JNS

JOURNAL OF NUTRITIONAL SCIENCE

RESEARCH ARTICLE

Prevalence of nutritional risk in the non-demented hospitalised elderly:

a cross-sectional study from Norway using stratified sampling

Helene K. Eide1,2*, Jūratė Šaltytė Benth3, Kjersti Sortland2, Kristin Halvorsen2 and Kari Almendingen2
1
Division of Medicine, Akershus University Hospital and Institute of Clinical Medicine, University of Oslo, Lørenskog, Norway
2
Department for Health, Nutrition and Management, Faculty of Health Sciences, Oslo and Akershus University College of Applied Sciences, Oslo,
Norway
3
Institute of Clinical Medicine, Campus Ahus, University of Oslo and HØKH, Research Centre, Akershus University Hospital, Lørenskog, Norway
(Received 31 August 2014 – Final revision received 31 August 2014 – Accepted 21 January 2015)

Journal of Nutritional Science (2015), vol. 4, e18, page 1 of 9 doi:10.1017/jns.2015.8

Abstract
There is a lack of accurate prevalence data on undernutrition and the risk of undernutrition among the hospitalised elderly in Europe and Norway. We
aimed at estimating the prevalence of nutritional risk by using stratified sampling along with adequate power calculations. A cross-sectional study was car-
ried out in the period 2011 to 2013 at a university hospital in Norway. Second-year nursing students in acute care clinical studies in twenty hospital wards
screened non-demented elderly patients for nutritional risk, by employing the Nutritional Risk Screening 2002 (NRS2002) form. In total, 508 patients
(48·8 % women and 51·2 % men) with a mean age of 79·6 (SD 6·4) years were screened by the students. Mean BMI was 24·9 (SD 4·9) kg/m2, and
the patients had been hospitalised for on average 5·3 (SD 6·3) d. WHO’s BMI cut-off values identified 6·5 % as underweight, 48·0 % of normal weight
and 45·5 % as overweight. Patients nutritionally at risk had been in hospital longer and had lower average weight and BMI compared with those not at risk
(all P < 0·001); no differences in mean age or sex were observed. The prevalence of nutritional risk was estimated to be 45·4 (95 % CI 41·7 %, 49·0) %,
ranging between 20·0 and 65·0 % on different hospital wards. The present results show that the prevalence of nutritional risk among elderly patients with-
out dementia is high, suggesting that a large proportion of the hospitalised elderly are in need of nutritional treatment.

Key words: Nutritional risk: Elderly: Hospital practice: Cross-sectional studies: Stratified sampling

Undernutrition and the risk of undernutrition constitutes a recent decades(3–5). Prevalence estimates vary even more in a
serious public health problem today and occurs frequently number of European studies(6–13). The present study, there-
among the hospitalised elderly in developed countries(1). fore, aims to add to the body of quality prevalence data by pro-
https://doi.org/10.1017/jns.2015.8 Published online by Cambridge University Press

However, the extent of the problem is not well described in viding prevalence estimates that meet strict methodological
the relevant literature and there is a lack of accurate prevalence criteria.
data in Europe and Norway. Many of the studies conducted Ageing results in physiological, psychological and social
are based on small or narrowly defined hospital populations, changes such as reduced lean body mass, impairment of
or use inadequate statistical sampling methods when collecting senses like taste and smell, loneliness and cognitive impairment
data – all of which affect the prevalence estimates in an – all of which may contribute to the development of under-
unfavourable way. Besides, different measurement methods nourishment(14), again exacerbated by the presence of acute ill-
are often employed as there is currently no clear consensus for ness(1,15). Moreover, most hospitalised elderly have chronic
a ‘gold standard’ method(1,2). Estimates between 50 and 75 % diseases and multiple diagnoses(16), which in turn increase
are reported in a few Norwegian studies conducted in the risk of undernutrition(1,15). If untreated, undernutrition

Abbreviations: NRS2002, Nutritional Risk Screening 2002; S1, student 1; S2, student 2.
* Corresponding author: Mrs Helene Kjøllesdal Eide, fax +47 69849008, email [email protected]

© The Author(s) 2015. The online version of this article is published within an Open Access environment subject to the conditions of the Creative
Commons Attribution license <http://creativecommons.org/licenses/by/3.0/>. 1
 journals.cambridge.org/jns

