, de
Van der Schueren
M(17)
Self-Evaluation of
Food Intake (SEFI)
tool estimating
food intake on a
1–10 scale
Patient Participant
Visual assessment of
food intake directly
after removal of meal
trays as 0 %, 25 %, 50 %,
75 % or 100 %
Nutrition assistants
and nursing staff
a Training refers to specific training on how to complete the index method.
R. Blaauw, G. Bischoff, M.A.E. de van der Schueren et al.
Clinical Nutrition ESPEN 72 (2026) 102899
6
Table 4
Included articles – Validity data pertaining to energy and/or protein intake.
Authors
Nutrients
measured
Sensitivity (95 % CI)
Specificity (95 % CI)
Correlation data
Bland-Altman bias
[Limits of agreement]
General comments
1
Bjornsdottir R,
Oskarsdottir ES,
Thordardottir FR,
Ramel A,
Thorsdottir I,
Gunnarsdottir I
[24]
Energy, protein
CC:
Energy: r = 0.92;
p < 0.001
Protein: r = 0.89;
p < 0.001
For all meals:
Energy: overestimation
bias = 45 kcal/day
[-231; 322 kcal/day]
Protein: bias = 1.5 g/
day [-14.0; 16.9 g/day]
For meals with intake ≤
50 % of meal served:
Energy:
underestimation
bias = 97 kcal/day
Protein:
underestimation
bias = 4.3 g/day
Cut-off for Inadequate
intake: ≤50 %
consumed
2
Budiningsari D,
Shahar S, Manaf ZA,
Susetyowati S. [25]
Energy, protein,
CHO, fat
Sensitivity to identify those
consuming ≤ 50 % of
intake:
Starch = 97 %
Animal protein = 98.2 %
Non-animal
protein = 93.9 %
Specificity to identify those
consuming ≤ 50 % of
intake:
Starch = 90 %
Animal protein = 75 %
Non-animal protein = 87 %
ICC:
Energy:
0.96 (0.94–0.97)
Protein:
0.91 (0.88–0.91)
CC:
Energy: r = 0.92;
p < 0.01
Protein: r = 0.849;
p < 0.01
Energy: [-108 to
115 kcal/day]
Protein: [-7.2 to 6.8 g/
day]
Accuracy for
assessment within 15 %
of true value:
Energy:
93.9 %; bias = 6.2 %
Protein:
90 %; bias = 10 %
3
Dekker IM, Langius
JAE, Stelten S, de
Vet HCW,
Kruizenga HM, de
van der Schueren
MAE [15]
Energy, protein
ICC:
Research assistants
Energy: 0.788 (-273,
56)
Protein: 0.905 (− 8.4,
1.0)
Trained researchers
Energy: 0.819 (− 193,
119)
Protein: 0.961 (− 4, 2.2)
Research assistants
Energy: bias = − 109
kcal; 7 % [-273; 56 kcal]
Protein: bias = − 3.7 g,
6 % [-8.4; 1.0 g]
Trained researchers
Energy: bias = 37 kcal;
2,3 % [-193; 119 kcal]
Protein: bias = 0.9 g,
1.5 % [-4.0; 2.2 g]
Accurate assessment of
intake by tool:
Research assistants:
Energy: 61 %
Protein: 61 %
Trained researchers:
Energy: 61 %
Protein: 79 %
4
Ghisolfi A, Dupuy C,
Gines-Farano A,
Lepage B, Vellas B,
Ritz P [12]
Energy
ICC:
Energy: 0.96
(0.94–0.97)
Energy: bias = 35 kcal
[ ± 420 kcal]
5
Guerdoux-Ninot E,
Flori N, Janiszewski
C, Vaille A, de
Forges H, Raynard
B, Baracos VE,
Thezenas S,
Senesse P [13]
Energy
80.8 % compared to intake
< or >25 kcal/kg/day
67.5 % (for low
food intake
<25 kcal/kg)
CC:
Energy:
Rho = 0.67, p < 0.05
Used cutoff of≤7 on the
ingesta-VVAS to
indicate low
intake as
compared to < or
>25 kcal/kg/day
Feasibility of
tool = 97.7 %
6
Husted MM,
Fournaise A,
Matzen L, Scheller
RA [20]
Energy, protein
Plate method
Energy:
bias = − 40.6 kcal
Protein: bias = − 1.0 g
Reduced plate method
Energy:
bias = − 118.8 kcal
Protein: bias = − 2.3 g
7
Energy, protein
(continued on next page)
