Skip to main content

A path analysis on the direct and indirect effects of the unit environment on eating dependence among cognitively impaired nursing home residents



This study aims to estimate the direct and indirect effects of the unit environment alongside individual and nursing care variables on eating dependence among residents who are cognitively impaired and living in a nursing home.


A multicentre observational study was carried out in 2017: 13 Italian nursing homes were involved in data collection. Included residents were aged > 65 at baseline, living in the considered facility for the last 6 months and during the entire study period and having received at least one comprehensive assessment. Data were collected (a) at the individual level: eating dependence using the Edinburgh Feeding Evaluation in Dementia Scale and other clinical variables; (b) at the nursing care level with daily interventions to maintain eating independence assessed with a checklist; and (c) at the nursing home level, using the Therapeutic Environment Screening Survey for Nursing Homes.


One thousand twenty-seven residents were included with an average age of 85.32 years old (95% CI: 84.74–85.89), mainly female (781; 76%). The path analysis explained the 57.7% variance in eating dependence. Factors preventing eating dependence were: (a) at the individual level, increased functional dependence measured with the Barthel Index (β − 2.374); eating in the dining room surrounded by residents (β − 1.802) as compared to eating alone in bed; and having a close relationship with family relatives (β − 0.854), (b) at the nursing care level, the increased number of interventions aimed at promoting independence (β − 0.524); and (c) at the NH level, high scores in ‘Space setting’ (β − 4.446), ‘Safety’ (β − 3.053), ‘Lighting’ (β − 2.848) and ‘Outdoor access’ (β − 1.225). However, environmental factors at the unit level were found to have also indirect effects by influencing the degree of functional dependence, the occurrence of night restlessness and the number of daily interventions performed by the nursing staff.


Eating dependence is a complex phenomenon requiring interventions targeting individual, nursing care, and environmental levels. The NH environment had the largest direct and indirect effect on residents’ eating dependence, thus suggesting that at this level appropriate interventions should be designed and implemented.

Peer Review reports


The progressive decline of functional dependence has been reported as the major cause of nursing home (NH) admission among cognitively impaired older individuals [1]. In the moderate stage of dementia, by following a hierarchical order, impairments among early-, middle- and late- loss activity of daily living (ADL) have been documented, with eating independence being among the last ADLs to deteriorate [2]. Initially, residents have been documented to refuse to eat [3], and then to manifest behavioural disorders, dysfunctions of eating mechanism, the inability to recognise food and how to use cutlery; in the late stages, complex compensatory and supportive measures during mealtime are required [4].

Aiming at preventing negative consequences such as malnutrition, anorexia, increased occurrence of pressure ulcers, dehydration, aspiration, multiple hospitalisation and mortality [5, 6], several interventions both at the resident and at the environmental levels [7] have been documented to date. Among the first, simple (e.g., offering verbal prompts, modified food and drinks and finger food [8]) to complex interventions (e.g., space retrieval, Montessori method [9]) have been studied. Moreover, the effectiveness of training programs aimed at educating caregivers and healthcare professionals on safe methods capable of guaranteeing optimal eating assistance [10, 11] has also been researched. At the environmental level, homelike dining rooms with limited noises or distractions during mealtime; appropriate meal service delivery styles and soft music during mealtime, have been documented to increase intake [10, 11]. However, these studies have been confined in the environment where meals are eaten [12]. Instead, residents live in complex NH environments for several years, not spending all day in the dining room where these changes have been recommended to be implemented.

According to Lin et al. [9] residents spend from 11.25 to 19.03 min to complete their meal; therefore, they are immersed in the dining environment for around 1 h and half a day, suggesting that they are more exposed to the influence of the entire physical environment of the NH unit that requires changes in their design choices [13].

However, to date only a few studies have been performed to identify the role of the NH environment alongside other individual and nursing care factors in initiating or in delaying the onset of disability in eating [14, 15]. Moreover, according to our best knowledge, no studies have investigated the possible direct and indirect effects of the NH environment on the degree of dependence in eating [14], for example influencing other variables directly affecting eating performance. Furthermore, given that eating is a social process [16], whether residents are used to eating at a table with other residents or alone in their bedroom has not been considered in available studies.

Therefore, the main aim of this study was to estimate the direct and indirect effects of the NH unit environment alongside individual and nursing care variables on residents’ eating dependence. We hypothesised that beyond the role of individual and nursing care factors [12] on the degree of dependence in eating, other environmental factors at the NH unit level, directly and indirectly increase and/or prevent eating impairments.


Study design

A multicentre pragmatic observational study design was performed in 2017, and here reported according to the STrengthening the Reporting of OBservational studies in Epidemiology studies [17].

Sample and setting

A total of 13 public NHs located in a rural area in the North-East of Italy where around 100,000 citizens were living at the time of the study and under the same Regional Health Service rules were preliminary assessed for their homogeneity in (a) their mission as long term facilities, (b) the amount of nursing care offered daily by nursing aides (NAs) and Registered Nurses (RNs) in around 75 min/day/resident, and (c) the admission resident criteria, as residents with moderate/severe functional dependence due to different health conditions - mainly dementia [18]. All NH were approached and all agreed to participate; these were equipped with an average of 86 beds (from 33 to 200, a total of 1161) and at the period of the study were hosting on average 83 residents (from 30 to 164).

Residents who were (a) > 65 years; (b) living in the same NH unit for the last 6 months and during the entire study period; and (c) who had reported in his/her records a comprehensive need assessment were deemed eligible. In Fig. 1, the flow diagram of residents included in the study has been reported.

Fig. 1

Eligible residents, included and reasons for exclusion. NH Nursing Home, n number

Variables and data collection process

The outcome variable of the study was the eating performance as measured using the Edinburgh Feeding Evaluation in Dementia scale (EdFED) [19] in its Italian validated version [20]. The tool consists of 10 items based upon a 3-point Likert scale (0 never, 1 sometimes, 2 often). It was completed by observing each resident during their entire mealtime.

