117-jhn 12314
117-jhn 12314
CLINICAL NUTRITION
Barriers to food intake in acute care hospitals: a report of
the Canadian Malnutrition Task Force
H. Keller,1 J. Allard,2 E. Vesnaver,3 M. Laporte,4 L. Gramlich,5 P. Bernier,6 B. Davidson,7
D. Duerksen,8 K. Jeejeebhoy9 & H. Payette10
1
Schlegel-UW Research Institute for Aging, University of Waterloo, Waterloo, ON, Canada
2
Department of Medicine, University Hospital Network, University of Toronto, Toronto, ON, Canada
3
University of Guelph, Guelph, ON, Canada
4
Réseau de Santé Vitalité Health Network, Cambellton, NB, Canada
5
Department of Medicine, University of Alberta, Alberta Health Services, Edmonton, AB, Canada
6
Jewish General Hospital, Montréal, QC, Canada
7
Canadian Malnutrition Task Force, Canadian Nutrition Society, Toronto, ON, Canada
8
Department of Medicine St-Boniface Hospital, University of Manitoba, Winnipeg, MB, Canada
9
Department of Medicine St-Michael Hospital, University of Toronto, Toronto, ON, Canada
10
Centre de recherche sur le vieillissement, CSSS-IUGS, Faculté de Médecine et des Sciences de la Santé, Universit
e de Sherbrooke, Sherbrooke,
QC, Canada
Keywords Abstract
dietetics, food service, malnutrition, meals,
nutrition screening. Background: Poor food intake is common in acute care patients and can
exacerbate or develop into malnutrition, influencing both recovery and
Correspondence outcome. Yet, research on barriers and how they can be alleviated is lack-
H. Keller, Schlegel-UW Research Institute for ing. The present study aimed to (i) describe the prevalence of food intake
Aging, University of Waterloo, Waterloo, N2l 3J1,
barriers in diverse hospitals and (ii) determine whether patient, care
Ontario, Canada.
or hospital characteristics are associated with the experience of these
Tel.: 519 888 4567
E-mail: [email protected] barriers.
Methods: Patients (n = 890; 87%) completed a validated questionnaire on
How to cite this article barriers to food intake, including perceptions of food quality, just before
Keller H., Allard J., Vesnaver E., Laporte M., their discharge from a medical or surgical unit in each of 18 hospitals across
Gramlich L., Bernier P., Davidson B., Duerksen D., Canada. Scores were created for barrier domains. Associations between these
Jeejeebhoy K., Payette H. (2015) Barriers to food barriers and selected patient characteristics collected at admission or
intake in acute care hospitals: a report of the
throughout the hospital stay and site characteristics were determined using
Canadian Malnutrition Task Force. J Hum Nutr
bivariate analyses.
Diet. 28, 546–557
doi:10.1111/jhn.12314 Results: Common barriers were being interrupted at meals (41.8%), not
being given food when a meal was missed (69.2%), not wanting ordered
food (58%), loss of appetite (63.9%) and feeling too sick (42.7%) or tired
(41.1%) to eat. Younger patients were more likely (P < 0.0001) to report
being disturbed at meals (44.6%) than older patients (33.9%) and missing a
meal for tests (39.0% versus 31.0%, P < 0.05). Patients who were malnour-
ished, women, those with more comorbidity, and those who ate <50% of
the meal reported several barriers across domains.
Conclusions: The present study confirms that barriers to food intake are
common in acute care hospitals. This analysis also identifies that specific
patient subgroups are more likely to experience food intake barriers.
Because self-reported low food intake in hospital was associated with several
barriers, it is relevant to consider assessing, intervening and monitoring bar-
riers to food intake during the hospital stay.
dietitian consultation had occurred. Food and fluid intake sons were based on a subset of participants as a result of the
were estimated by patients or families present at the meal applicability of the item in question. For example, the barrier
using the nutritionDay Form(3,21,24) for one meal on ‘did not get help when needed’ was only relevant for patients
3 days of the first week of stay and recorded as 0%, 25%, who ‘needed help with meals’. The primary analysis was to
50% or 100% consumption. These records were averaged determine associations among hospital, care and patient
and dichotomised for analysis as <50% and ≥50% con- characteristics and these food access barriers. Total domain
sumption of provided food. Hospital size and ward type, scores were used as the dependent variable, expect for the or-
as well as the proportion of food provided to patients ganisational items, where associations with each organisa-
that was outsourced were provided in a hospital question- tional barrier were calculated.