can result in a variety of negative consequences and is asso- experts. All second-year nursing students at the university col-
ciated with higher morbidity and mortality rates, more fre- lege in question who were undergoing their acute and clinical
quent complications, and longer hospital stays(1,15). care practice studies at the university hospital were instructed
In recent years, international and national guidelines have to screen elderly patients for nutritional risk. The bachelor
been published in Europe(17,18) and in Norway(19) to prevent nursing education programme has a particular focus on nutri-
and treat undernutrition in healthcare institutions effectively. tion, and the screening was an important part of the students’
The guidelines recommend that all patients in hospital care clinical training and education. To meet the substantial chal-
must be screened for nutritional risk on admission so that lenges related to undernourishment in the hospital setting, it
affected patients are identified(17–19), a recommendation man- is vital that nursing students receive proper education and
datory by law in Norway(19). The goal of nutritional risk training in nutritional risk screening. Involving students in
screening is to evaluate whether nutritional treatment is likely research activities is also of importance for the university col-
to influence the patients’ outcome(17). Such a screening aims lege offering the study programme to strengthen evidence-
to identify already undernourished patients and patients at based practice.
risk(17,19). A variety of nutritional risk screening tools have Totally, fourteen of sixteen medical and surgical somatic
been developed and published for use in the hospital set- wards at the university hospital were included in the study.
ting(20,21), most of them based on recent weight loss, food Additionally, one rehabilitation ward, one specialised short-
intake and BMI(2,17). Disease severity is also accounted for term unit, one emergency medicine ward and one cardiac
in some of the tools since stress metabolism may increase monitoring ward were included. Two wards were each divided
the patients’ nutritional needs(2,17). into two sub-wards due to differences in the patients’ diagno-
The increasing number of elderly individuals contributes to ses. Naturally, it was reasonable to assume that each of the
substantial challenges for the healthcare sector(16). Prevention twenty wards (Fig. 1) represented a homogeneous subgroup
and treatment of undernourishment in the elderly are thus of of the patient population. Data were therefore collected by
great importance and may yield both health- related and finan- using stratified sampling(22), with the wards defined as strata.
cial benefits. Awareness of incidence and prevalence estimates Stratified sampling is known to be the most representative
is central in highlighting the problem of undernourishment in of a population in the sense of minimised sampling error. A
the elderly, and is important for allocating healthcare statistician was responsible for the statistical sampling design.
resources. To our knowledge, no adequately designed preva-
lence study on undernutrition and the risk of undernutrition
has previously been conducted among the hospitalised elderly Selection of participants
in Norway. Since Norway represents a typical modern Western
Nine nutritional screening days were conducted in the academ-
society, such a study would provide important insights into the
ic years 2011/2012 and 2012/2013 (Fig. 1). The screening
problem of undernourishment among the hospitalised elderly
in Scandinavia as well as in Western Europe.
The present study is specifically targeted at estimating the
prevalence of nutritional risk among elderly hospitalised
patients. A stratified sampling technique reducing sampling
error was utilised in data collection to improve the representa-
tiveness of the sample. Adequate power calculations based on
rather strong assumptions were performed a priori to assure an
accurate estimate of the prevalence.
https://doi.org/10.1017/jns.2015.8 Published online by Cambridge University Press

Methods
Study design
A cross-sectional study was designed and carried out at one
university hospital in Norway. The university hospital operates
as both a local and regional hospital, thereby offering locally
based specialist healthcare services as well as more specialised
services. The hospital covers about 10 % of the Norwegian
population, providing healthcare services for approximately
half a million individuals living in urban and rural municipal-
ities. The patient population is heterogeneous with respect to
ethnicity and socio-economic factors, and could be considered
representative of Norwegian society. Fig. 1. Study design. In total 508 hospitalised elderly (≥70 years) patients par-
The study was developed by a collegium at a nursing bach- ticipated in the study. All second-year nursing students who were undergoing
their acute and clinical care practice studies conducted nutritional risk screen-
elor education programme in a multidisciplinary collaboration ing on twenty hospital wards. Nine nutritional screening days were conducted
with representatives from the university hospital and other in the academic years 2011/2012 and 2012/2013.
2
 journals.cambridge.org/jns

days were Tuesdays, Wednesdays and Thursdays to ensure a will benefit from nutritional treatment due to undernutrition
steady coverage of patients, as most patients are admitted on and/or increased nutritional needs resulting from disease(26).
Mondays and discharged on Fridays. The data were collected The screening form included an initial screening and a final
by 173 students. All the students were informed about the screening (Appendix 1). The final screening was conducted
study and introduced to the screening form by a clinical diet- if the answer was ‘yes’ to any one of the four questions in
itian at the start of the academic year. Shortly before each the initial screening. Patients with a total score of three or
screening day, clinical supervisors (lecturers and professors) more were classified as nutritionally at risk. All scorings of
from the university college met the students in small groups nutritional risk were checked by a clinical dietitian shortly
to go through the questionnaire and the screening form care- after each screening day.
fully. A research and development nurse at the university hos-
pital was employed by project funding to ensure better
Pilot and inter-rater agreement studies
communication with the wards and to inform the ward staff
about the screening. A pilot study involving 290 elderly patients and ninety-six
All elderly (≥70 years) patients admitted on the included nursing students was performed during the autumn of 2010
wards at 08.00 hours on the screening days were asked to par- and the spring of 2011 at the university hospital to test the
ticipate. Eligible patients were selected by the students in use of a nutritional risk screening form, as well as the addition-
cooperation with the ward nursing staff. Terminal patients, al questionnaire on the patients’ demographic characteristics.
i.e. patients assumed short-lived (less than 1 month) and The questionnaire was revised after the pilot study. The pilot
patients diagnosed with dementia were excluded. In addition, study also confirmed that the bachelor nursing education
patients experiencing language difficulties, being scheduled programme had an infrastructure that enabled the collection
for operations/examinations or unfit to participate were also of data.
excluded. As a large number of students was involved in data collec-
On the screening days the students filled in the question- tion for the present study, the data quality might be ques-
naire for each patient, including questions about age, sex, tioned. An inter-rater agreement study on age, weight and
length of hospital stay, weight, height, BMI and nutritional height was therefore carried out. Two nursing students (stu-
risk. The students measured weight and height whenever pos- dents 1 and 2; S1 and S2) familiar with the ordinary screening
sible, and screened the patients for nutritional risk. A specially study, but not a part of it, were trained to collect data for the
prepared manual instructed the students on how to fill in the agreement study. On the third and fourth screening days,
questionnaire and use the screening form properly. The stu- shortly after the ordinary screening was completed, S1 and
dents usually collected the data in pairs, making it possible S2 independently of each other screened repeateadly thirty
for them to verify each other’s work. Two individuals central patients on seven wards. Data collected from S1 and S2
to the research project were available to the students at the were later merged with the results of the ordinary screening
hospital on all screening days. for further analysis.