The tool has demonstrated strong psychometric properties in previous studies [e.g., 19, 20]. As the total score increases (from 0 to 20), the dependence in eating is higher. The item number 11, measuring the appropriate level of care required by the resident (only supportive/educative; partly or wholly compensatory), was also filled in by giving a score of 0, 1 or 2, respectively, according to the care delivered during the observed mealtimes.

The remaining variables were collected at the following levels, as reported in Table 1:

  1. 1)

    Individual level: in addition to some demographic variables (age, gender), the degree of functional dependence (Barthel Index, [21]), the cognitive performance (Cognitive Performance Scale, [22], the emotional status (Depression Rating Scale, [23]), the presence of pain and its intensity (Pain Intensity, [24]), the occurrence of some behavioural symptoms (night restlessness, verbal aggressiveness, physical aggressiveness occurrence) and the clinical instability (Clinical Instability Score, [25]) were considered. The presence of close/intimate relationships with family relatives [25] and where the resident was used to have breakfast, lunch and dinner (in his/her bedroom or in the dining room), and with whom (alone or with others) was also considered.

  2. 2)

    Nursing care level: routine interventions performed to maintain eating independence at the (a) resident level (e.g., verbal, behavioural or motivational prompts) and at the (b) environmental level (e.g., by promoting the desire to eat by stimulating smell and visual memory, creating and maintaining a peaceful environment allowing residents’ concentration) were recorded. These interventions emerged in a parallel study performed in the same NHs through focus groups [18] and then transformed into a checklist in the current study as a basis for observation.

  3. 3)

    Nursing Home level: in addition to the size (number of beds), the organisation of the NH in the number of units as a confined environment (with no commons areas to serve more than one unit) where a group of residents were living at the time of the study, cared for by a nursing team led by a nurse leader (26), was assessed. Therefore, the therapeutic properties of each NH unit environment were evaluated using the Therapeutic Environment Screening Survey for Nursing Home (TESS-NH) [26] composed of 84 items categorised into 13 domains as reported in Table 1. Authorisation to use the tool was obtained from the author (Prof. Sloane, correspondence available from authors). After having ensured the cross-cultural and conceptual equivalence, together with the face and content validation, the tool was subjected to the validation of other properties: as according to the Authors the tool is a checklist, only the inter-rater reliability, the test−retest, the criterion validity, the inter-dimension correlations and the internal consistency were evaluated (available from authors).

Table 1 Variables measured, metrics and validity of the measure, and source(s) of data collection

Path analysis findings: direct

Data collection process

Facilities were approached by the research team in the second semester of 2017. Different methods of data collection were used according to the level and nature of the data. In larger NHs data were collected over a few days:

  • Nursing Home level: two trained researchers with a nursing background visited each NH unit, taking around 2 h to complete the data collection with the TESS-NH [26]. Data were collected independently and then agreed upon, and discordances were discussed with a third researcher. Residents who were living in each NH unit on the day of the survey (n = 1080) were registered as eligible to be included in the study.

  • Outcome variable and individual level variables: after 3–4 weeks, on a day selected randomly by the principal investigator, the outcome was measured by observing each resident during lunch time in the dining room or at his/her bedside according to his/her routines. Four researchers with a nursing background and trained via a 4-h course in the use of the Edinburgh Feeding Evaluation in Dementia scale [19, 20] were involved. Only residents satisfying the inclusion criteria (n = 1027) were evaluated. For these residents the last complete comprehensive assessment performed and recorded in the NH and regional databases with the Val. Graf tool [25] including different measures was extracted after having received the appropriate authorization from the Ethical Committee.

  • Nursing care level variables: the daily interventions at the environment and at the individual level performed by the staff in the dining room or in the bedroom to maintain eating performance [18] were observed on the day of data collection at the individual level trough observations based on a checklist [18].

Bias control

At the resident level, selection bias [29] was prevented by including all residents living in the approached NHs. To avoid any misclassification [29] validated tools were used [19,20,21,22,23, 26]; moreover, their reliability was reassessed with the data collected in this study as reported in Table 1. Furthermore, while the comprehensive need assessment was performed by trained RNs, responsible for the care delivered to the residents, the outcome variable and the nursing care variables were assessed by researchers not involved in the daily care of residents after having received appropriate training and under the supervision of an expert researcher.

At the NH level, performance bias was prevented by including residents who were receiving the same amount of nursing care as established by regional rules. During the study period the NH policies were stable over time. Furthermore, to prevent any bias in observation, the NH unit environment evaluation with the TESS-NH tool [26] and by the same trained researchers not involved in other data collection and in daily care of residents.

Modelling and data analysis

On a preliminary fashion, Descriptive and inferential statistics were performed by computing frequencies, percentages and averages (with Standard Deviations [SD], ranges; or Confidence Intervals [CI] at 95%).

Then, in line with the study hypothesis, the Intra Class Correlation (ICC) was evaluated under random and fixed effects (CI at 95%, bootstrap method) to identify effect clusters at the NH units’ levels on the outcome variable. The ICC of the Edinburgh Feeding Evaluation in Dementia scale scores (19) at the NH level were 0.10 (95% CI: 0.03–0.19) and 0.06 (95% CI: 0.02–0.15) under random and fixed effects, respectively; at the NH units, these were 0.13 (95% CI: 0.06–0.20) and 0.10 (95% CI: 0.07–0.20) under random and fixed effects, respectively.

