Secondary data collected from the Making Most of Mealtimes (M3) study is used for this analysis. The M3 study is a multi-site, cross sectional study that collected data from 32 LTC homes in four Canadian provinces: Alberta, Manitoba, New Brunswick and Ontario [19]. Data were collected at the resident, dining room, and home levels. The purposes of M3 were to: 1) determine the food and fluid intake of residents in LTC and compare these to recommendations, and 2) identify the predictors of food and fluid intake. DEAP was specifically included to measure the physical mealtime environment as a potential determinant of food and fluid intake.
Sample
Eight LTC homes were purposively recruited in each of the four provinces. Homes that were considered for inclusion had: 1) been operating for at least 6 months, 2) a minimum of 50 residents that met the resident eligibility criteria, and 3) agreed to participate in the data collection providing full cooperation for all procedures. For-profit and not-for-profit homes were recruited and homes with special characteristics were chosen to promote sample diversity (e.g., culturally based homes) [19]. Within each home, data were collected on one to four randomly selected care units; 82 dining rooms were assessed during data collection. Eligible residents were randomly sampled from the units that were selected, with twenty residents included from each home; these residents were representative of the study units [19]. The eligibility criteria for resident participation included: 1) residing on the units selected, 2) being over the age of 65, 3) requiring a minimum of 2 h each day of nursing care, 4) residing in the home for at least 1 month, and 4) they, or a substitute decision maker, provided informed consent to participate if the resident had cognitive impairment (i.e., Cognitive Performance Scale 3+). Resident exclusion criteria included: 1) residing in the home for less than 1 month, 2) medically unstable at the time of recruitment (e.g. recent hospital transition), 3) short term admission at the time of recruitment, 4) requiring tube feeding, 5) deemed by home staff to be at the end of life, and/or 6) having an advanced directive that excluded them from research. A total of 640 residents were recruited with a final sample of 639, as one participant withdrew consent. Eligible staff included nursing, recreation and/or dietary that were regular part-time or full-time employees. A minimum of 10 employees working on the selected units were recruited for data collection [19]. A site survey was completed by home management to describe the homes and units included in the sample (e.g. profit status, part of a continuum of care, beds (total and study units), age of home, renovations in unit in past 5 years).
Measures
The DEAP tool
While the DEAP tool has not been tested for construct validity, this tool has demonstrated inter-observer reliability [13]. The homelikeness and functionality summary scales had good intraclass correlation coefficient values of 0.68 and 0.70, respectively [13]. The DEAP tool was used to assess the physical environment in each dining area by a trained research coordinator from each of the four provinces; it consists of primarily observational components as well as two questions asked of staff members. DEAP is designed to be an assessment of the physical space to compare and contrast physical features only, among dining spaces. Thus the observations are performed once at the beginning of data collection for each home, when the dining room was empty. Assessors were trained to walk throughout the dining space, observing the room/area from a variety of vantage points to complete the tool. Information was first recorded about the unit and the dining room, specifically: unit type (dementia care unit or general care unit). Numbers of tables chairs stools or chairs for staff and entry ways/exits were counted. By considering table/chair position for residents and vantage points for windows, percentage of residents with a clear view of the outside garden/green space was rated (≤24%, 25–49%, 50–74% 75%+; score 0 to 3). The following items were noted dichotomously as present (yes = 1) or absent (no= 0): use of adjustable tables; contrast between floor/table/dishes; rounded edges of furniture; presence of a posted menu; detergents/non-edibles secured; stove and other dangerous items secured; presence of a television and/or clock; dining room open between meals (i.e., door open, space not locked or closed off between meals); adjacent family kitchen with residential appliances, private family dining area; short distance from most bedrooms and visible from bedrooms; accessible washroom near dining room; accessible beverage services; and accessible main kitchen/servery (i.e., no barrier/door limiting resident access between meals). Data were also collected on the functionality of the space, including lighting intensity (0 = poor, 1 = reasonable, 2 = plenty), and glare (0 = strong, 1 = some, 2 = minimal). Safety and security information was assessed by categorizing the space on the size of the dining room, length of pathways for meal delivery, presence of obstacles/clutter, the ability of staff to view all residents and the use of restraints. Ratings for size, pathway and obstacles/clutter were: one (e.g. large institutional space >30 residents; long pathways for meal delivery >25 ft; several obstacles/ clutter), two (e.g. moderately large space [20–30 residents]; moderate length of pathways [15–25 ft]; some clutter/obstacles) and three (e.g. homelike space <20 residents; short pathways for meal delivery <15 ft; no clutter/obstacles) with a higher score indicating a more functional dining room. Based on the size of the dining space and layout of servery and tables, the space was rated on capacity for staff members to supervise residents (i.e., could they feasibly view all residents in the dining space or were there potential obstacles). Large space and/or awkward layout that would hinder staff observation of residents was given a score of 0, staff being able to easily view and access almost all residents was provided a score of 1 and staff being able to view all residents and get to them easily and quickly was give a score of 2. Social potential of the space was rated by the presence of a mixture of seating arrangements on a scale from zero to two, signifying one option (score = 0), 2 or 3 options (score = 1) and >3 options of seating arrangements (score = 2). Although most aspects of DEAP are observed by assessors, information on two questions is attained from staff. Assessors were trained to ask staff if residents’ opinions on the physical environment (e.g., light, noise, temperature; scored as yes = 1, no = 0, unsure = 9) were respected and acted on and if physical restraints were used in the dining room (yes = 1 and no = 0; unsure = 9). After completing DEAP, the assessor subjectively rated the overall space on two separate scales, homelikeness and functionality of the environment, with the scales ranging from 1(low) to 8 (high) [14]. For analysis only, researchers derived a total score by tallying individual variables/characteristics using the above item coding, resulting in a maximum DEAP score of 56, where a higher score indicates more physical features that are supportive of dining.