Data collection Sample size

Anthropometric measurements. Weight was measured After a literature review and discussions with experts in the
without shoes and outer clothes in either a standing or field, the proportion of elderly nutritionally at risk was
sitting position to the nearest 0·1 kg with the weight assumed to be 30 %. According to the standard statistical
apparatus available on the different wards, following usual power calculations, a total of 165 patients were needed to
hospital practice. Height was measured to the nearest 1 cm detect this large proportion with a 95 % CI of 10 % or less.
https://doi.org/10.1017/jns.2015.8 Published online by Cambridge University Press

with a non-elastic measuring tape either in a standing To account for a possible clustering effect within wards, an
position against a wall without shoes or alternatively with the intra-class correlation coefficient of 0·3 was assumed. The
half arm-span method if the patients had problems minimum number of patients required in the study to detect
standing(23), a reliable substitute for standing height for the a prevalence of 30 % nutritionally at risk with a 95 % degree
elderly(23,24). BMI was calculated as weight (kg) divided by of confidence with a true population estimate between 25 and
the square of height (m). The age-independent cut-off values 35 % was then estimated to be 522. Subsequently, on each
presented by the WHO(25) were used when categorising ward the number of elderly patients proportional to the
patients’ BMI. ward size was consecutively included in the sample. The size
of ward was defined as the daily average number of elderly
patients based on the records from the last 6 months provided
Assessment of nutritional risk. The translated version(19) of by the hospital’s analysis department. Sampling stopped on
(26)
the validated Nutritional Risk Screening 2002 (NRS2002) each ward when the intended number of patients was reached.
form from 2009 was used to identify patients nutritionally at
risk. The screening form is recommended by the European
Data analysis
Society for Clinical Nutrition and Metabolism (ESPEN)(17)
and the Norwegian Directorate of Health(19) for use in the Demographic and clinical characteristics were described as
hospital setting. The NRS2002 aims to detect patients who mean values and standard deviations or as frequencies and
3
 journals.cambridge.org/jns

percentages, as appropriate. Patient characteristics between on seven wards. Of 1059 patients with known participation sta-
those nutritionally at risk and not at risk were compared by tus, 145 patients (14 %) declined participation, while 390 (37 %)
a t test for independent samples for continuous variables were excluded according to predefined criteria. In total, 508
and Fisher’s exact test or χ2 test for categorical variables. patients (49 %) participated on the nine screening days. As a
The prevalence of nutritional risk was estimated as sug- consecutive inclusion of patients was performed, a somewhat
gested by Cochran(22) in the following way: a proportion of low participation rate does not affect the data quality. As esti-
patients nutritionally at risk in each stratum (ward), ph, was esti- mated by the intra-class correlation coefficient, the cluster effect
mated first; here h = 1,2, . . ., 20 is the ward indicator. Then in the data was only 5·4 %, which is considerably lower than the
weights Wh were defined as the ratio of a ward size Nh to 30 % assumed in power calculations. Consequently, a slightly
the total, defined as sum of all Nh, i.e. lower sample size than expected (n 522) could not influence
the precision of the prevalence estimate.
Nh
Wh = ,
N
Patient characteristics
where N = ∑h Nh. Then the weighted prevalence was calcu-
Patient characteristics are outlined in Table 1. Of a total of 508
lated as p = ∑h Wh ph. The variance of estimated prevalence
patients in the sample, 201 (39·6 %) were nutritionally at risk,
was then defined as
252 (49·6 %) were not nutritionally at risk, while nutritional risk
1
Nh2 (Nh − nh ) ph (1 − ph ) was unknown in fifty-five cases (10·8 %). Reasons for
var (p) = , unknown nutritional risk were missing data on weight for
N2 h Nh − 1 nh − 1
twelve of the patients (21·8 %), while eleven (20 %) could
where nh is the number of patients sampled in ward h. not recall previous weight. For the remaining thirty-two
Agreement in age, weight and height between the three patients (58·2 %), the students had not filled out the screening
students (S1, S2 and nursing students performing ordinary form correctly. There were no statistically significant differences
screening) was assessed by Bland–Altman analysis, where in mean age and sex between patients nutritionally at risk and
95 % limits of agreement were constructed. The 95 % limits patients not nutritionally at risk (Table 1). Notably, patients
of agreement define an interval in which 95 % of differences nutritionally at risk had been hospitalised for longer on the
between two scoring populations would lie. The acceptable day the measurements were taken and had lower average
limits were set a priori to ±1 year in age, ±2 kg in weight weight and BMI compared with the patients not at risk
and ±3 cm in height. Bias, defined as the mean difference (Table 1); all differences were statistically significant (P < 0·001).
between measurements of two students, was assessed by one- WHO BMI cut-off values(27) identified 6·5 % as underweight,
sample t test. 48·0 % as of normal weight and 45·5 % as overweight.
The statistical program IBM SPSS Statistics version 20 for
Windows was used for statistical analysis. P values below 0·05
Inter-rater agreement study
were considered statistically significant. All tests were two-sided.
Anonymous data files were analysed by a statistician. Descriptive analysis did not show any considerable differences
in mean age, weight and height (Table 2). Consequently, there
was no significant bias between pairs of students. Differences
Ethics between S1 and S2 were marginal. Deviations between the
The present study was conducted according to the guidelines nursing students performing ordinary screening and S1 or
laid down in the Declaration of Helsinki and procedures S2 were somewhat larger. The 95 % limits of agreement
involving human patients were approved by the university hos- were slightly wider than the prespecified acceptable limits.
https://doi.org/10.1017/jns.2015.8 Published online by Cambridge University Press