Then, taking into account the high cluster effect of the NH unit on the outcome variable, the path analysis model was developed: on a preliminary fashion, all variables collected were introduced in the saturated model and tested. The complexity of the correlation structure forced a more sparing variable selection by including only those consistent with the available conceptual frameworks describing the relation between some (a) individual-, (b) NH-, and (c) nursing care- levels variables. Specifically, the Chang and Roberts model [30] was considered documenting that feeding difficulties are based on individual factors (memory and cognitive impairments), but also on several contingent factors that have a probabilistic relation with these difficulties attributable to time or space patterns. Among these, social and psychological factors, as well as the dining environment, and culturally appropriate food choices, have been identified. Thus, collected variables consistent with the above-mentioned model were explored in their correlations with the outcome variable (Additional file 1) and those significantly correlated with each other were kept. Some of those not significantly correlated were also included in the model according to the evidence available in the field (e.g., depression [31]; pain intensity [32]; verbal aggressiveness [33]; clinical instability [18]; and the environmental and resident interventions [7]). Collinearities were also then assessed and removed as in the case of the TESS-NH global item score and the TESS-NH dimensions. Finally, a direction of each relationship among variables was assigned according to the Chang and Roberts model [30] and the study hypotheses.

Thus, the path analysis was performed by introducing the outcome variable (=eating dependence) as measured with the Edinburgh Feeding Evaluation in Dementia scale [19, 20] and as explanatory variables those identified in the model: according to the study hypothesis, some variables at the NH level (e.g. bed size, and units) and at the individual level (e.g. age, gender, Cognitive Performance Score, clinical instability, close relationship with the family relatives’) [34] were considered as exogenous variables [35], not influenced by variables introduced in the model. The remaining were considered as endogenous variables [35].

Direct and indirect effects were then explored by sequential multiple regression analyses: the standardised coefficients β were estimated for each variable. Standard Errors (Std.Err), Test Statistics (z-values) and p-values (P(>|z|) were also reported to perform the inferential analysis. Moreover, according to Tarling [35] we considered (a) direct effects, (b) indirect effects (by simply multiplying the path coefficients connecting the causal variable to the outcomes) and (c) total effects (as the sum of direct and indirect effects). Therefore, the model fit was assessed analysing the coefficients of determination (R2) specific of each regression.

All analyses were performed by using the SPSS Statistical Package version 24 and R Core Team (R Core Team, 2017).

On a preliminary fashion, the database was checked for missing values (< 1%) and these were managed adopting the Full Information Maximum Likelihood approach [36]. The R Package Lavaan [37] was used for model estimation. The statistical significance was set at p < 0.05.


Outcome variable

The 1027 residents included reported at the Edinburgh Feeding Evaluation in Dementia scale score on average 2.48 points (95% CI: 2.22–2.73); one fourth of them were in need of whole compensatory support (242; 23.6%) by carers as reported in Table 2.

Table 2 Outcome, individual, nursing care and NH level variables

Individual, nursing care and NH level variables

As reported in Table 2, at the individual level, the residents’ age was on average 85.32 years (95% CI: 84.74–85.89) and the majority were female (781; 76%). The average score on the Barthel Index was 25.25 (95% CI: 23.47–27.03); nearly half of the residents were moderate/severely cognitively impaired (471; 45.9%) as reporting a Cognitive Performance Score ≥ 3 and, similarly, at the Depression Rating Scale, around half were diagnosed with minor or major depressive disorders (459; 44.7%). Considering the Pain Intensity Index, an average score of 0.72 out of 3 (95% CI: 0.66–0.77) was recorded, while in some specific behaviour problems such as night restlessness, verbal and physical aggressiveness, the average scores out of 4 were 0.48 (95% CI: 0.42–0.55), 0.33 (0.28–0.38) and 0.14 (0.11–0.17), respectively. The residents’ clinical instability was on average 1.34 out of 4 (95% CI: 1.29–1.40). The majority of them had a close relationship with their relatives (710; 69.1%) and they were used to have their meals in the dining room surrounded by other residents (558; 54.3%).

At the nursing care level, on a daily basis, an average of 3.76 out of six environment interventions (95% CI: 2.98–4.55) with a large range (from two interventions in seven NHs, to all six in four NHs) were performed. On a daily basis, at the resident level on average 8.46 out of 10 interventions (95% CI: 7.39–9.52) were performed with one NH performing only one and eight performing all interventions included in the checklist.

At the NH level, facilities were composed from one to 4 units with on average 34 residents. At the TESS-NH global score, the average rating was 7 out of 10 (SD, 1.55) ranging from 5 (six NH units) to 10 (3 units). Average scores in each TESS-NH dimension have been reported in Table 2.

Path analysis

The model has explained the 57.7% variance in eating dependence as fully reported in the Additional file 2. As reported in Fig. 2, several individual level variables resulted as directly preventing self-feeding dependence. Higher scores on the Barthel Index had the greatest effect on preventing self-feeding dependence (β − 2.374). Compared to eating in bed alone, the following eating arrangements also resulted in preventing self-feeding dependence, although to a lesser extent: eating in the dining room with two residents (one on the left and one the right, β − 1.352), eating in the dining room with two residents (one of them in front, β − 1.577) or surrounded by other residents (β − 1.802) as compared to eating in bed, alone, have all emerged as preventing eating dependence. Moreover, having a close relationship with family relatives has also emerged as preventing eating dependence (β − 0.854).

Fig. 2

NH, individual and nursing care variables: direct effects on eating dependence as measured with the EdFED. In the boxes, β values are reported; the full specification of the model is reported in the Additional file 2. EdFED Edinburgh Feeding Evaluation in Dementia scale, NH Nursing Home

At the nursing care level, the increased number of interventions performed on a daily basis at the resident level has emerged as a protective factor for eating dependence (β − 0.524). At the NH levels, with higher effects, ‘Space setting’ (β − 4.446), ‘Safety (β -3.055) ‘Lighting’ (β -2.848) and ‘Outdoor access’ (β − 1.225) dimensions have all resulted as preventing eating dependence.