naire completed by the site coordinators. All sites pro- Independent sample t-tests or analysis of variance F
vided a tray service with pre-plated meals; some with a tests were used to determine the mean domain score for
selection of meals 24 h before the meal. Some hospitals a given categorised characteristic; SAS SURVEYREG was used
had no kitchens and trays were assembled at a regional to account for the clustered design and standard errors
provider. Porters, healthcare aides or other nursing staff are presented based on these estimates. Where required,
passed trays as part of their role; no hospital had a desig- post-hoc pairwise comparisons of means were performed
nated hostess with specialised skills or activities to sup- with Bonferroni corrections to account for multiple com-
port food intake. Association variables were chosen by parisons. For the organisational items where comparisons
the investigators in the present study for this exploratory were made at the item level, the Rao–Scott chi-squared
analysis because they were theoretically meaningful to test (corrected for clustered survey design) was used.
understanding potential food access issues(4,7).
A patient questionnaire used in previous research(7) to
Results
identify barriers to food intake and assess patient percep-
tions of their experience, including food quality and hun- Of the 1022 participants, 890 (87%) at least partially com-
ger, was modestly altered for the Canadian context. pleted the questionnaire. This subgroup, described below,
Specifically, a version omitting questions on food choice is the basis for this analysis. Among those who completed
was created for those hospitals that did not have selective some or all of the survey, 12.5% (n = 111) missed at least
menus. Patients completed this questionnaire (for 13 of one question, whereas 87.5% (n = 779) provided complete
18 hospitals reporting; 44.5% interviewer administered), answers to all questions. Table 1 provides the number per
just before their discharge. question completed. A patient’s domain score was only
The questionnaire consists of several items, which are calculated if all items within the domain were completed.
categorised into six domains: organisational, choice, hun- The items from the choice domain are limited to the 389
ger, eating difficulties, food quality and effects of illness. respondents with complete answers from the nine hospi-
Individual items were dichotomised from their original tals that provided a selective menu.
four-point scale into categories, ‘affected’ (score = 1) or The proportion of total nonrespondents in different
‘not affected’ (score = 0) as per the original scale(7) hospitals varied from 0% to 32%. Nonrespondents were
(M. Gulliford, personal communication). Affected items slightly older (66.8 years versus 63.9 years, P = 0.04),
indicate that the patient experienced the barrier to food more likely to be medical patients (81% versus 67.3%,
intake. All ‘affected’ items (i.e. score = 1) in a domain P = 0.002) and to be malnourished at admission (SGA-B:
were summed, resulting in the domain score. Thus, a 43.4% and C 19.5% versus 32.1% and 10.6%, respec-
higher domain score means that the patient was more tively, P <0.001) and to have cancer (28.6% versus
likely to experience barriers to food intake. Exploratory 18.9%, P = 0.008) than respondents. There was no differ-
factor analysis (not reported) indicated that questions ence in educational level, length of stay, gender or the
allocated to the organisational and choice domains were number of diagnoses by respondent status.