R. Blaauw, G. Bischoff, M.A.E. de van der Schueren et al.
Clinical Nutrition ESPEN 72 (2026) 102899
7
Table 4 (continued )
Authors
Nutrients
measured
Sensitivity (95 % CI)
Specificity (95 % CI)
Correlation data
Bland-Altman bias
[Limits of agreement]
General comments
Kawasaki Y, Sakai
M, Nishimura K,
Fujiwara K, Fujisaki
K, SM, Akamatsu R
[26]
CC:
Nurses – whole trays
Energy:
Rho = 0.91, p < 0.01
Protein:
Rho = 0.88, p < 0.01
Dietitians - whole trays
Energy:
Rho = 0.94, p < 0.01
Protein:
Rho = 0.89, p < 0.01
Dietitians - individual
foods
Energy:
Rho = 0.98, p < 0.01
Protein:
Rho = 0.96, p < 0.01
Nurses – whole trays
Energy:
bias = 41.4 kcal/day
[-121 to 147 kcal/day]
Protein: bias = 2.1 g/
day [-6.4 to 7.0 g/day]
Dietitians - whole trays
Energy:
bias = 36.4 kcal/day
[-122 to 106 kcal/day]
Protein: bias = 2,0 g/
day [-6.7 to 5.5 g/day]
Dietitians - individual
foods
Energy:
bias = 23,0 kcal/day
[-82 to 66 kcal/day]
Protein: bias = 1.0 g/
day [-4.3 to 3.9 g/day]
8
Ofei KT, Mikkelsen
BE, Scheller RA [21]
Energy, protein
ICC:
Energy: 0.99
(0.98–0.99)
Protein: 0.99
(0.98–0.99)
CC:
Energy: r = 0.99,
p < 0.01
Protein: r = 0.99,
p < 0.01
Energy: bias = 13.49 kJ
[-75.07 to 102.01 kJ]
Protein: bias = 0.04 g
[-1.5 to 1.59 g]
9
Palmer M, Miller K,
Noble S [18]
Energy, protein
CC:
Breakfast: r = 0.793;
p < 0.001
Lunch: r = 0.352; NS
Supper: r = 0.393; NS
Energy: bias = −
55 ± 317 kcal/day
Protein:
bias = 0.5 ± 16.2 g/day
Up to 93 % of daily food
records were
incomplete
10
Schumacker CSM,
Paulus MC, Boelens
YFN, van Zanten
ARH & Kouw
IWK
[16]
Energy, protein
Overestimation of food
consumption:
Food record charts
bias = 3.2 % [-25.7 to
31.9 %]
Digital photography
bias = 4.7 % [-26.2 to
35.5 %]
Food record charts vs
weighed food
Energy:
bias = − 4.9 ± 92.0 kJ
Protein:
bias = − 0.2 ± 1.3 g
Digital photography vs
weighed food
Energy:
bias = 1.0 ± 108.8 kJ
Protein:
bias = − 0.1 ± 1.3 g
Cut-off for Inadequate
intake: ≤50 %
consumed
Both Food Record
Charts and Digital
Photography
overestimated food
items if consumed
<50 % (range +3.0
to +11.5 %) and
underestimated when
consumption was
>50 % (range − 2.2
to − 6.7 %).
Both methods slightly
overestimated intake of
liquids (2,3–6,9 %) and
semi-solids (7,4–7,8 %)
compared to solids
(1,7-1,8 %).
11
Tan J, Lau KM, Ross
L, Kinneally J, Banks
M, Pelecanos A,
Young A [19]
Energy, protein
Original and revised
Energy: 100 % (69.2–100 %)
Protein: 100 % (71.5–100 %)
Original
Energy: 61.3 %
(49.7–71.9 %)
Protein: 62 %
CC:
Energy: r = 0.85,
p < 0.001
Based on cut point of
poor energy intake at
the meal≤1000 kJ
and≤10 g protein
R. Blaauw, G. Bischoff, M.A.E. de van der Schueren et al.
Clinical Nutrition ESPEN 72 (2026) 102899
8
(50.4 %–72.7 %)
Revised
Energy: 80 % (69.6–88.1)
Protein: 75.9 % (65–84.9 %)
Protein: r = 0.83,
p < 0.001
Agreement with
reference limits of
within 250 kJ and 2.5 g
protein:
Energy: 58.9 %
Protein: 44.4 %
Agreement with
reference limits of
within 450 kJ and 4.5 g
protein:
Energy: 77.8 %
Protein: 62.2 %
12
Thibault R, Goujon
N, Le Gallic E,
Clairand R, Sebille
V, Vibert J,
Schneider SM,
Darmaun D [14]
Energy
CC:
Total cohort:
Verbal format:
rho = 0.66, p < 0.0001
Visual format:
rho = 0.74, p < 0.0001
In-patients:
Verbal format:
rho = 0.73, p < 0.0001
Visual format:
rho = 0.74, p < 0.0001
Out-patients:
Verbal format:
rho = 0.32, p = 0.04
Underweight patients
(BMI<19):
Verbal format:
rho = 0.78, p < 0.0003
Visual format:
rho = 0.78, p < 0.0003
Overweight patients
(BMI≥ 25):
Verbal format:
rho = 0.39, p = 0.04
Feasibility:
Visual format = 98 %
Verbal format: 96 %
13
Tulloch H, Cook S,
Nasser R, Guo G,
Clay A [23]
Energy, protein
Daily average: 81 %
(62–94 %)
Breakfast: 75 % (53–89 %)
Lunch: 93 % (83–97 %)
Supper: 100 % (87–100 %)
Daily average: 88 %
(77–95 %)
Breakfast: 90 % (83–94 %)
Lunch: 72 % (61–81 %)
Supper: 70 % (60–77 %)
CC:
Energy Breakfast:
rho = 0.624, p < 0.001
Energy Lunch:
rho = 0.771, p < 0.001
Energy Supper:
rho = 0.892, p < 0.001
Breakfast:
bias = − 7±22 %
Lunch: bias = 3 ± 21 %
Supper: bias = 5 ± 18 %
Inadequate food
intake:≤50 %
consumption
Overall
agreement = 72 %
(Range 55–100 %)
Agreement to detect
intake ≤50 % = 88 %
14
Van den Berg G,
Van Noort H,
Borkent J,
Vermeulen H,
Huisman-de Waal
G, de Van der
Schueren M(17)
Energy, protein
CC:
Energy: r = 0.097
Protein: r = 0.167
ICC:
Lunch: 0.442
Dinner: 0.620
Associations between
nurse's records and
patient's self-
assessment indicated
that higher scores on
the 1–10 scale
corresponded to higher
energy and protein
values. Every ascending
point corresponded to
an increase in 52 kcal
(955 CI: − 8; 113) and
2.6 g protein (95 % CI:
0.4; 4.8).