On the other hand, factors directly increasing the likelihood of eating dependence at the individual levels were female gender (β 0.700), the increased Cognitive Performance Scores (β 0.600) and night restlessness (β 0.259) while the remaining variables entered in the model reported no significant contribution to the outcome variable. With the highest direct impact, some environmental dimensions as measured with the TESS-NH tool have increased the likelihood of eating dependence (‘Cleanliness’, β 6.574; ‘Maintenance’, β 2.919; ‘Familiarity’, β 2.326; and ‘Visual Tactile’, β 1.459). Also the unit’s bed size has increased the likelihood of eating dependence but with a limited direct impact (β 0.041).

As evident in Fig. 3, indirect effects have emerged mainly at the NH levels as measured with the TESS-NH. Functional dependence as measured with the Barthel Index [19] was prevented by the ‘Orientation/cueing’ (β − 1.003), ‘Maintenance’ (β − 0.139) ‘Familiarity’ (β − 0.137) dimensions, the number of units in the NH and the number of beds at the unit level (β − 0.071 and − 0.007, respectively).

Fig. 3

NH, individual and nursing care variables: indirect effects on eating dependence as measured with the EdFED. In the boxes, β values are reported; the full specification of the model is reported in the Additional file 2. EdFED Edinburgh Feeding Evaluation in Dementia scale, NH Nursing Home

In contrast, eating dependence was increased by ‘Space setting’ (β 0.446), ‘Exit control’ (β 0.251), ‘Lighting’ (β 0.246) and ‘Noise’ (β 0.006) dimensions.

Night restlessness was prevented by the ‘Space setting’ (β − 0.856), and increased by the ‘Orientation/cueing’ dimension (β 1.630) and the number of beds in the unit (β 0.008). Larger indirect effects of the NH environment have emerged on the nursing staff, where the amount of interventions performed on a daily basis at the resident level have been increased by ‘Space setting’ (β 3.837), ‘Visual tactile’ (β 1.478), ‘Exit control’ (β 1.037) and ‘Lighting’ (β 0.954) dimensions and prevented by ‘Orientation/cueing’ (− 5.725), ‘Cleanliness’ (− 2.833) ‘Familiarity’ (β − 1.939), ‘Noise’(β − 1.422) and ‘Maintenance’ (β − 0.713) dimensions. The number of interventions performed at the resident level were prevented by the increased Barthel Index (β − 0.668) and increased by the clinical instability (β 0.075).

According to Tarling [34] direct, indirect and total effects have been calculated (Table 3). NH level variables with the largest effects increasing eating-dependence were, in order, ‘Cleanliness’, ‘Orientation/cueing’, ‘Familiarity’ and ‘Maintenance’. NH level variables preventing eating dependence were appropriate ‘Space Setting’ and ‘Lighting’.

Table 3 Path analysis findings: direct, indirect and total effects [32]


Individual, nursing care and NH’s variables

More than 35% of residents required from partial to complete help in eating confirming that eating dependence is an increased care need in NHs [4, 38, 39].

The residents’ profile was similar to that reported in previous studies as highly functionally and cognitively impaired, and around four out of ten with a depressed mood [40]. Residents were moderately clinically stable as previously documented [41] and have shown less than one episode/week of behavioural problems such as night restlessness, verbal and physical aggressiveness, suggesting that there was a low incidence of episodes of resistance to cooperate in care [42]. The majority of them were in a close relationship with relatives, in line with previous studies [34].

At the nursing care level, interventions routinely delivered to promote eating performances in the dining environment were on average four, with a great variability across NHs (from two to six/NH). At the residents’ level, on average of more than eight interventions (from one to ten/NH) were performed. Staff attitudes, evidence-based knowledge, beliefs [43], tacit knowledge [44] regarding what is effective and not with each resident, as well as the culture and policies of the NH regarding mealtime [6] can explain these variations.

At the NH level, a moderate pleasant, positive and functional evaluation was evidenced, higher when compared to previous studies where an average of 5.75 has been reported [26]. Specifically, in some dimensions (‘Unit Autonomy’, ‘Exit control’, ‘Maintenance’, ‘Cleanliness’, ‘Safety’, ‘Lighting’, ‘Visual/tactile’, ‘Space setting’ and ‘Orientation/cueing’) the average scores were higher as compared to those documented previously [26, 45]. In the remaining dimensions (‘Outdoor access’, ‘Privacy’, ‘Noise’ and ‘Familiarity’) the average scores were lower as compared to previous studies [26, 45] thus suggesting areas of improvements.

According to the Intra Class Correlations findings, the proportion of the total variability in the outcome variable was limited but higher at the NH units’ level, suggesting that some factors embodied in the micro environment where residents live affect the eating dependence as documented by the following path analysis.

Path analysis findings

The analysis explained the 57.7% variance in the eating dependence: the cognitive decline contributed a modest amount to eating dependence; while the greatest contribution was functional dependence, as measured by the Barthel Index. This confirms that the decline of ADLs associated with cognitive decline is progressive and ends with eating dependence [46]. Being female and having night restlessness also contributed to the extent of dependence in feeding. The role of sex can be explained by the older age at which they joint the NH [47] which can imply also a lack of family support [48]. Night restlessness can be an indicator of psychological symptoms of cognitive decline, as well as the consequence of inactivity, that can lead to both an increased day time tiredness, difficulty engaging in activities and increased risk of sleep medication (e.g. benzodiazepine administration) due to staff burden [49].

According to the findings, residents who ate in the dining room close to or surrounded by two residents, rather than eating alone in the bedroom, were more likely to eat independently. The presence of family was also associated with eating independently. The social meaning of meals has already been recognised [16, 30, 38]: sitting surrounded by others can give residents the opportunity to mirror some behaviour [50] while significant others can offer a personalised support thus promoting high quality of interaction during mealtimes [51].

Notwithstanding the effect of individual variables, the NH level variables had a larger direct impact on self-feeding dependence. Less supportive environments have already been documented as significantly associated with eating excess disabilities [6]. We found that cleaner and maintained NH units were associated with an increased dependence in eating. This is possibly because a higher attention to these elements can prevent the degree of freedom to residents (e.g., to eat with fingers, to spill out) and the attitude of the staff to totally compensate his/her difficulties.