pital’s Internal Privacy Commission. Verbal informed consent These deviations, however, were caused by only a few values,
was obtained from all patients. Verbal consent was witnessed as identified by assessing the Bland–Altman plots.
and formally recorded. As the screening data were anonymous,
the study was exempted from review by the Regional Prevalence of nutritional risk
Committees for Medical and Health Research Ethics (refer-
ence no. 2011/2088 A). The researchers received anonymous- The prevalence of nutritional risk was calculated based on 453
ly completed questionnaires and screening forms from the patients (89·2 % of the total sample) where the nutritional risk
students and never met the patients. The ClinicalTrials.gov was available. The prevalence was estimated to be 45·4 (95 %
ID is NCT01977950 (http://www.clinicaltrials.gov). CI 41·7, 49·0 %) (Table 3). Detailed estimates of the number
of patients nutritionally at risk on each ward are presented in
Table 3. The prevalence rates ranged between 20·0 and 65·0 %
Results on different wards.
Participation
Discussion
All elderly patients were approached on the nutritional screening
days (Fig. 1). Only approximate information on participation The present results demonstrate that undernourishment is a
status was known due to some students’ incomplete reporting serious public health problem among the hospitalised elderly
4
 journals.cambridge.org/jns

Table 1. Patient characteristics

(Mean values and standard deviations, or numbers of subjects and percentages)

Total sample Nutritionally at risk Nutritionally not at risk

Mean SD Mean SD Mean SD P

Subjects –
n 508 201 252
% 39·6 49·6
Sex (n) 502 199 249 0·506*
Men
n 257 98 131
% 51·2 42·8 57·2
Women
n 245 101 118
% 48·8 46·1 53·9
Age (years) (n) 505 201 250
79·6 6·4 80·1 6·1 79·4 6·4 0·262†
Length of stay (d) (n) 498 198 247
5·3 6·3 6·0 7·8 4·6 4·9 0·023†
Body weight (kg) (n) 492 200 249
71·3 16·5 63·8 15·7 76·6 15·1 <0·001†
BMI (kg/m2) (n) 492 199 250
24·9 4·9 22·4 4·6 26·7 4·4 <0·001†
BMI (WHO categories)(25) <0·001‡
Underweight: ≤18·49 kg/m2
n 32 32 0
% 6·5 100 0
Normal weight: 18·5–24·99 kg/m2
n 236 115 103
% 48·0 52·8 47·2
Overweight: ≥25 kg/m2
n 224 52 147
% 45·5 26·1 73·9

* Fisher’s exact test.

† t Test for independent samples.
‡ χ2 test.

in a modern Western society. For the total sample, the esti- overrepresented, making it unclear if the numbers are repre-
mated prevalence of nutritional risk was as high as 45 %, sug- sentative. Further, due to possible similarities in patient
gesting that nearly half of the elderly patients without dementia characteristics within the same ward, the presence of a cluster
were in a need of appropriate nutritional treatment. The find- effect within each ward was assumed in the power calculations
ings suggest that much can be done to improve the nutritional in the present study. Power calculations taking into account
status of the hospitalised elderly. Defining ways to prevent and such a cluster effect correctly result in sample sizes larger
treat this condition effectively in the hospital setting should than those of standard power calculations, assuring an
therefore be given immediate high priority. This is the first adequate number of patients in the study. As estimated by
prevalence study on this scale conducted among the hospita- the intra-class correlation coefficient, the cluster effect in our
https://doi.org/10.1017/jns.2015.8 Published online by Cambridge University Press