not consistent with those identified by the investigators(7). Table 2 presents the prevalence of key patient charac-
As a result, domain scores were not created for these two teristics; for example, 52.6% of patients included in the
areas. Descriptive analyses were only completed for choice analysis were over the age of 65 years. Most of the partic-
items because half of the hospitals did not have a selective ipants used in this analysis were from academic hospitals
menu. (63.7%) and medical units (67.3%). Slightly more than
Descriptive statistics were completed on patient question- half of participants were over the age of 65 years (52.6%)
naire responses and domains using proportions and the and male (52.2%). Most patients had a Charlson Comor-
mean (SE) for domain scores; SAS SURVEYFREQ procedures bidity Index (CCI) less than or equal to two (58.9%) and
were used to take into account the clustered design (SAS, ver- 18.6% had an active cancer. Malnutrition (SGA-B or -C)
sion 9.3; SAS Institute Inc., Cary, NC, USA). Some compari- was relatively common at 42.4%, although relatively few
Organisational
Disturbed by activities, noises or unpleasant smells 345/866 38.9
Interrupted by the hospital staff 371/887 41.8
Missed meals by not being available when they were served 174/876 19.9
Missed meals because of avoiding food for tests 306/881 34.7
When meals missed, not given hospital food by staff† 251/363 69.2
Did not get help when needed (restricted to patients who possibly needed help, n = 218‡) 92/218 42.2
Does not want food that has been ordered§ 218/376 58.0
Did not receive ordered food§ 102/369 27.6
Choice
Meals not served at times that suit patient 104/877 11.9
Do not understand how to complete the menu selection sheet§ 15/387 3.9
Not being able to choose preferred foods§ 90/386 23.3
Not enough information on menu to make selection§ 140/379 36.9
Hunger
Visitors bring in food because patient is hungry 262/869 30.1
Become hungry between meals that are too far apart 212/869 24.4
Felt hungry but could not ask staff for food 100/869 11.5
Felt hungry and wanted something to eat but no food was available from the hospital 106/869 12.2
Eating difficulties
In an uncomfortable position to eat 236/867 27.2
Difficulty reaching food 171/867 19.7
Difficulty cutting up food 140/867 16.1
Difficultly opening packets/unwrapping food 261/867 30.1
Difficulty feeding self 75/867 8.7
Not enough time to eat all the food 64/867 7.4
Needed help to eat meals 68/867 7.8
Quality/satisfaction with food; dissatisfied with:
Taste 242/841 28.8
Appearance 137/841 16.3
Smell 153/841 18.1
Portion size 163/841 19.4
Temperature of food 177/841 21.0
Effects of illness on food intake
Loss of appetite 531/862 63.9
Sickness 368/862 42.7
Pain 322/862 37.4
Tired 354/862 41.1
Worried 216/862 25.1
Depressed 173/862 20.0
Breathing difficulties 147/862 17.1
Chewing or swallowing difficulties 131/862 15.2
⁄
Number with positive answer/number who answered the question.
†
Patients answered ‘did not miss a meal’ excluded.
‡
Patients answered ‘did not need any help’ excluded.
§
Only for hospitals with selective menus. Questions adapted from Naithani et al.(7).
patients had a dietitian consultation (27%). Approxi- Common barriers (>30% affected) included: not given
mately one-third (29.8%) consumed less than 50% of food when meal was missed (69.2%); did not get help
their tray during the first week of their admission. when needed (42.2%); did not want the food they had
ordered (42%); meal interrupted by staff (41.8%); being