Abbreviations used: ICC=Interclass correlation coefficient. Presented as ICC (95 % Confidence interval [CI]).
CC=Correlation coefficient. Presented as Pearson correlation: r; p-value or Spearman correlation: rho, p-value.
NS – Non significant difference.
R. Blaauw, G. Bischoff, M.A.E. de van der Schueren et al.
Clinical Nutrition ESPEN 72 (2026) 102899
9
point equaling 130 kcal and 5 g protein. They used a food record
based on standard portion sizes as the reference method for
breakfast, lunch and snacks, but with the dinner meal weighed the
food before and after consumption. Ghisolphi et al. [12] used the
Caloric Intake Tool (CIT) to visually estimate food consumption
based on a 4-point scale, with a fixed pre-determined energy value
of every one point equaling 160 kcal according to menu calcula-
tions to a total of 640 kcal. Based on total meal points, the final
caloric value is calculated by the proportional consumption
thereof e.g. if only 50 % of a meal scoring 4 points was consumed,
total energy intake was 320 kcal [12]. In the Australian study by
Tan et al. [19], the Meal Intake Points (MIP) system was used on
adult hospitalized patients. Menu items were assigned points
based on the nutritional value in comparison to a common de-
nominator of 500–1000 kJ and 5–10 g protein for 1 point, up to a
maximum of 4 points per meal/tray. They report results of both the
original version of the MIP, as well as the adapted final version [19].
The results from the latter version are included in this scoping
review. The Pictorial Dietary Assessment Tool (PDAT) [25] provides
pictures of three food groups (staple food e.g. rice, porridge; ani-
mal food sources e.g. egg, chicken, and non-animal food sources
e.g. tempeh, tofu). The pictures represent different proportions of
food consumption with corresponding energy and protein content
[25].
Digital technology was used by two studies. Dietary Intake
Monitoring System (DIMS) consists of a device with a digital
camera and a weighing scale to measure plate weight before and
after consumption. Data is transmitted to a tablet programmed
with food analysis software [21]. Schumacker et al. [16] took pic-
tures of food trays after meal consumption and asked healthcare
workers to assess intake based on a 6-point scale. Based on the
pre-weight of the meals and the proportional intake, the infor-
mation was converted to energy and protein intake.
Summary: A variety of measurement tools have been used for
assessment of food intake, ranging from VEMs using a 10-point
scale, to VEMs with plate pictorials with or without allocated
predefined energy and protein values, to using photographs of
meals.
3.3.2. Validity of index methods
The wording of the original GLIM framework requires an energy
estimation to a precision of 50 % of energy requirements [1]. Please
refer to Table 4 for the validation results of the included studies
discussed below.
3.3.2.1. 10-point scale concept. Ingesta-VVAS defined low intake as
patients consuming less than 25 kcal/kg/day and this correspond
to a value of ≤7 on the 10-point scale [13]. A good sensitivity of
80.8 % accurately identified individuals having low intake, while
32.5 % were incorrectly identified as not having low intake
(specificity 67.5 %). The Ingesta-VVAS had a fair correlation with
energy intake (rho = 0.67, p < 0.05) and was considered feasible
(rating of 97.7 %) [13]. Thibault et al. [14] reported a fair to good
correlation between scoring and energy intake (verbal form
rho = 0.66 and visual form rho = 0.74). Although most people felt
comfortable providing verbal answers (98 %), the visual form
provided better accuracy [14]. Van den Berg et al. [17] found higher
scoring corresponded to higher intake levels, where every
ascending point corresponded to a 52 kcal and 2.6 g protein
increase.
3.3.2.2. Visual plate pictorials. Using cut-offs of consuming ≤50 %
of meals, Bjornsdottir et al. reported good accuracy with an un-
derestimation for energy (bias = 97 kcal/day) and protein
(bias = 4.3 g/day). In general the tool overestimated slightly for
energy (bias = 45 kcal) and protein (bias = 1.5 g) [24]. Husted et al.
used a 5-option model (plate model) and a 4-option model
(reduced plate model), and reported better results for the 5-
options model, with an energy underestimation of 6 % and pro-
tein underestimation of 10 % (energy bias = 40.6 kcal and protein
bias = 0.5 g) versus the 4-options tool that underestimated protein
by 22 % (energy bias = 118.8 kcal and protein bias = 2.3 g) [20].
Although Palmer et al. reported good accuracy with small under-
estimation for both energy (bias = 55 kcal) and protein
(bias = 0.5 g), only 7 % of their daily food records were complete
[18].