lised elderly in Norway. data was considerably lower than that assumed in power calcu-
The major strengths of the present study are its proper stat- lations. Thus, even though somewhat smaller than planned,
istical sampling design and the adequate power calculations the sample of the present study can be considered sufficient
that were carried out before the study. The differences in for a reliable prevalence estimate.
the patients’ diagnoses on different wards, comprising relative- The study sample comprised nearly all somatic medical and
ly homogeneous units, make the stratified sampling a preferred surgical wards at the university hospital, in addition to four
technique in the hospital population. This sampling technique associated wards. The hospital offers locally based specialist
ensures sufficient representation of each ward, which might be healthcare services as well as services that are more specialised.
difficult to achieve with simple random sampling. In addition, In this way, the sample covers a heterogeneous population of
it tends to produce more precise estimates of population para- elderly hospitalised patients with a large variety of potential
meters as compared with simple random sampling, since the diagnoses. Moreover, the heterogeneity of the patient popula-
variances of the entire sample are based on the variances with- tion makes it comparable with Norwegian society as a whole.
in each stratum(28). Even though other studies with large sam- Unfortunately, for ethical and practical reasons it was not pos-
ple sizes have produced prevalence estimates with high sible to include the patients from the psychiatric division and
precision(7,8,10–12), they have either sampled from certain patients diagnosed with dementia in the present study. The
types of wards or by consecutively including all admitted estimated prevalence therefore cannot be generalised to the
patients. Consequently some wards may have been under- or entire elderly population at the university hospital.
5
 journals.cambridge.org/jns

Table 2. Descriptive statistics of age, weight and height collected by included(14). The present results are strengthened by the fact
nursing students (performing ordinary screening), student 1 (S1) and
that similar figures have been shown in Europe(13).
student 2 (S2), and bias between students (including P values for
one-sample t tests) and 95 % limits of agreement (LoA) The hospital ward composition may make an impact on the
total prevalence estimate, as our data indicate that the propor-
Age (years) Weight (kg) Height (m) tion of patients nutritionally at risk varied in wards. It has been
Nursing students argued that large hospitals tend to differ from other hospitals
n 30 29 30 in terms of ward composition by providing more specialised
Mean 77·90 69·97 1·68 care, which could affect the case mix of the studied population
5·20 15·07 0·12
S1
SD
and further effect the prevalence estimate(29). However, by
n 30 29 30 providing more specialised care in addition to locally based
Mean 77·83 70·89 1·66 specialist healthcare services, large hospitals usually handle a
SD 5·38 15·35 0·13 wider variety of potential diagnoses, and sampling from
S2
n 30 29 30
large hospitals will therefore ensure more representative data.
Mean 77·83 70·89 1·66 The results from other Norwegian(3–5) and European stud-
SD 5·38 15·34 0·13 ies(6–13) reporting the prevalence of undernutrition and the risk
Nursing students v. S1 of undernutrition among the hospitalised elderly have shown
Bias 0·07 –0·92 0·03
P 0·961 0·818 0·415
variable prevalence rates. This is presumably due to methodo-
95 % LoA –1·71, 1·85 –3·56, 1·71 –0·17, 0·23 logical differences and weaknesses, and the results are often
Nursing students v. S2 not representative of the studied population and/or can sel-
Bias 0·07 –0·92 0·03 dom be generalised to a larger part of the elderly population
P 0·961 0·818 0·437
95 % LoA –1·71, 1·85 –3·56, 1·71 –0·14, 0·19 at the hospital studied. Different measurement methods,
S1 v. S2 such as screening tools and BMI cut-offs, are also often
Bias 0·00 0·003 –0·001 employed, which makes it challenging and even impossible
P 1·00 0·999 0·976
to compare the results. Three studies that have employed
95 % LoA N/A –0·17, 0·17 –0·09, 0·09
the NRS2002 to identify nutritional risk have reported either
N/A, not applicable. lower (22–28 %)(12), higher (54 %)(6) or similar (42 %)(13)
rates compared with the present results. However, the pub-
Generalisations from cross-sectional studies are always chal- lished lower rate only reflects nutritional risk on hospital
lenging. The results, however, clearly show the extent of the admission(12). As nutritional status often deteriorates during
problem of undernutrition and the risk of undernutrition hospital stays(30) and undernourished patients in general are
among the hospitalised elderly in Norway today, although hospitalised longer(1,15), rates on admission will usually be
the estimated prevalence would probably have been even lower than estimates covering the entire hospitalised popula-
higher if elderly patients diagnosed with dementia had been tion. Moreover, none of the three studies used proper

Table 3. Total prevalence estimate and proportions of patients nutritionally at risk on each ward
(Numbers of subjects and percentages)

Total sample Nutritionally at risk Unknown

Ward n % n % n %

Orthopaedics 1 40 7·9 8 20·0 8 20·0

https://doi.org/10.1017/jns.2015.8 Published online by Cambridge University Press