Environmental familiarity has emerged as also being associated with increased eating dependence as already documented by Keller et al. [38], possibly because familiar, non-institutional furnishings; however, fully offering a familiar environment can be really challenging in a context of safety measures (e.g. automated beds) and where residents from different cultures co-exist (e.g., the need to share a bedroom). Besides, the ‘sense of home’ is multifactorial, not only including the building and interior design but also familiar food and drinks [52]. Although to a lesser extent, the ‘Visual/tactile stimulation’ dimension has emerged as also associated with an increased eating dependence. Possible explanations include excessive distracting stimuli that should be further studied in underlying mechanisms.

On the other side, allowing space for residents, a safe setting, appropriately lit, and open to the external environment were associated with a reduced eating dependence. Perhaps this is because all of these factors affect resident’s engagement in activity [52, 53] thereby promoting independence. Light, noise, and temperature levels have been already documented as reducing self-feeding abilities [27].

The majority of the above-mentioned environmental factors have been reported to have a moderate indirect effect on the degree of functional dependence and on night restlessness, while higher indirect effects on interventions performed by nursing staff, in some dimensions preventing (‘Orientation/cueing’, ‘Cleanliness’, and ‘Familiarity’) while in others increasing their occurrence (‘Space setting’ and ‘Visual tactile’). This seems to confirm that environmental factors at the unit level can have both direct effects and a mediator effects by explaining the 81.1% variance in the nursing care interventions at the resident level. These findings can be interpreted in two ways: on one side the increased quality of some dimensions of the unit environment (e.g., high orientation, cueing, familiarity) can prevent specific stimulation of the resident during mealtime because the environment is mistakenly perceived as a substitute for individualised care. Alternatively, the high maintenance and cleanliness can discourage attempts of the staff to stimulate to eat alone, with a focus on ensuring the highest level of cleanliness and order as expected by the NH. In this light, not only the dining atmosphere as already documented [27, 54] but also the entire environmental factors at the unit level [55] can affect the eating performance due to its intersection both with eating dependence and the care delivered by the staff.

Differently, the number of interventions performed by nurses at the dining room level (e.g. reducing noises, distractions), have reported no direct effects on eating dependence. The variance arisen in these interventions was mainly explained by the quality of the unit environment as measured by the TESS-NH tool. This suggests that the overall quality of the unit environment affects the number of interventions performed at the dining room, but the negative indirect effects in some cases (e.g., ‘Exit control’, ‘Lighting’) and the positive indirect effects in others (‘Space setting’) requires further investigation.


The outcome variable was measured one time, by observing lunch; residents have been documented to have variations in their eating performance over the day and the time, e.g. with an increased degree of cooperation and the physiological capacity to eat at breakfast [18, 40]. Data on residents were collected from their routine assessments as stored in the database; as a consequence, no data on medications were collected. Data on nursing staff as for example, the staff-to-resident, were not collected given the homogenous care offered in the included NHs; in addition, no data regarding staff attitudes or knowledge regarding how to promote eating independence were collected [7].

Given that no interventions can be considered to date as being a gold standard aimed at maintaining or increasing eating performances in residents living in NH [8], we identified interventions via focus groups and then included them in a checklist. The checklist was then used to document the strategies observed in NHs. Further studies aimed at assessing the check list validity is suggested. Moreover, we evaluated only the number of interventions performed in a set of possible interventions all attempting to stimulate each resident to eat independently according to the available research in the field: the intensity applied to these interventions to each resident has not been measured [28], suggesting that further studies should also consider this aspect and not only the number of interventions performed.


The study findings suggest that the environment of the NH unit generates both direct and indirect positive and negative effects on eating performance, while the amount of environmental interventions enacted by nurses at the dining level have all reported no significant effects.

Eating dependence is a complex phenomenon requiring multiple interventions. Apart from individual unmodifiable predisposing factors (female gender, cognitive decline), some modifiable factors such as: decreased functional dependence, decreased night restlessness, eating in the dining room with others and the presence of close family members, can all reduce eating dependence. At the nursing care level, the number of interventions performed daily to maintain self-feeding independence can prevent self-feeding dependence. However, the largest direct and indirect effect on self-feeding dependence was the quality of the NH unit environment suggesting that there is a need to consider the whole environment where the resident live and not only that of the dining room. Changes are required not only at the dining room level, but in the entire NH unit that should not be ‘perfect’: increased scores in some dimensions (e.g., cleanliness), emerged as both direct and indirect effects on increased eating feeding dependence. Accordingly, further studies aimed at evaluating the best environment capable of maximising eating performance are recommended also with the intent to provide cumulative evidence and to inform the environment design decisions.

Moreover, in studies testing the effectiveness of interventions at the resident level, the quality of the environment should also be evaluated and documented given its role in mediating the degree of eating dependence.

Availability of data and materials

The datasets used and/or analysed during the current study are available from the corresponding author on request.



Activity of Daily Living


Barthel Index


Confidence Intervals


Cognitive Performance Scale


Depression Rating Scale


Edinburgh Feeding Evaluation in Dementia scale


Intra Class Correlation


Nursing aides


Nursing home


Registered Nurse


Standard Deviations


Completely standardized solution


Standard error

Standardized latent variable coefficient


Therapeutic Environment Screening Survey for Nursing Home


  1. 1.

    Fong JH, Mitchell OS, Koh BS. Disaggregating activities of daily living limitations for predicting nursing home admission. Health Serv Res. 2015;50(2):560–78.

    PubMed  Article  Google Scholar 

  2. 2.

    van Het Bolscher-Niehuis MJ, den Ouden ME, de Vocht HM, Francke AL. Effects of self-management support programmes on activities of daily living of older adults: a systematic review. Int J Nurs Stud. 2016;61:230–47.