Fig. 2. Measurement tools for assessment of dietary intake. The figure graphically depicts the four types of measurement tools used to assess dietary intake across the included
studies. Panel A: Visual estimation method (VEM) using a 10-point scale with slider ranging from 1 (I did not eat anything) to 10 (I ate everything); Panel B: VEM based on plate-
model pictures. In this example 5 plate options are used with 0 % (no food was consumed), 25 % (a quarter of food was consumed), 50 % (half of food was consumed), 75 % (three-
quarters of food was consumed) and 100 % (all food was consumed). The grey shaded area indicates food left on the plate and the white area indicates food consumed; Panel C:
Similar to Panel B, but with points allocated based on the proportion of food consumed. The points are given a specific energy (kcal) and protein (g) value based on the dietary
analysis of the menu served; Panel D: The use of digital technology to determine nutritional composition of food consumed. It consists of a camera to photograph the plate of food
before and after consumption, a scale to weigh the food to determine gram value, a computer programme with food analysis software and a nutritional analysis printout.
R. Blaauw, G. Bischoff, M.A.E. de van der Schueren et al.
Clinical Nutrition ESPEN 72 (2026) 102899
10
Inadequate intake, defined as consuming ≤50 % of meals, was
used as the reference in the study by Tulloch et al. [23]. The daily
average intake was correctly estimated in 72 % of cases. Those with
inadequate intake were correctly identified in 88 % of cases. On
average, sensitivity was 81 %, however, they reported variations
between the three meals of the day, with best accuracy at supper
(Se 100 %, rho 0.89), followed by lunch (Se 93 %, rho = 0.77) and
breakfast (Se 75 %, rho = 0.62). Accuracy decreased at the opposing
ends of the scale with underestimation in the 0–25 % group and
overestimation in the 75–100 % consumption groups [23].
Kawasaki et al. [26] interpreted whole trays using an 11-point
scale, and reported that 61 % of trays were categorized with
90–100 % accuracy. They did report a difference in the accuracy of
measurements performed by nursing assistants and dietitians,
with more accurate assessments performed by the latter group
(energy bias = 36.4 kcal; protein bias = 1.0 g) versus the nursing
assistants (energy bias = 41.4 kcal; protein bias = 2.1 g). Dietitians
performed even better when scoring individual foods (energy
bias = 23 kcal; protein bias = 1.0 g). Intake of certain food groups
(dairy and grains) and consumption of modified textured foods
(minced meals) affected the accuracy of the data [26].
3.3.2.3. Visual plate pictorials with associated predefined nutritional
value. Dekker et al. [15] reported different accuracy results for the
research assistants and dietitians, with less bias reported for the
latter group. Energy was underestimated by 2.3 % (bias = 37 kcal)
versus 7 % (bias = − 109 kcal) and protein underestimation by 1.5 %
(bias = 0.9 g) versus 6 % (bias = − 3.7 g) (dietitians versus research
assistants respectively). A good interclass correlation (ICC) of
0.79–0.96 between energy and protein was reported for both
groups [15]. Ghisolfi et al. [12] reported good ICC between nursing
staff trained to complete the tool, with an energy bias = 35 kcal.
Items with larger caloric content produced a slight energy un-
derestimation with the CIT compared to weighed plate waste [12].
A cut-off point for poor intake per meal, which corresponded to
50 % of the nutritional value of the meals provided, was set at ≤
1000 kJ and ≤10 g protein by Tan et al. [19]. Both the original and
revised versions of MIP produced 100 % sensitivity for both energy
and protein. Specificity improved in the revised version to 80 % for
energy and 75.9 % for protein, but it showed lower accuracy to
correctly identify higher intakes. They stated a priori that good
agreement would be achieved if the index method scored within
250 kJ and 2.5 g protein compared to the reference test. This was
achieved for 58.97 % of meals for energy and 44.4 % of meals for
protein. If the limits of agreement were extended to within 450 kJ
and 4.5 g protein, it was achieved in 77.8 % (energy) and 62.2 %
(protein) of meals. Good correlation was achieved between food
weighing and the MIP (energy: r = 0.85; protein: r = 0.83) [19].
Budiningsari et al. [25] reported accuracy within 15 % of the true
value for more than 90 % of cases for both energy (bias = 6.2 kcal/
day) and protein (bias = 10 g/day). Different staff assessors agreed
with each other with an ICC of 0.96 (energy) and 0.91 (protein). A
very good correlation between weighing and PDAT was reported
for energy (r = 0.929) and protein (r = 0.859). The percentage of
individuals with inadequate food intake and identified by PDAT as
at risk for intakes ≤50 %, were good for all food groups (sensitivity
for staple food 97 %, animal foods 98.2 % and non-animal foods
93.9 %).
3.3.2.4. Digital technology. Very small underestimation of energy
(bias = 13.49 kJ) and protein (bias = 0.04 g) consumption, with
good agreement (ICC 0.99 for both energy and protein) and
excellent correlation (r = 0.99 for energy and r = 0.98 for protein)
was reported using digital camera technology [21]. Similarly, good
accuracy was reported by Schumacker et al. [16] with an
overestimation of food intake <5 % using photography
(bias = 4.7 %) or food record charts (bias = 3.2 %). The actual impact
on energy and protein intake was negligible with < 5 kJ and 0.2 g
protein difference from weighed records. Overestimation of intake
was more prevalent when overall consumption was less than 50 %,
or in the case of consumption of liquids or semi-solid foods [16].
