Modified Texture Food Use is Associated with Malnutrition in Long Term Care: An Analysis of Making the Most of Mealtimes (M3) Project

Abstract

Objective

Modified texture food (MTF), especially pureed is associated with a high prevalence of under-nutrition and weight loss among older adults in long term care (LTC); however, this may be confounded by other factors such as dependence in eating. This study examined if the prescription of MTF as compared to regular texture food is associated with malnutrition risk in residents of LTC homes when diverse relevant resident and home-level covariates are considered.

Design

Making the Most of Mealtimes (M3) is a cross-sectional multi-site study.

Setting

32 LTC homes in four Canadian provinces.

Participants

Regular (n= 337) and modified texture food consumers (minced n= 139; pureed n= 68).

Measurements

Malnutrition risk was determined using the Mini Nutritional Assessment short-form (MNA-SF) score. The use of MTFs, and resident and site characteristics were identified from health records, observations, and standardized assessments. Hierarchical linear regression analyses, accounting for clustering, were performed to determine if the prescription of MTFs is associated with malnutrition risk while controlling for important covariates, such as eating assistance.

Results

Prescription of minced food [F(1, 382)=5.01, p=0.03], as well as pureed food [F(1, 279)=4.95, p=0.03], were both significantly associated with malnutrition risk among residents. After adjusting for age and sex, other significant covariates were: use of oral nutritional supplements, eating challenges (e.g., spitting food out of mouth), poor oral health, and cognitive impairment.

Conclusions

Prescription of minced or pureed foods was significantly associated with the risk of malnutrition among residents living in LTC facilities while adjusting for other covariates. Further work needs to consider improving the nutrient density and sensory appeal of MTFs and target modifiable covariates.

Introduction

Malnutrition among older adults living in long term care (LTC) has many negative outcomes including increased risk of morbidity and mortality, impaired cognition, poor functional capacity and psychological well-being (1, 2, 3, 4, 5, 6), which is costly in human and health care terms (3, 5, 6, 7, 8, 9, 10). Poor food intake is considered a primary cause of malnutrition in LTC (11, 12) and is influenced by a variety of factors related to mealtime experience, meal access, and meal quality (7, 13). Dysphagia is estimated to occur in up to 60% of residents resulting from conditions common to this group (13, 14, 15) and is known to be associated with malnutrition (4, 12, 16, 17, 18, 19). However, it may not be dysphagia itself that is the cause of malnutrition, but rather the changes in food intake that result from management of dysphagia or the presence of concomitant factors such as dementia. Modified texture foods (MTF) are a common management strategy for dysphagia (6, 16, 17, 20) and prior work suggests deficits in the quality of MTF and especially the pureed foods used in LTC (21, 22, 23, 24, 25, 26). As well, persons who require MTF are more likely to have eating difficulties or require eating assistance (6, 17, 27, 28, 29). These challenges can also result in reduced food intake and malnutrition (6, 17, 27, 28), regardless of swallowing capacity or type of diet prescribed. It becomes difficult in these highly vulnerable residents to identify how best to support their nutritional status. Prior research has demonstrated that MTF use is associated with a high prevalence of malnutrition and weight loss among older adults in LTC (1, 6, 7, 16, 20, 30, 31, 32, 33). Yet, research considering MTF as a potential basis for malnutrition has been limited to small and homogenous samples (11, 12, 16, 18, 25, 34), which have not considered the confounding effects of a variety of covariates such as eating assistance, cognitive impairment and other factors (3, 12, 16, 17, 19, 35). Thus, is its unclear if MTF on their own are a root cause of malnutrition and should be the focus of intervention. The Making the Most of Mealtimes (M3) prevalence study identified that pureed diets were associated with lower energy intake, but when an interaction with eating challenges was considered it was no longer a significant predictor (13). Further, the provision of frequent eating assistance resulted in higher energy intake. This analysis demonstrated that the association between MTF and food intake is complex. Significant bivariate associations with MTF use have also been found in this sample, demonstrating the potential confounding effect of a variety of covariates (e.g., length of admission, dysphagia risk, dementia, use of oral nutritional supplements, micronutrient supplements, eating challenges and oral health) (29). What is not known is the independent effect of these covariates on the outcome of malnutrition risk; this knowledge would help to target limited resources for intervening and preventing malnutrition in LTC. The aim of this analysis, using the M3 data set, was to further understand the association of MTF use, and specifically minced and pureed textures in comparison to regular texture, on the risk for malnutrition when diverse relevant resident-level covariates, that are associated with MTF use, are considered.