Ear–Nose–Throat/Gynaecology 15 3·0 7 46·7 2 13·3

Vascular/Thorax 18 3·5 10 55·6 2 11·1
Gastro Surgery lower 26 5·1 13 50·0 2 7·7
Gastro Surgery upper 18 3·5 9 50·0 0
Urology 22 4·3 13 59·1 0
Orthopaedics 2 46 9·1 19 41·3 9 19·6
Emergency Medicine 23 4·5 7 30·4 1 4·3
Neurology + Endocrinology 28 5·5 9 32·1 1 3·6
Infectious Medicine 1 37 7·3 14 37·8 8 21·6
Neurology/Stroke 14 2·8 4 28·6 2 14·3
Infectious Medicine 2 13 2·6 4 30·8 0
Rehabilitation Neurology 36 7·1 10 27·8 6 16·7
Renal Medicine 15 3·0 7 46·7 1 6·7
Heart Medicine 30 5·9 8 26·7 3 10·0
Lung Medicine 50 9·8 30 60·0 4 8·0
Cardiology Medicine 37 7·3 9 24·3 2 5·4
Cardiac Monitoring 6 1·2 1 16·7 2 33·3
Haematology 14 2·8 6 42·9 1 7·1
Specialised Short-Term Unit 20 3·9 13 65·0 1 5·0
Total prevalence estimate (%) 45·4
95 % CI (%) 41·7, 49·0

6
 journals.cambridge.org/jns

statistical sampling methods for estimating prevalence involved in data collection might be seen as a shortcoming
rates(6,12,13), and only medical wards were included in the sam- of the study. We can also assume that the students had limited
ples of the studies reporting either lower or higher rates(6,12). research experience. However, the inter-rater agreement study
The studies that report much higher prevalence rates com- exhibited an acceptable quality of the screening data.
pared with that in the present study have often used the Moreover, individuals central to the planning and conducting
screening form Mini Nutritional Assessment Tool of the present study were experienced in using students for
(MNA)(3,6–9,11), which has been shown to identify more the collection of research data(36). The students also recieved
patients nutritionally at risk compared with the supervision before each nutritional screening day, to secure
NRS2002(6,31,32). The MNA is specifically developed and the data collection.
recommended for use on elderly patients(17,33). However, In conclusion, the prevalence of nutritional risk among eld-
being of old age is also taken into account in the erly without dementia was high, suggesting that a large propor-
NRS2002(26). The NRS2002 may also be a more appropriate tion of hospitalised elderly patients are in need of nutritional
screening form for use with the acutely diseased elderly treatment. The present study demonstrates how a close multi-
since the MNA does not consider the effect of stress metab- disciplinary collaboration between a university hospital and a
olism on nutritional needs(6,32,34,35). Furthermore, as the nursing bachelor education programme can facilitate the con-
screening was part of the students’ clinical training and educa- ducting of a larger research study by involving students in
tion in the present study, it was important to choose a screen- research activities.
ing tool commonly used in the hospital setting in Norway,
recommended by the Norwegian Directorate of Health.
We observed no age difference between patients nutritional- Acknowledgements
ly at risk and patients not at risk. This was somewhat surpris-
We thank all the participants and all the nursing students for
ing since advanced age is a known risk factor of
their participation. We would also like to thank the nursing
undernourishment(14). On the other hand, this could be just
students’ clinical supervisors for their contributions, and the
an effect of the inclusion criteria (age ≥70 years), since younger
practicum coordinators at the university college and at the uni-
patients were not included in the present study sample, com-
versity hospital. Our thanks also go to the individuals involved
pared with other studies that have found an effect of age on
in planning the present study.
undernourishment(10,12,13). In the present study we did not
Financial support for the present study was received from
control for other patient characteristics, for example, multi-
the South-Eastern Norway Regional Health Authority (grant
morbidity, and the effect of age might have been dominated
no. 2719007), internal funding from the participating univer-
by other factors.
sity hospital (grant no. 2619013) and the Department of
A limitation of the present study is that nutritional risk was
Health, Nutrition and Management (internal funding),
unknown in 11 % of the sample, most often due to incomplete
Faculty of Health Sciences, Oslo and Akershus University
screening forms. We observed no systematic incomplete data;
College of Applied Sciences. K. A. obtained funding. The
hence the impact of missing data on the prevalence estimate in
financial contributors had no role in the design, analysis or
the present study is considered to be minor. Another reason
writing of this article.
for missing data on nutritional risk was that a few patients
K. A. was project leader; K. A., J. S. B. and K. S. designed
could not recall previous weight, and the question of
and conducted the pilot study; K. A., J. S. B., H. K .E.,
whether screening forms that require data on recent weight
K. H. and K. S. designed the research; H. K. E. and
loss, like NRS2002, are suitable for the entire elderly hospita-
K. S. conducted the research; J. S. B. performed the statistical
lised population, can be raised. However, as we excluded
analysis; K. A., J. S. B., H. K. E., K. H. and K. S. wrote the
patients diagnosed with dementia, this information was lacking
https://doi.org/10.1017/jns.2015.8 Published online by Cambridge University Press