Summary: The 10-point Ingesta-VVAS missed 32.5 % of in-
dividuals with inadequate intake. This decreased to 12 % using
visual plate pictorials. Tools tend to be less accurate at opposing
ends of intake levels, also when assessing different food consis-
tencies, food groups, different meals and therapeutic diets versus
regular meals. Using digital technology proves to provide accurate
data, however, the human component is still required to estimate
proportional intake and preprogramme the device with nutritional
values of menu items.
3.3.3. Accuracy of the index tools used by various healthcare
professionals in estimating food intake
Both studies using VEM based on a 1–10 point scale found that
98 % of participants could answer the verbal form completed by
the patients themselves or dietitians, indicating good agreement
[13,14]. The 4-option VEM plate models of Bjornsdottir et al. [24]
and Palmer et al. [18] showed good accuracy for nursing staff
completing the tools with an underestimation of energy of less
than 100 kcal (bias of 97 kcal and 55 kcal respectively). Similarly,
Ghisolfi et al. [12] reported an energy bias of 35 kcal for nursing
staff trained to complete the tools. These differences are relatively
small and should not influence the outcome of using these tools in
dietary assessment. The study by Palmer et al. [18] did, however,
report differences between the meals of the day, with the biggest
underestimation of both energy and protein intake noted for
breakfast. This aspect could be considered during the training of
the staff. Even though it is part of their routine clinical practice,
nursing assistants, not trained in the method being tested, had
significantly lower accuracy in estimating food intake using an 11-
point scale compared to dietitians, without specific training in
completion of the tools [26]. In the case of Tulloch et al. [23] food
service workers adequately trained to complete the tools could
correctly determine daily intake of patients in 72 % of cases and
could correctly identify patients with inadequate intakes (≤50 % of
intake) in 88 % of cases. The results of this study have important
clinical implications, as it showed that adequately trained food
service workers can accurately assess patients’ intake quickly
when removing trays. Nutrition assistants could accurately
determine energy and protein for 61 % of cases, and under-
estimated intakes in up to a third of cases using the Rate-the-Plate
method [15]. In the same study, adequately trained researchers
reached the same level of accuracy for energy (61 %), but could
determine protein with higher accuracy (79 %) [15]. Using the
PDAT, dietitians, nurses and serving assistants showed that staff
from different backgrounds could accurately complete the tool
with a good agreement amongst them (ICC of more than 0.90) [25].
It is also clear that employing digital technology resulted in good
agreement between assessors without receiving specific training
(ICC of 0.99) [21].
Summary: Good results are shown when the various tools were
completed by patients themselves, nursing staff, food service
workers and dietitians. It is also clear that adequate training on the
completion of the tools enhances accuracy.
4. Discussion
This scoping review appraised measures for food intake of adult
patients in healthcare settings and reports on their validity, reli-
ability and potential for application for determining the food
R. Blaauw, G. Bischoff, M.A.E. de van der Schueren et al.
Clinical Nutrition ESPEN 72 (2026) 102899
11
intake criteria of GLIM. Determination of the food intake variable
within GLIM should be easy to complete by most healthcare pro-
fessionals with limited nutrition training. For this reason, we
limited our search to tools that take less than 5 min to complete
(e.g. less than 20 items), with minimal participant burden. Com-
mon dietary intake assessment methods (e.g. 24-hour recall, food
frequency questionnaire) were thus not considered. Our focus was
also to identify tools that could extrapolate data to percentage
energy intake, as the original GLIM framework operationalized
reduced food intake as ≤ 50 % of intake [1,2].
Assessment of food intake is essential for the diagnosis of
malnutrition. Various techniques for dietary assessment are available
with different levels of ease and time required for completion,
analysis and interpretation [28,29]. In general, most techniques
require a certain amount of training and skills to implement.
Weighed records are considered to provide the most accurate results
[30] and were used as the reference method in many studies included
in this scoping review. Interestingly, only four studies conducted food
weighing before and after consumption [12,20,21,23,27], while the
remaining studies compared post-meal weights to an assumed
standard portion of food served [18,19,24,25,31].
We found various visual estimation methods reported, with
different levels of accuracy and sophistication. These ranged from
a 1–10-point scale without any visuals, to visuals of various plate
options, with and without pre-determined nutritional value, to
digital technology. A VEM is defined as “a method or tool that aids
a rater to estimate the proportion or amount of food and fluids
consumed during eating and drinking occasions, subsequently
allowing for nutrient analysis” [8].
Comparing the accuracy of the plate model with 5 options
versus the one with 4 options, Husted et al. [20] reported smaller
bias for both energy and protein with the 5-options [20]. Keller
et al. also noted that factors affecting accuracy of VEM include
portion-controlled meals (smaller portion sizes) and adding
additional food items to trays. The latter resulted in a 3.8 times
higher chance of misclassification (OR = 3.85) [31].
Visuals with associated energy and protein values attached,
accurately reflected nutritional intake. These methods, however,
require significant effort at the food production end (food service
unit) to ensure use of standardized menus and nutritional analysis,
as it restricts menu flexibility. But it does save time at the user-end
(ward level). This need for accurate menu analysis was stressed by
a few studies [12,15].