Materials and Methods

Study design and participants

The M3 project, a cross-sectional multi-site study, was the basis for this analysis (7, 36). A total of 32 LTC homes were recruited across four provinces (8 homes per province) in Canada: Alberta (AB), Manitoba (MB), Ontario (ON), and New Brunswick (NB) (36). Diversity in the sample was achieved by including both for-profit and not-for-profit homes, both urban and rural homes, faith-based homes, and homes where a high proportion of individuals were from cultural minorities (36). Within each LTC home, one to four units with residents that met the eligibility criteria were randomly selected. Care was taken to ensure the selection of a dementia care unit where one existed. Twenty eligible residents (over the age of 65 years; residing in the selected unit and LTC home for a least one month) were randomly recruited per LTC home. They or their decision maker provided informed consent to participate. Residents were excluded if they: 1) were medically unstable (i.e., recent hospitalization); 2) were a short term admission; 3) required tube feeding; 4) were not eating because they were receiving end of life care; 5) did not routinely eat in the dining room; 6) had advanced orders excluding them from any research; or 7) they or their decision makers were unable to understand English, French and, in the case of Ontario, Cantonese. This protocol received approval from research ethics boards from the University of Waterloo (ORE 20056), University of Alberta (Pro00050002), University of Manitoba (J2014:139), Université de Moncton (1415-022), and University Health Network, University of Toronto (16-5051-DE). Some individual LTC homes also required ethics approval from their local/regional research ethics committee.

Data collection and measures

The MNA-SF was completed for each participant by a trained project coordinator; lower scores indicate increased risk of malnutrition (37, 38). The prescribed food texture for each participant and various resident characteristics were gathered from their health records. As a total of 67 different labels were used by the 32 LTC homes, coding rules were created to categorize MTF into logical groups (29, 39). Prior literature was used to identify covariates to be included in the analysis including: age, sex, length of admission (months), total number and type of major diagnoses using the interRAI diagnostic list (e.g., dementia), supplementation (oral nutritional, micronutrient; each coded Yes/No), total number of drugs, and if family routinely provided food for the resident (Yes/No). The interRAI LTCF cognitive performance scale (CPS; dichotomized for analysis ≤2 vs 3+, where 0 represents intact cognition and 6 represents very severe impairment) (36, 40) was used to categorize residents on cognitive ability. ‘Risk of dysphagia' was a composite measure, based on prescription of thickened fluids, failing a screening procedure with applesauce and water, or being observed by researchers to cough or choke at meals (36, 41). Trained dental hygienists completed a standardized oral assessment (36, 42), including a summary rating on their perception that the oral health status could influence food intake of the participant (1= not/unlikely influenced; 5= food intake significantly impacted by oral health; dichotomized for analysis 1 vs. 2+).