paper; K. A., J. S. B., H. K. E., K. H. and K. S. had primary

for only a few patients in the present study. On some wards a
responsibility for the final content; all the authors read and
greater number of patients were excluded than on others.
approved the final manuscript.
There is a risk that attitudes among the ward nursing staff
There are no conflicts of interest.
may have led to unnecessary exclusion of patients found
unfit to participate, which could have an impact on the esti-
mate. Unfortunately, there was no detailed information on
References
patient exclusion in the present study, and future studies
should note the importance of obtaining such information. 1. Stratton RJ, Green CJ & Elia M (2003) Disease-Related Malnutrition:
The multidisciplinary collaboration was essential for carrying An Evidence-Based Approach to Treatment, 1st ed. Wallingford: CABI.
2. Rasmussen HH, Holst M & Kondrup J (2010) Measuring nutrition-
out the present study. Using students in this clinical study al risk in hospitals. Clin Epidemiol 2, 209–216.
enabled a collection of a large dataset using limited resources. 3. Ranhoff A, Gjoen A & Mowe M (2005) Screening for malnutrition
Moreover, the students gained insight into how a large multi- in elderly acute medical patients: the usefulness of MNA-SF. J Nutr
disciplinary research study is planned and carried out, and they Health Aging 9, 221–225.
acquired important research-based professional knowledge 4. Mowe M & Bohmer T (1991) The prevalence of undiagnosed pro-
tein–calorie undernutrition in a population of hospitalized elderly
and training in nutritional risk screening. The screening may patients. J Am Geriatr Soc 39, 1089–1092.
also have induced increased competence among hospital 5. Mowe M, Bøhmer T & Kindt E (1994) Reduced nutritional status
ward staff. On the other hand, a large number of students in an elderly population (>70 y) is probable before disease and
7
 journals.cambridge.org/jns

possibly contributes to the development of disease. Am J Clin Nutr 23. Kwok T & Whitelaw MN (1991) The use of armspan in nutritional
59, 317–324. assessment of the elderly. J Am Geriatr Soc 39, 492–496.
6. Holst M, Yifter-Lindgren E, Surowiak M, et al. (2012) Nutritional 24. Nygaard HA (2008) Measuring body mass index (BMI) in nursing
screening and risk factors in elderly hospitalized patients: associ- home residents: the usefulness of measurement of arm span. Scand J
ation to clinical outcome? Scand J Caring Sci 27, 953–961. Primary Health Care 26, 46–49.
7. Vanderwee K, Clays E, Bocquaert I, et al. (2011) Malnutrition and 25. World Health Organization (2006) BMI classification. http://apps.
nutritional care practices in hospital wards for older people. J Adv who.int/bmi/index.jsp?introPage=intro_3.html (accessed October
Nurs 67, 736–746. 2013).
8. Vanderwee K, Clays E, Bocquaert I, et al. (2010) Malnutrition and 26. Kondrup J, Rasmussen HH, Hamberg O, et al. (2003) Nutritional
associated factors in elderly hospital patients: a Belgian cross- Risk Screening (NRS 2002): a new method based on an analysis
sectional, multi-centre study. Clin Nutr 29, 469–476. of controlled clinical trials. Clin Nutr 22, 321–336.
9. Volkert D, Saeglitz C, Gueldenzoph H, et al. (2010) Undiagnosed 27. Beck AM & Ovesen L (1998) At which body mass index and
malnutrition and nutrition-related problems in geriatric patients. degree of weight loss should hospitalized elderly patients be consid-
J Nutr Health Aging 14, 387–392. ered at nutritional risk? Clin Nutr 17, 195–198.
10. Pirlich M, Schütz T, Norman K, et al. (2006) The German hospital 28. Kelsey JL, Whittemore AS, Evans AS, et al. (1996) Methods in
malnutrition study. Clin Nutr 25, 563–572. Observational Epidemiology, 2nd ed. New York: Oxford University
11. Söderström L, Thors Adolfsson E, Rosenblad A, et al. (2013) Press.
Mealtime habits and meal provision are associated with malnutrition 29. Westergren A, Wann-Hansson C, Borgdal EB, et al. (2009)
among elderly patients admitted to hospital. Clin Nutr 32, 281–288. Malnutrition prevalence and precision in nutritional care differed
12. Imoberdorf R, Meier R, Krebs P, et al. (2010) Prevalence of under- in relation to hospital volume – a cross-sectional survey. Nutr J
nutrition on admission to Swiss hospitals. Clin Nutr 29, 38–41. 8, 20.
13. Lucchin L, D’Amicis A, Gentile MG, et al. (2009) An Italian investi- 30. McWhirter JP & Pennington CR (1994) Incidence and recognition
gation on nutritional risk at hospital admission: The PIMAI (Project: of malnutrition in hospital. BMJ 308, 945–948.
Iatrogenic MAlnutrition in Italy) study. e-SPEN 4, e199–e202. 31. Raslan M, Gonzalez MC, Gonçalves Dias MC, et al. (2010)
14. Morley JE (1997) Anorexia of aging: physiologic and pathologic. Comparison of nutritional risk screening tools for predicting clinical
Am J Clin Nutr 66, 760–773. outcomes in hospitalized patients. J Nutr 26, 721–726.
15. Norman K, Pichard C, Lochs H, et al. (2008) Prognostic impact of 32. Bauer JM, Vogl T, Wicklein S, et al. (2005) Comparison of the Mini
disease-related malnutrition. Clin Nutr 27, 5–15. Nutritional Assessment, Subjective Global Assessment, and
16. World Health Organization (2002) Active Ageing: a Policy Framework. Nutritional Risk Screening (NRS 2002) for nutritional screening
Geneva: World Health Organization. and assessment in geriatric hospital patients. Z Gerontol Geriatr 38,
17. Kondrup J, Allison SP, Elia M, et al. (2003) ESPEN guidelines for 322–327.
nutrition screening 2002. Clin Nutr 22, 415–421. 33. Guigoz Y, Vellas B & Garry PJ (1996) Assessing the nutritional sta-
18. Beck AM, Balknas UN, Furst P, et al. (2001) Food and nutritional tus of the elderly: The Mini Nutritional Assessment as part of the
care in hospitals: how to prevent undernutrition – report and guide- geriatric evaluation. Nutr Rev 54, 59–65.
lines from the Council of Europe. Clin Nutr 20, 455–460. 34. Drescher T, Singler K, Ulrich A, et al. (2010) Comparison of
19. The Norwegian Directorate of Health (2009) Nasjonale faglige retning- two malnutrition risk screening methods (MNA and NRS
slinjer for forebygging og behandling av underernæring (National Professional 2002) and their association with markers of protein malnutrition
Guidelines on Prevention and Treatment of Malnutrition). Oslo: The in geriatric hospitalized patients. Eur J Clin Nutr 64, 887–893.
Norwegian Directorate of Health. 35. Neelemaat F, Meijers J, Kruizenga H, et al. (2011) Comparison of
20. Anthony PS (2008) Nutrition screening tools for hospitalized five malnutrition screening tools in one hospital inpatient sample.
patients. Nutr Clin Pract 23, 373–382. J Clin Nurs 20, 2144–2152.
21. van Bokhorst-de van der Schueren MAE, Guaitoli PR, Jansma EP, 36. Sortland K, Gjerlaug AK & Harviken G (2013) Vektdokumentasjon,
et al. (2014) Nutrition screening tools: does one size fit all? A sys- kroppsmasseindeks, måltidsfrekvens og nattefaste blant eldre sykeh-
tematic review of screening tools for the hospital setting. Clin jemsbeboere – en pilotstudie (Weight documentation, body mass
Nutr 33, 39–58. index, meal frequency and overnight fast among elderly in nursing
22. Cochran WG (1963) Sampling Techniques, 2nd ed. New York: Wiley. home – a pilot study). Vard Nord Utveckl Forsk 33, 41–45.
https://doi.org/10.1017/jns.2015.8 Published online by Cambridge University Press