Incorporating intake analysis by means of pre-programmed
nutritional analysis, can assist with identifying at-risk in-
dividuals at the ward level. This was also reported by Budiningsari
et al. [25] who found very good accuracy in predicting individuals
at risk of ≤50 % intake. They stated that the use of a system with
pre-programmed nutritional analysis for every menu item saves
time and assists in the calculation of nutritional content. It does
require standardized meals and detailed menu analysis before-
hand. This was emphasized in the study by Palmer et al. [18]. These
authors reported that researchers were unable to determine daily
intake in 93 % of the food intake charts, due to missing data in the
pre-programmed nutritional analysis, or due to inadequate sup-
plier information, resulting in the inability to determine up to
450 kcal and 24 g protein at meals. This led to an inability to
determine agreement between the index and reference methods
[18]. The problem was therefore not with using the VEM, but the
accuracy of the data available in food intake charts [18].
Digital technology provided the most accurate estimation of food
intake, however, the clinical application and usability in current
resource-constrained environments are questioned. In general,
image-based methods (digital) can affect intake and result in bias
associated with the knowledge that an image will be taken before
and after meals, as this process can affect behavior. Using technology
is also dependent on the availability and tech-savviness of partici-
pants [29]. A positive component of digital methodology, is that
memory is not an issue and portion sizes can be more accurately
captured [29] and digital technology can be implemented with
minimal interruptions during meal times [32]. Currently, a human
component is still required to assess proportional intake based on
the photographs taken and further analysis for nutritional compo-
sition. It therefore saves time at the ward level but requires the same
detailed processes for the analysis as for other tools.
In general, various factors affect participants’ ability to rate
their food intake. This can include the type of meals received,
where differences were reported in the accuracy of estimating
regular diets compared to therapeutic diets, with regular diets
being less accurate [27]. Food consistency also impacts accuracy of
assessment, with a lower sensitivity reported for fluids compared
to solids [22]. Variations were also found in the accuracy of
reporting specific meals, where Palmer et al. reported accuracy of
especially lunch and supper to be very poor [18], to Tulloch et al.
reporting the opposite, with best results at supper [23].
Barriers to reporting accurate intake are more commonly re-
ported in women [29,33,34], younger participants [33], those
consuming less than 50 % of a meal [33], underweight individuals
[33,34], overweight individuals [14], those suffering comorbidities
[33] and mental issues (depression and anxiety) [34]. Other
components that influence accuracy of reporting include types of
meals (therapeutic versus regular diets), as mentioned above [27],
and setting (outpatients versus in-patients) [14].
Most tools reviewed have the ability to accurately identify
those individuals with inadequate intake, however, the ability to
extrapolate the data for determination of actual energy re-
quirements to comply with the original GLIM variable, is chal-
lenging. Using a VEM with more options, for instance 11 options
[26], provides estimations to an accuracy levels of 10 %, whereas
the more commonly used 4 or 5 option VEMs, can only provide
data to an accuracy level of 20–25 % of actual intake [18,20,24].
These tools can more easily identify percentage meal intake than
percentage actual energy requirements.
Dietary intake tools can be completed by various health care
professionals. However, adequate training of staff, especially non-
dietetic professionals, is needed to improve the accuracy of results
[8,15]. Among staff members (nurses and nursing assistants), those
who received training in completing VEM demonstrated greater
knowledge, paid more attention to detail (e.g., removing dish
covers), and showed improved understanding of portion sizes [35].
All tools should be supported by nutrition support protocols to
ensure that at-risk cases are escalated for appropriate treatment.
Such protocols must define the number of inadequate meals
permitted and the intake thresholds that trigger intervention.
4.1. Strengths and limitations
Strengths in this study are the use of multiple databases for the
search, the use of standard procedures and double-blind review of
included references. Only publications written in English, and
those that estimated energy and protein were included. Another
limitation is that references focused on healthy individuals, ath-
letes, individuals with overweight or with noncommunicable
diseases, and pregnancy were excluded, but appropriately limiting
the generalizability to the clinical setting.
4.2. Conclusions
Different index methods were used primarily in hospital set-
tings to determine food intake and specifically energy intake.
R. Blaauw, G. Bischoff, M.A.E. de van der Schueren et al.
Clinical Nutrition ESPEN 72 (2026) 102899
12
These methods were generally accurate as compared to a reference
method, and more specifically individuals with reduced food
intake can be identified. These brief food intake measures can be
completed by various members of the health care team, as well as
participants themselves, with training improving accuracy.
Although the tools can identify inadequate intake, which is
required for the GLIM criteria, they do not identify sufficiency of
this intake relative to the individual patient's energy requirement.
This component of the GLIM framework needs reconsideration, as
establishing an individual energy requirement in clinical settings
requires specialist skills. Clinicians can use the various methods of
visual estimation found in this scoping review with confidence to
determine food intake in clinical settings.