Additional covariates included were based on mealtime observations, carried out by trained research assistants who measured food intake as well as assessed mealtime interactions with residents. The standard and valid Edinburgh Feeding Evaluation in Dementia Questionnaire (Ed-FED) which assesses eating and feeding issues in older adults with dementia (36, 43) was collected at a breakfast, lunch and dinner. The single question on Ed-FED (i.e., Does the resident require physical help with feeding/eating?) was left categorized as three groups for the analysis (0= no physical assistance required; 1= sometimes physical assistance is required; and 2= often physical assistance is required to eat during mealtimes). For these three meals the research assistants also recorded nine other observed eating behaviours (i.e. chewing problems, lack of energy) using an ordinal scale (1= not applicable or never; 2= sometimes; 3= often) (43, 44, 45, 46). Scores for these nine additional questions were summed and termed Other Eating Challenges (36). The rounded average scores for the three meals were used in the analyses. For up to nine observed meals, research assistants collected meal duration for each participant. For one meal per day, they used the Relational Behaviour Scale (RBS) to gather data regarding the quality of eating assistance provided to each resident requiring full eating assistance (47). The RBS contains three questions each scored from one to seven (1= low intensity of relational care and 7= high intensity of care) (47). The totals for each meal was averaged for up to three meals and used in the analysis.

Statistical analysis

Data was analyzed using Statistical Analysis System software for Windows (Version 9.4). A hierarchical multiple linear regression analysis, accounting for clustering within province, home, and unit, was performed to assess the association of the MNA-SF total score with the prescription of minced or pureed as compared to regular texture food while adjusting for relevant covariates. Initial full models were based on covariates found in bivariate analysis to be associated with MNA-SF (p<0.2; age, sex, length of admission; dementia diagnosis, dysphagia risk; number of medications, oral nutritional supplement use; micronutrient supplement use; family brings in food; CPS; Ed-FED and Other Eating Challenges; oral health affects food intake; average length of meal; receiving eating assistance) and backwards elimination was used to develop the final models. Variables within the models that contributed the least based on their R square value were manually deleted from the models at each step until at least all variables in the model had a p-value of <0.05 and were based on the smallest decreases in the model sum of squares value as non-significant variables were removed. Interactions between Ed-FED and CPS, and also Ed-FED and prescribed MTF in the final model were tested. Further, multicollinearity was assessed by inspection of the correlation matrix to detect high correlations (R square values) among predictor variables.

Results

Characteristics and MNA-SF scores of residents prescribed modified and regular textures

Table 1 presents resident characteristics of the M3 sample (n= 638) and the average MNA-SF score for various categories of covariates. A total of 47.2% (n= 301) were prescribed any MTF, with 21.8% (n= 139) minced and 10.7% (n= 68) pureed food. The MNA-SF score was significantly lower with MTF use (e.g. regular texture food=11.37 SD 2.13; pureed= 8.41 SD 2.25) (F(4, 633)= 32.81, p<0.0001). MNA-SF scores were also significantly lower for participants who: had a diagnosis of dementia (t(515)= 5.33, p<0.0001); a higher CPS score (more cognitive impairment) (t(631)= 11.75, p<0.0001); used an oral nutritional supplement (t(294)= 10.38, p<0.0001); required physical assistance during mealtimes (F(2, 630)= 67.55, p< 0.0001); or were rated as having oral health affecting food intake (t(539)= 3.77, p=0.0002). MNA-SF scores were significantly higher for participants who used micronutrient supplementation (t(212)= -3.40, p= 0.001) or had family bring in food for them (t(634)= -3.85, p= 0.0001).

Table 1.

Resident characteristics of the M3 sample and MNA-SF scores (n=638)