    Article  Google Scholar 

  3. 3.

    Saucedo Figueredo MC, Morilla Herrera JC, San Alberto Giraldos M, López Leiva I, León Campos Á, Martí García C, et al. Validation of the Spanish version of the Edinburgh Feeding Evaluation in Dementia Scale for older people with dementia. PLoS One. 2018;13(2):e0192690.

    PubMed  PubMed Central  Article  CAS  Google Scholar 

  4. 4.

    Liu W, Cheon J, Thomas SA. Interventions on mealtime difficulties in older adults with dementia: a systematic review. Int J Nurs Stud. 2014;51(1):14–27.

    PubMed  Article  Google Scholar 

  5. 5.

    Lane NE, Wodchis WP, Boyd CM, Stukel TA. Disability in long-term care residents explained by prevalent geriatric syndromes, not long-term care home characteristics: a cross-sectional study. BMC Geriatr. 2017;17(1):49.

    PubMed  PubMed Central  Article  Google Scholar 

  6. 6.

    Liu W, Galik E, Boltz M, Nahm ES, Resnick B. Optimizing eating performance for older adults with dementia living in long-term care: a systematic review. Worldviews Evid-Based Nurs. 2015;12(4):228–335.

    PubMed  Article  Google Scholar 

  7. 7.

    Herke M, Fink A, Langer G, Wustmann T, Watzke S, Hanff AM, Burckhardt M. Environmental and behavioural modifications for improving food and fluid intake in people with dementia. Cochrane Database of Syst Rev. 2018;7:CD011542.

    Google Scholar 

  8. 8.

    Abdelhamid A, Bunn D, Copley M, Cowap V, Dickinson A, Gray L, et al. Effectiveness of interventions to directly support food and drink intake in people with dementia: systematic review and meta-analysis. BMC Geriatr. 2016;16:26.

    PubMed  PubMed Central  Article  CAS  Google Scholar 

  9. 9.

    Lin LC, Huang YJ, Su SG, Watson R, Tsai BWJ, Wu SC. Using spaced retrieval and Montessori-based activities in improving eating ability for residents with dementia. Int J Geriatr Psychiatry. 2010;25(10):953–9.

    PubMed  Article  Google Scholar 

  10. 10.

    Bunn DK, Abdelhamid A, Copley M, Cowap V, Dickinson A, Howe A, et al. Effectiveness of interventions to indirectly support food and drink intake in people with dementia: eating and drinking well IN dementiA (EDWINA) systematic review. BMC Geriatr. 2016;16:89.

    PubMed  PubMed Central  Article  Google Scholar 

  11. 11.

    Douglas JW, Lawrence JC. Environmental considerations for improving nutritional status in older adults with dementia: a narrative review. J Acad Nutr Diet. 2015;115(11):1815–31.

    PubMed  Article  Google Scholar 

  12. 12.

    Herke M, Fink A, Langer G, Wustmann T, Watzke S, Hanff AM, Burckhardt M. Environmental and behavioural modifications for improving food and fluid intake in people with dementia (protocol). Cochrane Database Syst Rev. 2015;2:CD011542.

    Google Scholar 

  13. 13.

    Easton T, Milte R, Crotty M, Ratcliffe J. Where’s the evidence? A systematic review of economic analyses of residential aged care infrastructure. BMC Health Serv Res. 2017;17(1):226.

    PubMed  PubMed Central  Article  Google Scholar 

  14. 14.

    Slaughter SE, Hayduk LA. Contributions of environment, comorbidity, and stage of dementia to the onset of walking and eating disability in long-term care residents. J Am Geriatr Soc. 2012;60(9):1624–31.

    PubMed  Article  Google Scholar 

  15. 15.

    Slaughter SE, Morgan DG. Functional outcomes of nursing home residents in relation to features of the environment: validity of the professional environmental assessment protocol. J Am Med Dir Assoc. 2012;13:487.

    PubMed  Article  Google Scholar 

  16. 16.

    Amella EJ, Aselage MB. Mealtime difficulties. In: Boltz M, Capezuti E, Fulmer T, Zwicker D, editors. Evidence-based geriatric nursing protocols for best practice. 4th ed. New York: Springer Publishing Company; 2012. p. 453–68.

    Google Scholar 

  17. 17.

    von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP. The Strengthening the reporting of observational studies in epidemiology (STROBE) statement: guidelines for reporting observational studies. J Clin Epidemiol. 2008;61(4):344–9.

    Article  Google Scholar 

  18. 18.

    Palese A, Grassetti L, Bandera D, Zuttion R, Ferrario B, Ponta S, et al. High feeding dependence prevalence in residents living in Italian nursing homes requires new policies: findings from a regionally based cross-sectional study. Health Policy. 2018;122(3):301–8.

    PubMed  Article  Google Scholar 

  19. 19.

    Watson R. The Mokken scaling procedure (MSP) applied to the measurement of feeding difficulty in elderly people with dementia. Int J Nurs Stud. 1996;33(4):385–93.

    CAS  PubMed  Article  Google Scholar 

  20. 20.

    Bagnasco A, Watson R, Zanini M, Rosa F, Rocco G, Sasso L. Preliminary testing using Mokken scaling of an Italian translation of the Edinburgh feeding evaluation in dementia (EdFED-I) scale. Appl Nurs Res. 2015;28(4):391–6.

    PubMed  Article  Google Scholar 

  21. 21.

    Mahoney FI, Barthel DW. Functional evaluation: the Barthel index. A simple index of independence useful in scoring improvement in the rehabilitation of the chronically ill. Md State Med J. 1965;14:56–61.

    Google Scholar 

  22. 22.

    Morris JN, Fries BE, Mehr DR, Hawes C, Phillips C, Mor V, Lipsitz LA. MDS Cognitive Performance Scale. J Gerontol. 1994;49(4):M174–82.

    CAS  PubMed  Article  Google Scholar 

  23. 23.