8
 journals.cambridge.org/jns

Appendix 1. The Norwegian version of Nutritional Risk Screening 2002 (NRS2002) from 2009 (in English)(19)

Initial screening: Yes No

1. Is BMI <20·5 kg/m2?
2. Has the patient lost weight within the last few weeks?
3. Has the patient had a reduced dietary intake in the last weeks?
4. Is the patient severely ill?
Yes: If the answer is ‘Yes’ to any question, the final screening is performed
No: If the answer is ‘No’ to all questions, the patient is rescreened at weekly intervals. If the patient is scheduled for a major operation, a preventive nutritional
care plan is considered to avoid the associated risk status
Final screening:
Score Nutritional status Score Severity of disease
0 Normal nutritional status 0 Not ill
1 Weight loss 5–10 % in the last 3 months 1 A patient with a chronic disease or a
and/or dietary intake of 50–75 % of patient who has undergone minor
requirement for a week or more surgery. Studies have been conducted
on patients with liver cirrhosis, kidney
failure, chronic lung disease and
cancer, and on patients with collum
femoris fracture, after
cholecystectomy, and laparoscopic
surgery
2 BMI 18·5–20·5 kg/m2 and/or recent weight 2 A patient with significantly reduced
loss 10–15 % in the last 3 months and/or general condition due to illness.
dietary intake of 25–50 % of requirement Studies have been conducted on
for a week or more patients with severe pneumonia,
inflammatory bowel disease with fever,
acute renal failure, major surgery such
as colectomy and gastrectomy, ileus,
anastomosis leakage and repetitive
operations
3 BMI ≤18·5 kg/m2 and/or recent weight loss >15 % 3 A patient is seriously ill. Studies have
in the last 3 months and/or dietary intake of 0–25 % been conducted on patients with large
of requirement for a week or more apoplexy, severe sepsis, intensive care
unit patients (APACHE >10), bone
marrow transplants, major head
injuries, burns >40 % and severe acute
pancreatitis
Explanation of the final screening:
The patient scores from 0 to 3 for nutritional status
The patient scores from 0 to 3 for disease severity
A score of 1 is added for patients older than 70 years
If total score is ≥3, the patient is nutritionally at risk, and nutritional treatment must be initiated
If total score is <3, the patient is not nutritionally at risk. Screening must be repeated after 1 week

APACHE, Acute Physiology and Chronic Health Evaluation.

https://doi.org/10.1017/jns.2015.8 Published online by Cambridge University Press