	Descriptive Statistics	MNA-SF Score [mean (SD)]
Age (years), mean (SD)	86.81 (7.80)	10.64 (2.53)
Sex, % (n)
Male	31.03 (198)	10.89 (2.42)
Female	68.97 (440)	10.52 (2.57)
Number of residents on prescribed food textures, % (n)
Regular fooda	52.82 (337)	11.37 (2.13)*
Modified texture foodb	47.18 (301)	9.81 (2.68)
Bite-sized	14.26 (91)	11.12 (2.18)
Minced	21.79 (139)	9.69 (2.79)
Pureed	10.66 (68)	8.41 (2.25)
Liquidized	0.47 (3)	7.33 (4.04)
Length of admission (months), mean (SD) [median]	27.79 (27.16) [20.00]	10.64 (2.53)
Number of diagnoses, mean (SD)	5.41 (2.02)	10.64 (2.53)
Dementia diagnosis, % (n)
Yes	65.20 (416)	10.27 (2.60)*
No	34.80 (222)	11.32 (2.23)
Dysphagia risk, % (n)
Yes	59.09 (377)	10.54 (2.50)
No	40.91 (261)	10.77 (2.56)
Number of drugs, mean (SD)	7.50 (3.45)	10.64 (2.53)
Oral nutritional supplementation, % (n)
Yes	30.72 (196)	9.05 (2.77)*
No	69.28 (442)	11.34 (2.06)
Micronutrient supplementation, % (n)
Yes	76.65 (489)	10.84 (2.37)*
No	23.35 (149)	9.96 (2.86)
Family routinely brings food in for resident, % (n)1
Yes	35.38 (225)	11.15 (2.37)*
No	64.62 (411)	10.35 (2.57)
interRAI LTCF scales, CPS, % (n)2
≤2	44.23 (280)	11.81 (1.95)*
≥3	55.77 (353)	9.72 (2.53)
Physical assistance required during mealtimes, % (n)2
Often	11.85 (75)	8.24 (2.31)*
Sometimes	11.37 (72)	9.33 (2.92)
Never	76.78 (486)	11.21 (2.19)
EDFED, mean (SD)2	12.37 (2.25)	10.65 (2.53)
Other Eating Challenges, mean (SD)2	10.64 (1.65)	10.65 (2.53)
Relational Behaviour Scale, mean (SD)3	16.49 (3.32)	8.58 (2.29)
Oral health rating, % (n)4
Affects consumption	49.47 (280)	10.41 (2.67)*
Does not affect consumption	50.53 (286)	11.19 (2.19)
Average length of meal (minutes), mean (SD)1	40.18 (13.05)	10.64 (2.53)

Descriptive statistics were used to compute means for continuous variables and frequencies for categorical variables. Numerical and ordinal data are mean (standard deviation) and median where appropriate. Categorical and ordinal data are % (n). Student’s t-test and an ANOVA were performed, values with * indicate p<0.05 across categories. Abbreviations: SD= standard deviation. Total n=638, where

^1.n= 636

^2.n= 633

^3.n= 99

^4.n= 566

^a.Total number of residents prescribed regular texture food is 338 from the total M3 sample of 639 residents, however one resident prescribed regular texture food had a missing MNA-SF total score thus the total number of residents prescribed a regular texture used for this analysis is 337.

^b.Modified texture food= bite-sized; minced; pureed; or liquidized.

There was a significant and negative association with MNASF for minced (β= -0.56, standard error (SE)= 0.25, t(382)= -2.24, p= 0.03); and pureed textures (β= -0.89, standard error (SE)= 0.40, t(279)= -2.23, p= 0.03; Table 2) as compared to the regular texture when adjusting for covariates; being on either of these textures increased risk for malnutrition. Age was negatively associated with MNA-SF score for the minced consumers only. Covariates were consistent between minced and pureed models, excepting that oral health was not included in the final model for the minced vs. regular texture comparison. Covariates that were associated with MNA-SF included: CPS scores (more impaired had more nutrition risk), Ed-FED (more eating challenges had more nutritional risk), prescription of oral supplements (more nutrition risk), and oral health affecting food intake rating (more nutrition risk). Interaction effects tested between CPS and Ed-FED scores were not significant, as well as the interaction effect between prescription of MTF and Ed-FED score.

Table 2.