disturbed by activity, noise or smell (38.9%); found it
Frequency of meal experiences and barriers to food intake
hard to choose foods because of insufficient information
The proportion of patients experiencing various barriers on menu (36.9%); missed meals as a result of tests
to food intake in hospital is presented in Table 1. (34.7%); and difficulty opening packages/unwrapping
Table 2 Prevalence and comparison of mean (SE) domain scores† with patient, hospital and care characteristics
Overall mean (SE) [range] 0.78 (0.07)‡ [0–4]§ 1.17 (0.09) [0–7] 1.04 (0.10) [0–5] 2.38 (0.16) [0–8]
Age (years)
<65 47.4 (421) 0.93 (0.09) 0.93 (0.09) 1.04 (0.12) 2.45 (0.16)
≥65 52.6 (468) 0.65 (0.06)*** 1.39 (0.13)** 1.04 (0.12) 2.31 (0.20)
Gender
Female 47.8(425) 0.77 (0.09) 1.29 (0.13) 1.11 (0.13) 2.85 (0.19)
Male 52.2 (464) 0.80 (0.08) 1.06 (0.09) 0.98 (0.11) 1.94 (0.17)***
SGA at admission
Well nourished (A) 57.6(511) 0.73 (0.08) 1.00 (0.09)a,b 0.90 (0.09)a 1.95 (0.16)a,b
Mild/moderate malnutrition (B) 32.1(285) 0.82 (0.07) 1.30 (0.11)b,c 1.22 (0.13)a 2.83 (0.23)b
Severe malnutrition (C) 10.3(91) 0.94 (0.12) 1.78 (0.27)a,c*** 1.23 (0.21)* 3.41 (0.29)a***
Number of diagnoses
One diagnosis 576 0.71 (0.08)a 1.07 (0.10) 1.01 (0.11) 2.18 (0.17)a
Two diagnoses 226 0.95 (0.11)a 1.35 (0.16) 1.14 (0.14) 2.76 (0.23)a
Three diagnoses 88 0.81 (0.07)* 1.36 (0.19) 0.96 (0.17) 2.70 (0.28)*
Presence of cancer
No 81.4 (723) 0.78 (0.08) 1.19 (0.10) 0.99 (0.11) 2.28 (0.18)
Yes 18.6 (165) 0.83 (0.07) 1.07 (0.17) 1.27 (0.16) 2.83 (0.24)
CCI at admission
CCI ≤ 2 58.9 (516) 0.77 (0.08) 1.04 (0.10) 0.94 (0.12) 2.22 (0.17)
CCI > 2 41.1 (360) 0.82 (0.07) 1.37 (0.12)* 1.21 (0.12) 2.62 (0.20)*
C-reactive protein
<10 g/L 29.2 (206) 0.84 (0.07) 1.00 (0.08) 0.98 (0.15) 2.11 (0.21)
≥10 g/L 71.1 (508) 0.80 (0.08) 1.39 (0.13)* 1.14 (0.10) 2.64 (0.18)*
Admission ward type
Medical 67.3 (563) 0.79 (0.08) 1.16 (0.09) 1.00 (0.11) 2.21 (0.18)
Surgical 32.7 (273) 0.73 (0.12) 1.21 (0.16) 1.07 (0.16) 2.53 (0.17)
Hospital type
Academic 63.7 (567) 0.77 (0.07) 1.11 (0.08) 1.00 (0.10) 2.49 (0.15)
Community 36.2 (323) 0.81 (0.15) 1.27 (0.21) 1.10 (0.22) 2.18 (0.35)
Percentage of outsourced food (data from 15 hospitals only)
1–49 24.9 (185) 0.65(0.13) 0.79(0.13)a 0.93 (0.25) 1.68 (0.41)
50–84 32.0 (237) 0.87 (0.13) 1.38 (0.19)a 1.23 (0.22) 2.71 (0.26)
85–100 43.0 (319) 0.83(0.14) 1.30(0.10)* 0.96 (0.16) 2.38(0.15)
Had registered dietitian visit
No 73 (650) 0.75 (0.08) 1.16 (0.10) 1.00 (0.11) 2.14 (0.15)
Yes 27 (240) 0.87 (0.09) 1.20 (0.13) 1.15 (0.13) 3.00 (0.25)***
Food intake <50% during the 1st week
No 70.2 (587) 0.76 (0.09) 1.13 (0.10) 0.85 (0.11) 1.93 (0.17)
Yes 29.8 (249) 0.81 (0.08) 1.33 (0.15) 1.51 (0.11)*** 3.38 (0.17)***
Pre-admission weight loss ≥ 5%
No 69.6 (589) 0.74(0.07) 1.03 (0.09) 0.98 (0.12) 2.17 (0.17)
Yes 30.4 (257) 0.85 (0.10) 1.29 (0.14) 1.14 (0.12) 2.80 (0.22)**
Takes oral supplements pre-admission
No 79.9 (708) 0.77(0.08) 1.09 (0.10) 1.00 (0.11) 2.26 (0.15)
Yes 20.1 (178) 0.83 (0.08) 1.47 (0.14) 1.15 (0.15) 2.85 (0.25)**
†
Higher scores on domains of hunger, eating difficulties, food quality and illness effects on food intake (columns) indicate more barriers in that
domain.
‡
Means (SE of estimate based on SURVEYREG and Taylor series linerariation).
§
Potential range on a domain score that is equal to the number of total items in the domain.
Alpha superscripts (i.e. a,b,c) indicate that a statistically significant difference exists between groups that have the same superscript.