Author contributions
Ren
ee Blaauw: Conceptualization; Project administration;
Literature searches; Methodology; Screened all relevant articles
for eligibility; Full article evaluation; Writing - original draft;
Writing - review & editing; Gert Bischoff: Screened all relevant
articles for eligibility; Writing - review & editing; Marian de van
der Schueren: Screened all relevant articles for eligibility; Writing
- review & editing; Charlene Compher: Screened all relevant ar-
ticles for eligibility; Writing - review & editing; Krista Haines:
Screened all relevant articles for eligibility; Writing - review &
editing; Nicole Kiss: Screened all relevant articles for eligibility;
Writing - review & editing; Charles Lew: Screened all relevant
articles for eligibility; Writing - review & editing; Ainsley Malone:
Screened all relevant articles for eligibility; Writing - review &
editing; Claudia Maza: Screened all relevant articles for eligibility;
Writing - review & editing; Nancy Stoner: Screened all relevant
articles for eligibility; Writing - review & editing; Heather Keller:
Screened all relevant articles for eligibility; Software; Full article
evaluation; Writing - review & editing. Conceptualization; Data
curation; Formal analysis; Funding acquisition; Investigation;
Methodology; Project administration; Resources; Software; Su-
pervision; Validation; Visualization; Writing - original draft;
Writing - review & editing.
Declaration of generative AI and AI-assisted technologies in
the writing process
During the preparation of this work ChatGPT was used to
generate Fig. 2. The authors reviewed the content for factual cor-
rectness and take full responsibility for the content of the pub-
lished article.
Funding
This work did not receive any funding.
Conflict of interest
The authors declare no conflict of interest related to this work.
Appendix A. Supplementary data
Supplementary data to this article can be found online at
https://doi.org/10.1016/j.clnesp.2025.102899.
References
[1] Cederholm T, Jensen GL, Correia M, Gonzalez MC, Fukushima R,
Higashiguchi T, et al. GLIM criteria for the diagnosis of malnutrition - a
consensus report from the global clinical nutrition community. Clin Nutr
2019;38(1):1–9.
[2] Jensen GL, Cederholm T, Correia M, Gonzalez MC, Fukushima R,
Higashiguchi T, et al. GLIM criteria for the diagnosis of malnutrition: a
consensus report from the global clinical nutrition community. JPEN - J
Parenter Enter Nutr 2019;43(1):32–40.
[3] Cederholm T, Barazzoni R, Rothenberg E, Wickramasinghe K, Cuerda C,
Schneider SM, et al. A new diagnosis code in ICD-11 for undernutrition in
adults - a historic achievement for the clinical nutrition community. Clin
Nutr 2025;55:219–21.
[4] de van der Schueren MAE, Keller H, Consortium G, Cederholm T, Barazzoni R,
Compher C, et al. Global Leadership Initiative on Malnutrition (GLIM):
guidance on validation of the operational criteria for the diagnosis of protein-
energy malnutrition in adults. Clin Nutr 2020;39(9):2872–80.
[5] Barazzoni R, Jensen GL, Correia M, Gonzalez MC, Higashiguchi T, Shi HP, et al.
Guidance for assessment of the muscle mass phenotypic criterion for the
Global Leadership Initiative on Malnutrition (GLIM) diagnosis of malnutri-
tion. Clin Nutr 2022;41(6):1425–33.
[6] Cederholm T, Jensen GL, Ballesteros-Pomar MD, Blaauw R, Correia M,
Cuerda C, et al. Guidance for assessment of the inflammation etiologic cri-
terion for the GLIM diagnosis of malnutrition: a modified Delphi approach.
Clin Nutr 2024;43(5):1025–32.
[7] Jensen GL, Cederholm T, Ballesteros-Pomar MD, Blaauw R, Correia M,
Cuerda C, et al. Guidance for assessment of the inflammation etiologic cri-
terion for the GLIM diagnosis of malnutrition: a modified Delphi approach.
JPEN - J Parenter Enter Nutr 2024;48(2):145–54.
[8] Heighington-Wansbrough AJ, Gemming L. Dietary intake in hospitals: a
systematic literature review of the validity of the visual estimation method
to assess food consumption and energy and protein intake. Clin Nutr ESPEN
2022;52:296–316.
[9] Ferguson CE, Tatucu-Babet OA, Amon JN, Chapple LS, Malacria L, Myint
Htoo I, et al. Dietary assessment methods for measurement of oral intake in
acute care and critically ill hospitalised patients: a scoping review. Nutr Res
Rev 2025;38(1):81–94.
[10] Veritas Health I. Covidence systematic review software. Melbourne,
Australia: Veritas Health Innovation; 2025.
[11] Page MJ, McKenzie JE, Bossuyt PM, Boutron I, Hoffmann TC, Mulrow CD, et al.
The PRISMA 2020 statement: an updated guideline for reporting systematic
reviews. BMJ 2021;372:1–9.
[12] Ghisolfi A, Dupuy C, Gines-Farano A, Ritz P. Validation of a new tool: the
calorie intake tool, to easily estimate the energy intake of diseased aged
patient. J Nutr Health Aging 2014;18(10):857–60.
[13] Guerdoux-Ninot E, Flori N, Janiszewski C, Vaill
e A, de Forges H, Raynard B,
et al. Assessing dietary intake in accordance with guidelines: useful corre-
lations with an ingesta-Verbal/Visual analogue scale in medical oncology
patients. Clin Nutr 2019;38(4):1927–35.
[14] Thibault R, Goujon N, Le Gallic E, Clairand R, Sebille V, Vibert J, et al. Use of
10-point analogue scales to estimate dietary intake: a prospective study in
patients nutritionally at-risk. Clin Nutr 2009;28(2):134–40.