Hierarchical multivariate linear regression analysis of malnutrition risk and use of modified texture food

	Minced vs. regular texture N=469. F-value= 4.20 (p<0.0001) R2=0.37		Pureed vs. regular texture N= 364. F-value= 3.84 (p <0.0001) R2= 0.40
Covariates	ß	SE	ß	SE
Minced fooda	-0.56	0.25*	—	—
Pureed fooda	—	—	-0.89	0.40*
Age	-0.04	0.01*	-0.02	0.01
Maleb	-0.38	0.24	-0.19	0.26
Oral nutritional supplementationb	-1.52	0.25*	-1.51	0.29*
Cognitive performance scale scored	-1.01	0.25*	-0.85	0.29*
Ed-FED score	-0.33	0.07*	-0.34	0.08*
Oral health affects consumptione	—	—	-0.56	0.26*

Abbreviations: ß= standardized beta parameter estimate; SE= standard error; * p < 0.05; a. Reference category= regular texture food. b. Reference category= female. c. Reference category= no. d. Reference category= ≤2. e. Reference category= oral health does not affect consumption.

Discussion

Our findings indicate that the prescription of minced or pureed food textures are negatively associated with MNA-SF scores in residents in LTC homes when relevant resident level covariates, often associated with MTF use were considered. This analysis further confirms that dysphagia risk alone is not sufficient to predict malnutrition, but rather it is the diet that is used to manage dysphagia that is relevant. This suggests that the sensory and nutrient quality of the MTF is of importance for preventing malnutrition. Existing literature has suggested that pureed food contains lower amounts of energy, protein and micronutrients compared to regular texture and other modified textures (e.g., chopped, minced) (21, 22, 23, 24, 25, 26), and minced foods along with pureed foods tend to be visually unappetizing and have a decreased palatability (6, 11, 17, 20, 28, 48). Food intake analysis in the M3 sample, demonstrates that MTF consumers (minced, pureed, and liquidized) have lower intakes of energy, protein and several micronutrients (49). Therefore, interventions that improve the nutrient density and sensory appeal of MTF need to be implemented in LTC. Prior research suggests success with making food more appealing (50, 51, 52, 53) and enhancing ingredients to improve intake of energy and protein (31, 54, 55). Standards and policy to make these initiatives the norm for long term care residents who require MTF are required to drive practice change.

Eating difficulties, the inability to eat independently, and a diagnosis of dementia can lead to malnutrition and are commonly associated with dysphagia and prescription of MTF (4, 5, 6, 7, 16, 17, 20). This analysis, for the first time, disentangled these relationships demonstrating the importance of these individual variables, even when adjusting for others. Prior work has confirmed the associations identified between age (17, 19, 35, 56), oral nutritional supplementation, significant cognitive impairment, eating challenges, poor oral health and malnutrition (17, 19, 35, 57, 58). Oral supplementation prescription is likely to have been an intervention to address existing malnutrition, rather than causing malnutrition (3, 18, 57, 59, 60, 61, 62). The continued association between cognitive level and malnutrition when adjusting for eating difficulties, MTF, and oral health suggests that dementia itself or un-modeled factors associated with dementia (e.g., agitation, wandering) are also relevant for nutritional status (3, 12, 60, 62). This study found that residents had a lower MNA-SF score when their oral health was rated to have an effect on food intake. This is in contrast to a study by Verbrugghe and colleagues which found no significant association between malnutrition and deglutition (ability to chew, manipulation of food in mouth, and swallowing) when using the Minimal Eating Observation Form- Version II (17). Differences in assessment of oral health likely explain this difference as M3 used experts and a standardized oral health exam. Based on this analysis, oral health care is an important area for development and implementation of feasible interventions. Interventions to improve oral health have been shown to improve food intake, toothaches, and jaw pain (57, 63). Eating assistance itself and the quality of this assistance (Relational Behaviour Scale) were not independently associated with malnutrition in this sample. LTC standards in some provinces require a maximum of two residents being assisted at one time by one staff (64) as well as training (64, 65, 66, 67) staff to ensure safe eating practices; these standards may explain this lack of association. Further, a prior analysis demonstrated that residents who required frequent eating assistance had better food intake than those who sometimes required eating assistance (13); thus, receiving eating assistance leads to improved food intake (68, 69, 70, 71). As dementia progresses, residents have more and more eating challenges (Ed-FED) which are important determinants of food intake (13) and were associated with malnutrition risk in this analysis. Interventions to support self-feeding and other eating challenges are needed to improve food intake among LTC residents (71, 72, 73).