*P <0.05, **P <0.001, ***P <0.0001. SGA, subjective global assessment; CCI, Charlson Comorbidity Index.
food (30.1%). The most common food quality barrier food was outsourced. More food quality domain barriers
was the taste of food (28.8%), whereas other sensory were seen in malnourished patients and those with reduced
problems were reported by less than 20% of patients. Ill- food intake in the first week, indicating an increased dissat-
ness effects on food intake were more common, with isfaction with food quality. High scores in the illness effects
63.9% reporting poor appetite, 42.7% reporting sickness, domain were seen for females, the malnourished (SGA-B
41.1% reporting tiredness and 37.4% reporting pain at or -C), sicker (two versus one diagnosis, >2 CCI, CRP ≥10/
meals. Visitors brought in food for 30.1% of respondents; g/L), those with reduced food intake during the first week,
however, it is unclear why this occurred (e.g. preferences and those reporting a pre-admission weight loss of >5%
not being met, not enough to eat, a special treat, etc.). and/or reporting using oral nutritional supplements pre-
admission. Patients with higher scores on the effects of ill-
ness domain were also positively associated with a dietitian
Barrier domain scores
visit [3.0 (SE 0.25) versus 2.14 (SE 0.15)] compared to
The maximum score for a domain was dependent on the those with lower scores on this domain.
number of items contained within the domain because
scores are dependent on the number of items for an indi-
Associations with organisational barriers
vidual patient that were experienced or ‘affected’ (e.g. two
items affected in the hunger domain would give a score of Organisational barrier questions were analysed individually
2 for that patient). Because the potential number of items rather than as a domain score and the results are shown in
per domain varied, the means cannot be compared directly Table 3. No association was found with hospital type or
to determine the most common barriers: hunger (mean size, diet technician visit, presence of cancer, taking oral
0.78, minimum = 0, maximum = 4), difficulties with eat- nutritional supplements or antibiotic treatment and thus
ing (mean 1.17, minimum = 0, maximum = 7), food qual- these characteristics are not provided in the table. Being
ity (mean 1.04, minimum = 0, maximum = 5) and effects disturbed during meals was more likely to occur for youn-
of illness (mean 2.38, minimum = 0, maximum = 8). To ger patients, those malnourished at admission, eating
make this comparison, we considered having any item as <50% of their food during the first week of admission, or
‘affected’ in a domain to mean that the domain was experi- being in a hospital with a higher proportion of outsourced
enced. Using this strategy, the most common domain bar- food. Similarly, the sicker, malnourished at admission, or
riers were (in order): effects due to illness (70.5%), those in hospitals with a higher proportion of outsourced
difficulties self-feeding (49.1%) and food quality (48.75%). food, were more likely to be interrupted by staff during
The hunger domain was experienced by 46.4%. their meal. Higher proportions of those reporting missing
meals because of tests were younger, malnourished at
admission, had a pre-admission weight loss of >5%, had
Associations with food intake barrier domains
<50% food intake during their first week of admission and
Hospital type, size, diet technician visit, antibiotic treat- saw the dietitian during their hospitalisation. Furthermore,
ment, and following a special diet pre-admission were not surgical patients and those with inflammation or <50%
associated with barrier domain scores and were thus not food intake during the first week of admission were more
included in Table 2. Other significant associations likely not be given replacement food when meals were
between patient, hospital or care characteristics (e.g. hav- missed. Those who did not receive help needed at meal-
ing a dietitian visit) compared to the four domains of times were also not as sick (CRP < 10 g/L) and did not see
hunger, eating difficulty, food quality and effects of illness a dietitian during their admission. Not being able to
are presented in Table 2. Means of the domains are pro- choose the foods that they liked was associated with being
vided in cells. For example, the mean hunger domain an older adult, female, malnourished, with one or two
score for those under the age of 65 years was 0.93 (SE diagnoses (versus three), following a special diet pre-
0.09) and this was significantly higher than for those aged admission, consuming <50% food in their first week of
≥ 65 years (0.65 SE 0.06), indicating that these younger hospitalisation, having a dietitian visit or being located in
adults experienced more hunger during hospitalisation. a hospital that had some but not all food outsourced.
The results are presented by domain.