    Burrows AB, Morris JN, Simon SE, Hirdes JP, Phillips C. Development of a minimum data set-based depression rating scale for use in nursing homes. Age Aging. 2000;29(2):165–72.

    CAS  Article  Google Scholar 

  24. 24.

    Fries BE, Simon SE, Morris JN, Flodstrom C, Bookstein FL. Pain in U.S. nursing homes: validating a pain scale for the minimum data set. Gerontologist. 2001;41(2):173–9.

    CAS  PubMed  Article  Google Scholar 

  25. 25.

    Pascazio L, Morosini P, Bembich S, Nardone I, Clarici A, Barbina L, et al. Description and validation of a geriatric multidimensional graphical instrument for promoting longitudinal evaluation. Arch Gerontol Geriatr. 2009;48(3):317–24.

    PubMed  Article  Google Scholar 

  26. 26.

    Sloane PD, Mitchell CM, Weisman G, Zimmerman S, Foley KM, Lynn M, et al. The therapeutic environment screening survey for nursing homes (TESS-NH): an observational instrument for assessing the physical environment of institutional settings for persons with dementia. J Gerontol B Psychol Sci Soc Sci. 2002;57(2):S69–78.

    PubMed  Article  Google Scholar 

  27. 27.

    Kovach CR, Noonan PE, Schlidt AM, Wells T. A model of consequences of need-driven, dementia-compromised behavior. J Nurs Scholarsh. 2005;37(2):134–40 discussion 140.

    PubMed  Article  Google Scholar 

  28. 28.

    Liu W, Jao YL, Williams K. The association of eating performance and environmental stimulation among older adults with dementia in nursing homes: a secondary analysis. Int J Nurs Stud. 2017;71:70–9.

    PubMed  PubMed Central  Article  Google Scholar 

  29. 29.

    Bornhöft G, Maxion-Bergemann S, Wolf U, Kienle GS, Michalsen A, Vollmar HC, et al. Checklist for the qualitative evaluation of clinical studies with particular focus on external validity and model validity. BMC Med Res Methodol. 2006;6:56.

    PubMed  PubMed Central  Article  Google Scholar 

  30. 30.

    Chang CC, Roberts BL. Feeding difficulty in older adults with dementia. J Clin Nurs. 2008;17:2266–74.

    PubMed  Article  Google Scholar 

  31. 31.

    Bales CW, Locher JL, Salzman E. Handbook of clinical nutrition and aging. 3rd ed. New York: Humana Press; 2015.

    Google Scholar 

  32. 32.

    van Dalen-Kok AH, Pieper MJ, de Waal MW, Lukas A, Husebo BS, Achterberg WP. Association between pain, neuropsychiatric symptoms, and physical function in dementia: a systematic review and meta-analysis. BMC Geriatr. 2015;15:49.

    PubMed  PubMed Central  Article  Google Scholar 

  33. 33.

    Konno R, Kang HS, Makimoto K. The best evidence for minimizing resistance-to-care during assisted personal care for older adults with dementia in nursing homes: a systematic review. JBI Libr Syst Rev. 2012;10(58):4622–32.

    PubMed  Google Scholar 

  34. 34.

    Cohen LW, Zimmerman S, Reed D, Sloane PD, Beeber AS, Washington T, et al. Dementia in relation to family caregiver involvement and burden in long-term care. J Appl Gerontol. 2014;33(5):522–40.

    PubMed  Article  Google Scholar 

  35. 35.

    Tarling R. Statistical modeling for social researchers. Principles and practice. 1st ed. New York: Routledge; 2008.

    Google Scholar 

  36. 36.

    Arbuckle JL. Full information estimation in the presence of incomplete data. In: Marcoulides GA, Schumacker RE, editors. [Eds] Advanced structural equation modeling: Issues and Techniques. Mahwah: Lawrence Erlbaum Associates; 1996.

    Google Scholar 

  37. 37.

    Yves R. Lavaan: an R package for structural equation modeling. J Stat Softw. 2012;48(2):1–36.

    Google Scholar 

  38. 38.

    Keller HH, Carrier N, Slaughter SE, Lengyel C, Steele CM, Duizer L, et al. Prevalence and determinants of poor food intake of residents living in long-term care. J Am Med Dir Assoc. 2017;18(11):941–7.

    PubMed  Article  Google Scholar 

  39. 39.

    Liu W, Galik E, Boltz M, Nahm ES, Resnick B. Optimizing eating performance for older adults with dementia living in long-term care: a systematic review. Worldviews Evid-Based Nurs. 2015;12(4):228–35.

    PubMed  Article  Google Scholar 

  40. 40.

    Palese A, Menegazzi G, Tullio A, Zigotti Fuso M, Hayter M, Watson R. Functional decline in residents living in nursing homes: a systematic review of the literature. J Am Med Dir Assoc. 2016;17(8):694–705.

    PubMed  Article  Google Scholar 

  41. 41.

    Tyler DA, Feng Z, Leland NE, Gozalo P, Intrator O, Mor V. Trends in postacute care and staffing in US nursing homes, 2001-2010. J Am Med Dir Assoc. 2013;14(11):817–20.

    PubMed  Article  Google Scholar 

  42. 42.

    Konno R, Kang HS, Makimoto K. A best-evidence review of intervention studies for minimizing resistance-to-care behaviours for older adults with dementia in nursing homes. J Adv Nurs. 2014;70(10):2167–80.

    PubMed  Article  Google Scholar 

  43. 43.

    Evripidou M, Charalambous A, Middleton N, Papastavrou E. Nurses’ knowledge and attitudes about dementia care: systematic literature review. Perspect Psychiatr Care. 2019;55(1):48–60.

    PubMed  Article  Google Scholar 

  44. 44.

    McAdam R, Mason B, McCrory J. Exploring the dichotomies within the tacit knowledge literature: towards a process of tacit knowing in organizations. J Knowledge Manag. 2007;11(2):43–59.