Study limitations

As this was a cross-sectional study, causality cannot be inferred between risk factors and malnutrition. The MNA-SF was used to determine malnutrition risk and the full MNA may be more appropriate to administer to older adults living in LTC (37, 38). Diet textures required classification to go from 67 diet types and misclassification may have occurred; further there was lack of standardization of these textures across homes. Similarly, a conservative view on dysphagia risk was taken and may have overestimated prevalence. Finally, not all covariates that could potentially influence food intake and thus be associated with malnutrition were modeled. As the primary purpose was to determine if MTF is associated with malnutrition, a parsimonious strategy was adopted based on theory and factors known to confound MTF, dysphagia and malnutrition. Other covariates such as mealtime experience and mealtime environment (e.g., glare on floor, no table setting contrast, background noise) (63, 64, 65) could affect food intake and nutritional status and should be examined in future studies.

Conclusion

MTF use is common in LTC and as demonstrated in this analysis, is associated with malnutrition risk when adjusting for other covariates. As upwards of a third of residents required MTF (e.g., minced, pureed), greater attention to the sensory and nutrient quality of these foods is needed to prevent malnutrition that is so common in this group of residents. Single component interventions targeting the oral health of residents (68, 69, 71, 74); restorative dining (57, 63); and the nutrient density, visual appeal, and textural and sensory qualities of MTF (24, 48, 51, 75, 76, 77, 78) have been found to be effective, but need to be researched more broadly using a multi-component intervention format to target the complexity of malnutrition risk in LTC. The ultimate goal of the M3 prevalence study was to identify opportunities to improve food and fluid intake among residents, and thus improve their nutritional status by informing future intervention studies.

Statement of Authorship

The primary author, Vanessa Vucea, conducted data analyses and developed the first manuscript draft in collaboration with Heather H. Keller. Co-author, Jill M. Morrison, provided statistical support for data analyses. Co-authors (Heather H. Keller, Jill M. Morrison, Lisa M. Duizer, Alison M. Duncan, Natalie Carrier, Christina O. Lengyel, Susan E. Slaughter, and Catriona M. Steele) reviewed and provided input for this manuscript based on their specific research expertise. All authors reviewed and approved this manuscript.

Conflict of Interest

Vanessa Vucea, Alison M. Duncan, Natalie Carrier, and Christina O. Lengyel declare that they have no conflicting interests associated with this research. Heather H. Keller reports grants from Canadian Institutes of Health Research, outside the submitted work; Speakers bureau: Abbott Nutrition, Nestle Health Sciences; and Co-Chair Canadian Malnutrition Task Force. Jill M. Morrison reports grants from Canadian Institutes of Health Research, outside the submitted work. Lisa M. Duizer reports grants from Toronto Rehabilitation Institute, outside the submitted work. Susan E. Slaughter reports grants from Canadian Institutes of Health Research, outside the submitted work. Catriona M. Steele reports grants from Canadian Institutes of Health Research, grants from National Institute of Deafness and Other Communication Disorders, outside the submitted work; and Board member of the International Dysphagia Diet Standardisation Initiative.

Funding Sources

The M3 project was funded by the Canadian Institutes of Health Research (CIHR).

Ethical Standards

This study received approval from research ethics boards from the University of Waterloo (ORE 20056), University of Alberta (Pro00050002), University of Manitoba (J2014:139), Université de Moncton (1415-022), and University Health Network, University of Toronto (16-5051-DE). Some individual LTC homes also required ethics approval from their local/regional research ethics committee.