Higher scores (i.e. more barriers) in the hunger domain
Discussion
were seen for younger patients and those having at least
two diagnoses. Higher scores in the eating difficulty The prevalence of malnutrition at the time of admission to
domain were seen in older, malnourished (SGA-B or -C), hospital is high (45%)(3). Nutritional status remains poor
and sicker (CCI > 2, CRP ≥ 10 g/L) patients, as well as or deteriorates throughout the hospital stay, whereas it
those who were in hospitals where a higher proportion of improves in some patients(1). Barriers that impair food
552
Not given food Not able to
Disturbed during Staff interrupted Missed meals when missed Did not get help choose foods that
meals during meals because of tests meals when needed like
Mild/moderate malnutrition (B) 284 41.2 284 44.0 283 40.6 135 67.4 85 35.3 130 28.5
Severe malnutrition (C) 90 50.0* 90 55.6* 89 43.8** 40 77.5 31 54.8 46 32.6**
Number of diagnoses 886 887 881 363 218 386
1 Diagnosis 574 38.2 575 43.1 573 33.2 232 69.0 121 48.8 267 23.2
2 Diagnoses 224 42.0 224 37.1 220 40.9 100 69.0 68 35.3 90 28.9
3 Diagnoses 88 36.4 88 45.5 88 29.5 31 71.0 29 31.0 29 6.9*
CCI at admission 872 873 867 359 216 373
CCI ≤ 2 513 37.6 514 37.9 510 35.5 205 69.8 114 42.1 203 21.7
CCI > 2 359 39.6 359 46.2* 357 34.5 154 68.2 102 41.2 170 27.1
C-reactive protein 710 711 705 299 183 281
<10 g/L 205 37.1 205 38.5 202 41.1 90 60.0 42 59.5 102 23.5
≥10 g/L 505 42.4 506 45.7 503 33.6 209 73.2* 141 41.1* 179 25.7
Admission ward type 832 833 829 350 211 386
Medical 559 38.1 560 42.5 559 33.1 224 65.2 141 41.8 289 21.5
Surgical 273 39.2 273 37.0 270 36.7 126 79.4** 70 42.9 97 28.9
Pre-admission weight loss 842 843 837 350 201 362
<5% 586 37.2 587 39.7 584 32.2 226 69.0 122 41.0 258 21.3
≥5% 256 43.0 256 46.1 253 42.7** 124 71.0 79 48.1 104 24.0
Following a special diet pre-admission 528 528 526 200 114 255
No 348 33.1 348 37.1 347 35.7 133 63.2 74 36.5 169 23.7
Yes 180 34.4 180 33.9 179 33.0 67 65.7 40 47.5 86 32.6***
Had registered dietitian visit 886 887 881 363 218 386
No 647 36.8 648 39.5 647 32.1 252 68.7 150 45.3 294 20.4
Yes 239 44.8 239 48.1 234 41.9** 111 70.3 68 35.3* 92 32.6***
Food intake during first week <50% 832 833 828 338 203 366
No 584 37.0 585 42.2 584 31.3 221 66.1 129 40.3 274 19.7
Yes 248 46.0* 248 42.7 244 41.8* 117 76.9* 74 44.6 92 35.9**
Percentage of outsourced food 738 739 736 307 178 352
1–49 185 23.2 185 23.2 185 30.8 62 67.7 34 47.1 84 20.2
H. Keller et al.
portions having these characteristics noted in the first column. For example, for the item ‘disturbed at meals’, younger, malnourished and patients reporting low food intake during the first week
Each question item focused on organisational aspects is presented in the columns, whereas patient, hospital and care characteristics that have been categorised are presented in rows, with pro-
of admission, as well as patients at hospitals reporting 85–100% of food as being outsourced, were more likely to experience this barrier. SGA, subjective global assessment; CCI, Charlson Com-
% affected
choose foods that
the present study is consistent with the limited research to
17.3**
43.1
date(7,14,16,24). Illness effects, eating difficulties and organi-
Not able to
196
72
like
28.1
47.5
when needed
69.0
71.0
are needed for all patients, and not only those who are
identified to be malnourished at admission.
meals
100
145
47.6*
85–100
50–84
changing mealtime practices is a promising strategy but interpretation of these results. For most of these items,
has been noted to be challenging to implement(27,29). having an organisational barrier translated into eating less
than 50% of the food provided during the first week of
admission (Table 3). Previous work has also noted this
Factors associated with barriers
association(19). Surprisingly, younger patients were more
In addition to documenting the prevalence of barriers to likely to be disturbed during the meal, either from noise/
food intake in the Canadian hospital system, the present smells or interruptions by medical staff, potentially as
study is novel in that it explores the potential associations more invasive procedures are required for these patients.