    Article  Google Scholar 

  45. 45.

    Slaughter S, Calkins M, Eliasziw M, Reimer M. Measuring physical and social environments in nursing homes for people with middle – to late stage of dementia. J Am Geriatr Soc. 2006;54(9):1436–41.

    PubMed  Article  Google Scholar 

  46. 46.

    Inouye SK, Westendorp RG, Saczynski JS. Delirium in elderly people. Lancet. 2014;383(9920):911–22.

    PubMed  Article  Google Scholar 

  47. 47.

    Watson R, Palese A, Zuttion R, Ferrario B, Ponta S, Hayter M. Identifying longitudinal sustainable hierarchies in activities of daily living. Arch Gerontol Geriatr. 2017;71:122–8.

    PubMed  Article  Google Scholar 

  48. 48.

    Montesanto A, De Rango F, Pirazzini C, Guidarelli G, Domma F, Franceschi C, Passarino G. Demographic, genetic and phenotypic characteristics of centenarians in Italy: Focus on gender differences. Mech Ageing Dev. 2017;165(Pt B):68–74.

    PubMed  Article  Google Scholar 

  49. 49.

    Feng Z, Falkingham J, Liu X, Vlachantoni A. Changes in living arrangements and mortality among older people in China. SSM Popul Health. 2016;3:9–19.

    PubMed  PubMed Central  Article  Google Scholar 

  50. 50.

    Hessler JB, Schäufele M, Hendlmeier I, Junge MN, Leonhardt S, Weber J, Bickel H. Behavioural and psychological symptoms in general hospital patients with dementia, distress for nursing staff and complications in care: results of the general hospital study. Epidemiol Psychiatr Sci. 2018;27(3):278–87.

    CAS  PubMed  Article  Google Scholar 

  51. 51.

    Rizzolatti G, Fabbri-Destro M, Cattaneo L. Mirror neurons and their clinical relevance. Nat Clin Pract Neurol. 2009;5(1):24–34.

    PubMed  Article  Google Scholar 

  52. 52.

    Aselage MB, Amella EJ, Watson R. State of the science: alleviating mealtime difficulties in nursing home residents with dementia. Nurs Outlook. 2011;59(4):210–4.

    PubMed  Article  Google Scholar 

  53. 53.

    van Hoof J, Verbeek H, Janssen BM, Eijkelenboom A, Molony SL, Felix E, et al. A three-perspective study of the sense of home of nursing home residents: the views of residents, care professionals and relatives. BMC Geriatr. 2016;16(1):169.

    PubMed  PubMed Central  Article  Google Scholar 

  54. 54.

    Edwards JS, Gustafsson IB. The room and atmosphere as aspects of the meal: a review. J Foodservice. 2008;19(1):22–34.

    Article  Google Scholar 

  55. 55.

    Elf M, Nordin S, Wijk H, Mckee KJ. A systematic review of the psychometric properties of instruments for assessing the quality of the physical environment in healthcare. J Adv Nurs. 2017;73(12):2796–816.

    PubMed  Article  Google Scholar 

  56. 56.

    Garante per la Protezione dei Dati Personali. Autorizzazione n. 3/2012 - Autorizzazione al trattamento dei dati sensibili da parte degli organismi di tipo associativo e delle fondazioni. Downloaded from: Accessed 8 Aug 2019.

Download references


Not applicable.


No funding was obtained for this study.

Author information




AP, RW, MH conceived and managed the study. AP, AD, TK, ML collected data; AP and LG designed the model, analysed the data; AP, LG and VB discussed the meanings of the findings. RW and MH assisted in the study design, helped to draft the manuscript and supervised it. All authors read and approved the final manuscript.

Corresponding author

Correspondence to Alvisa Palese.

Ethics declarations

Ethics approval and consent to participate

The Ethical approval to access the database where data of residents were stored was given by the Regional Ethical Committee - Comitato Etico Unico Regionale F.V.G., Italy (prot. N. 29747, 2018) in the context of a large retrospective study. According to the ‘Autorizzazione n. 9/2012 – Autorizzazione general al trattamento dei dati personali effettuato per scopi di ricerca scientifica’ (Authorization 9/2012 – General authorization for the processing of personal data carried out for scientific research purposes, [56]), the written informed consent of the residents was not required. To access the NHs the approval was provided from the Health Care Institution responsible and each NH involved (prot. N. 66935, 2017) who, according to the observational nature of the study, not-intrusive and not changing the usual care provided to the residents, did not require the collection of their written consent to participate. Specifically, data collected via observation was the that same routinely collected and stored in the database above-mentioned according to the minimum data set of the Nursing Homes involved. However, given that eating difficulties were assessed in different times according to the needs of the residents and their admission in the Nursing Home, eating performance was re-assessed on a specific day by trained researchers with the intent to increase the rigour of the study.

Before the data collection, all residents were informed with regards to the study aims on personal basis by the PI (Alvisa Palese); in the dining rooms there were also available informative sheets regarding the aims of the ongoing study projects. The day of the observation, the information regarding the study aims was repeated. All residents gave their verbal informed consent to participate.

Consent for publication

Not applicable.

Competing interests

The authors declare that they have no competing interests.

Additional information

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Additional file 1.

Correlations between explanatory variables (at the individual, nursing care and NH levels) and the outcome variable.

Additional file 2.

Path Analysis Findings.

Rights and permissions

Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (, which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver ( applies to the data made available in this article, unless otherwise stated.

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Palese, A., Grassetti, L., Bressan, V. et al. A path analysis on the direct and indirect effects of the unit environment on eating dependence among cognitively impaired nursing home residents. BMC Health Serv Res 19, 775 (2019).

Download citation


  • Eating dependence
  • Feeding difficulties
  • Mealtime difficulties
  • Nursing home
  • Path analysis
  • Dementia
  • Cognitive decline