[15] Dekker IM, Langius JAE, Stelten S, de Vet HCW, Kruizenga HM, de van der
Schueren MAE. Validity of the "Rate-a-Plate" method to estimate energy and
protein intake in acutely ill, hospitalized patients. Nutr Clin Pract 2020;35(5):
959–66.
[16] Schumacker CSM, Paulus MC, Boelens YFN, van Zanten ARH, Kouw IWK.
Dietary food record charts and digital photography effectively estimate
hospital meal consumption. Clin Nutr ESPEN 2025;66:115–20.
[17] van den Berg G, van Noort H, Borkent J, Vermeulen H, Huisman-de Waal G, de
van der Schueren M. Self-assessment of food intake by hospitalized patients:
accuracy and acceptability of the Simple Evaluation of Food Intake (SEFI®); a
pilot study. 2025. Unpublished, https://hdl.handle.net/2066/321029.
[18] Palmer M, Miller K, Noble S. The accuracy of food intake charts completed by
nursing staff as part of usual care when no additional training in completing
intake tools is provided. Clin Nutr 2015;34(4):761–6.
[19] Tan J, Lau KM, Ross L, Kinneally J, Banks M, Pelecanos A, et al. Development of
a new tool to monitor and identify inadequate oral intake in hospital. Nutr
Diet 2021;78(3):296–304.
[20] Husted MM, Fournaise A, Matzen L, Scheller RA. How to measure energy and
protein intake in a geriatric department – a comparison of three visual
methods. Clin Nutr ESPEN 2017;17:110–3.
[21] Ofei KT, Mikkelsen BE, Scheller RA. Validation of a novel image-weighed
technique for monitoring food intake and estimation of portion size in
hospital settings: a pilot study. Public Health Nutr 2019;22(7):1203–8.
[22] McCullough J, Keller H. The My Meal intake tool (M-MIT): validity of a patient
self-assessment for food and fluid intake at a single meal. J Nutr Health Aging
2018;22(1):30–7.
[23] Tulloch H, Cook S, Nasser R, Guo G, Clay A. Food service workers: reliable
assessors of food intake in hospitalized patients. Can J Diet Pract Res
2019;80(1):30–3.
[24] Bjornsdottir R, Oskarsdottir ES, Thordardottir FR, Ramel A, Thorsdottir I,
Gunnarsdottir I. Validation of a plate diagram sheet for estimation of
energy and protein intake in hospitalized patients. Clin Nutr 2013;32(5):
746–51.
[25] Budiningsari D, Shahar S, Abdul Manaf Z, Susetyowati S. A simple dietary
assessment tool to monitor food intake of hospitalized adult patients.
J Multidiscip Healthc 2016;9:311–22.
R. Blaauw, G. Bischoff, M.A.E. de van der Schueren et al.
Clinical Nutrition ESPEN 72 (2026) 102899
13
[26] Kawasaki Y, Sakai M, Nishimura K, Fujiwara K, Fujisaki K, Shimpo M, et al.
Criterion validity of the visual estimation method for determining patients'
meal intake in a community hospital. Clin Nutr 2016;35(6):1543–9.
[27] Amaral YG, Penaforte FRdO, AraÚJo LBd, Japur CC. Can hospitalized patients
adequately estimate their own food intake? A cross-sectional pilot study. Rev
Nutr 2022;35:1–13.
[28] Dao MC, Subar AF, Warthon-Medina M, Cade JE, Burrows T, Golley RK, et al. Di-
etary assessment toolkits: an overview. Public Health Nutr 2019;22(3):404–18.
[29] Burrows TL, Ho YY, Rollo ME, Collins CE. Validity of dietary assessment
methods when compared to the method of doubly labeled water: a sys-
tematic review in adults. Front Endocrinol (Lausanne) 2019;10:850.
[30] Rutishauser IH. Dietary intake measurements. Public Health Nutr
2005;8(7A):1100–7.
[31] Kawasaki Y, Akamatsu R, Tamaura Y, Sakai M, Fujiwara K, Tsutsuura S. Dif-
ferences in the validity of a visual estimation method for determining
patients' meal intake between various meal types and supplied food items.
Clin Nutr 2019;38(1):213–9.
[32] Winzer E, Luger M, Schindler K. Using digital photography in a clinical
setting: a valid, accurate, and applicable method to assess food intake. Eur J
Clin Nutr 2018;72(6):879–87.
[33] Keller H, Allard J, Vesnaver E, Laporte M, Gramlich L, Bernier P, et al. Barriers
to food intake in acute care hospitals: a report of the Canadian malnutrition
task force. J Hum Nutr Diet 2015;28(6):546–57.
[34] Curtis LJ, Valaitis R, Laur C, McNicholl T, Nasser R, Keller H. Low food intake in
hospital: patient, institutional, and clinical factors. Appl Physiol Nutr Metabol
2018;43(12):1239–46.
[35] Kawasaki Y, Tamaura Y, Akamatsu R, Sakai M, Fujiwara K. Differences
in
characteristics
of
raters
who
use
the
visual
estimation
method in hospitals based on their training experiences. Clin Nutr
2019;38(1):240–5.
R. Blaauw, G. Bischoff, M.A.E. de van der Schueren et al.
Clinical Nutrition ESPEN 72 (2026) 102899
14