between barriers and some multilevel characteristics Older adults were more likely than younger adults to find
(patient, unit, hospital) that can help to target interven- it difficult to choose foods that they liked, suggesting that
tions and practice change. When comparing the domain current hospital menus may not meet food preferences for
scores by these characteristics (Table 2), it was not sur- this age cohort. Regardless of age, not being provided food
prising to see the significant association between the age when meals were missed occurred frequently (approxi-
of patients and the hunger domain. Younger patients mately 70%), suggesting that processes are needed to track
likely have increased energy needs and standard hospital this challenge for all patients and identify mechanisms for
trays may not provide sufficient energy, especially for ensuring that patients are not left without food for multi-
recovery. The identified positive association between hun- ple meals(4). Women were also more likely than men to
ger and more diagnoses could mean a higher metabolic find it difficult to make food choices and previous work
demand and thus the need for more energy and nutrients. has found similar differences in gender and hospital food
Based on these findings, younger patients and those with satisfaction(10).
a greater number of conditions may be targets for Malnourished patients are more likely to miss meals,
increased food provision in hospital. be interrupted, be disturbed at meals, and find it chal-
As expected, older, malnourished and sicker patients lenging to choose foods they wanted to eat than those
had more eating difficulties. However, the association who were not malnourished. It is apparent that greater
between eating difficulties and outsourced food is a little focus on the malnourished subgroup of patients is needed
surprising but may potentially be explained by increased to ensure that preferred foods are provided and meals are
packaging of outsourced food. To ensure this is not a protected from barriers. Similarly, the more unwell a
spurious association, future work examining usability of patient, the more likely they are to be interrupted by staff
outsourced packaging is worthy of further study. It is also at mealtimes and surgical patients are more likely than
not surprising that those with malnutrition on admission medical patients not to be provided a replacement tray
and those with low food intake in the first week had when a meal is missed. These barriers and their associa-
more dissatisfaction with food quality. There is likely a tions further affirm the concept that meals and food need
bidirectional relationship between malnutrition and taste to be seen and treated with the same importance as medi-
perception(30,31) and low intake is expected if patients do cine by the entire team(13).
not like the taste of the food. Following a therapeutic diet before admission was asso-
All of the factors significantly associated with the illness ciated with having difficulty making food choices in hos-
effects domain are consistent with our understanding of pital, and having difficulty choosing food in hospital was
how disease impacts the patient(9), except for the associated with a dietitian consultation. It is unclear from
increased scores in this domain for female compared to these analyses whether the consultation was undertaken
male patients. Based on these results, patients admitted to address this issue or resulted in this challenge for
with weight loss, using oral nutritional supplements in patients. Additionally, those with other organisational
the community and having more conditions on admis- barriers were more likely to have a dietitian consultation
sion could be targeted for enhanced nutrition interven- than those without these barriers and it is unclear from
tions to minimise the effects of illness on food intake. this analysis if the consultations were a result of the barri-
Screening at admission is a mechanism that can help to ers experienced or because the patient was unwell and
identify these patients. As seen in these data, these these barriers resulted from their condition. Finally, dis-
patients were also more likely to have a dietitian visit. turbances and staff interruptions appeared to occur more
However, this analysis does not tell us the effects of that frequently for patients in hospitals with 85% or more
intervention. outsourced food. It is unclear whether the tray delivery
Organisational items were not summarised into a systems or some other mechanism led to this significant
domain score. Rather than commenting by item, we association around interruptions. Greater outsourcing was
describe the characteristics of patients who were likely to negatively associated with lack of food choice and sug-
report these challenges in an attempt to have a meaningful gests that outsourcing may provide a larger food choice
for patients. Further investigations on menu length and patient population are more likely to experience these
variety are needed to confirm this assertion. barriers, including those who are malnourished on admis-
sion. Because improving food intake supports both the
recovery(1) and morale of patients(17): (i) barriers to food
Study limitations, strengths and recommendations for
intake should be routinely assessed within a hospital to
future research
better understand the patients’ perceptions and needs(19);
Although the present study is one of the few investigations (ii) multifactorial interventions are needed as a result of
that have attempted to further understand barriers to food the range of barriers to food intake(2,18,33); (iii) selected
intake in hospital, it is not without its limitations. Because groups of patients are more likely to report barriers (e.g.
of the research design, causality or directionality cannot be elderly, malnourished, those who were consuming oral
determined. Although generally representative of Canadian nutritional supplements before admission, etc.) and thus
medical and surgical patients(3), it is unknown whether the targeting these patients for interventions at admission,
sample included in this analysis is representative of patients potentially through screening, may be a way to support
in other countries. Self-report of variables including, food food intake; and (iv) policies to support food intake at
intake and use of supplements before admission, may have the hospital and unit level need to be fully implemented
lead to some misclassification bias. Misclassification of the to support food intake(34). Implementation research that
food intake variable may have also occurred for longer-stay focuses on these nutrition care processes is needed and
patients because only the first week of self-reported meal should be part of a future research agenda.
intake was used in the analysis. A social-ecological perspec-
tive considering the multiple levels of influence on hospital Acknowledgments
food intake was considered, although this was limited by
the covariates available for analysis. Future work should The support of the 18 hospitals and the site investigators
more broadly consider theoretical concepts(32) when plan- and coordinators who collected these data is gratefully
ning studies, ensuring that the measurement of a variety of acknowledged, as well as the patients who participated in
potential factors across several levels of influence (e.g. the present study. Special recognition is extended to
patient, unit, hospital, region) and consider multivariate Anastasia Teterina for her support with the statistical
modelling to better understand the factors associated with analysis. HK is in an endowed research position as the
barriers to food intake. Furthermore, patients who com- Schlegel Research Chair, Nutrition & Aging (2012–2017).
pleted some or all of the questionnaire were different from
those who did not complete the tool; this is not surprising
because the questionnaire is 37 items in length and poten- Conflict of interests, source of funding and
tially comprised too high a burden for the oldest and sick- authorship
est patients. Because malnutrition at admission was a
common covariate with mealtime experiences and chal- The authors declare that they have no conflicts of interest.
lenges in the present study, further work should be carried Funding was provided from the Canadian Nutrition
out aiming to identify food intake barriers in hospital for Society, which received unrestricted grants from Ab-
these most vulnerable patients who could not complete the bott Nutrition, Baxter, Fresenius-Kabi Canada, Nestle
survey. Social desirability bias may have influenced Health Sciences and Pfizer. JA, HK, KJ, ML, DD, LG,
responses for patients who were assisted by the project HP and PB are also members of Abbott Nutrition
coordinator for completion of the survey. Future work and/or Nestle Health Sciences speakers bureau. No
could also determine a minimal set of food access barriers other relationships or activities that could appear to
as indicators(5) that could be translated into mealtime audit have influenced the submitted work are declared.
tools to support as well as track changes in practices. The HK, ML, PB, EV and BD identified key covariates for
strengths of this research are the large and diverse sample, this exploratory analysis. HK and JA were responsible
the high completion rate of the questionnaire and consid- for the quality of the data collection, with DD, ML,
eration of multiple levels of influence and not just patient KJ, HP providing expertise on data collection and
characteristics on mealtime and food intake challenges. analysis where required for the main study. EV was
the data manager and conducted the initial analyses.
BD was the project manager during the data collection,
Summary attending to quality control with respect to data collec-
In summary, food access barriers are common in Cana- tion. All authors critically reviewed the manuscript and
dian hospitals and potentially translate into poor food approved the final version submitted for publication.
intake during hospitalisation. Specific subgroups of the
31. Ng K, Woo J, Kwan M et al. (2004) Effect of age and disease better food service? A study of compliance in Danish
on taste perception. J Pain Symptom Manage 28, 28–34. hospitals and nursing homes from 1995 to 2002-2003. Eur
32. Keller H, Carrier N, Duzer L et al. (2014) Making the J Clin Nutr 61, 129–134.
most of mealtimes (M3): grounding mealtime 34. Vanderwee K, Clays E, Bocquaert I et al. (2011)
interventions with a conceptual model. J Am Med Dir Malnutrition and nutritional care practices in hospital
Assoc 15, 158–161. wards for older people. J Adv Nurs 67, 736–746.
33. Mikkelsen B, Beck AM & Lassen A (2007) Do
recommendations for institutional food service result in