DEAP staff training and protocol
Research coordinators were trained to complete all M3 measures, including the DEAP, during 3-days of in-person training. Specific to the DEAP, researchers reviewed the tool with coordinators question by question, to clarify intent of items. Pictures of LTC physical dining rooms were used to demonstrate the physical qualities to be attended to when scoring. The research coordinators then observed four dining rooms to practice their assessments; results were compared and clarification provided where required to promote consistency among raters [19].
Theoretical constructs for comparison
Two standardized measures collected at the home and dining room level were used to assess construct validity. The 50 item staff-completed Person-Directed Care (PDC) questionnaire assesses staff’s perceptions with respect to personhood, comfort care, autonomy, knowing the person and support for relationships [20]. The staff PDC questionnaire has demonstrated face validity and conceptually distinct constructs (Cronbach’s alpha 0.86–0.91) [20]. In the M3 study, the staff PDC questionnaire was completed by 10–20 staff that provided care in the study units. The Mealtime Scan (MTS) is an observational tool that assesses the physical and psychosocial environments [21] while a meal is being consumed. Assessors observe an entire mealtime from beginning to end (on average 1 h), assessing various physical, social and care activities with respect to mealtimes as a meal is in process. As each meal is unique, dependent on time of day, who is present and the types of interactions that occur, multiple observations are required to assess the mealtime environment. This tool is face valid and is based on theoretical domains of physical, social, and relationship and person-centred practices in the dining room [21]. Embedded within the MTS is the Mealtime- Relational Care Checklist (M-RCC) that can be used on its own to assess relationship and person-centred care (R/PCC) practices and behaviours exhibited by staff with residents during mealtimes. The MTS includes three summary scales, on a scale from 1(low) to 8 (high), to assess the physical, social and person-centred environments. The MTS has been deemed an inter-rater reliable tool with good intraclass correlations (0.65–0.85) for the three summary scales and M-RCC [21]. The MTS was completed by the trained provincial coordinator and occasionally by the trained research assistants due to scheduling challenges with data collection. Assessments of the dining environment using the Mealtime Scan were completed 4–6 times in each unit’s dining room (n = 82) with observations at breakfast, lunch and dinner; the mean of scales was used in analyses. Individual items on MTS (e.g. number of residents, staff, residents:staff) were extracted to help describe the dining environments.
Resident level measures were also used to determine the DEAP’s construct validity. The interRAI Long Term Care Form is a standardized assessment tool used to gather information on health, cognitive and quality of life domains of LTC residents [22]. The trained provincial coordinators collected this data by interviewing staff members that were familiar with the resident’s current care and behaviour; interviewing was necessary as a) some provinces did not routinely use the interRAI instruments, b) to promote consistency among assessments, and c) as care needs and characteristics were desired to be concurrent with other data collected in M3 [19]. The items from the interRAI Long Term Care Form that were used to determine construct validity of DEAP were the Cognitive Performance Scale (CPS; maximum score of 6, with a higher score indicating a higher degree of cognitive impairment) and the Depression Rating Scale (DRS; maximum score of 33, with a higher score indicating increased presence of depressive symptoms) [22,23,24]. Malnutrition risk was measured using the Mini Nutritional Assessment- Short Form (MNA-SF), a valid and reliable instrument for assessing nutritional risk [25, 26]. Food intake for each resident was collected for three non-consecutive days, including one weekend day; assessments of food intake for an individual resident typically occurred during a 10-day period. All three meals for each day (nine meals in total) were observed and food intake determined by weighing main plate items before and after meal consumption. Side dishes and beverages were estimated considering the home’s portion sizes and capacity of serving dishes. Food and fluid intake between meals were estimated by researchers asking staff/resident/or family members what had been consumed between meals and estimating portion sizes. Food and fluid consumption after the evening meal was estimated and recorded by home staff. The detailed process for collecting food and fluid intake data is presented in the protocol paper [19]. LTC recipes were gathered and assembled in the nutrient analysis program Food Processor (version 10.14.1) and average energy and protein intake was estimated across the days of intake for each resident in this analysis.
Statistical analysis
Two analyses were used to determine construct validity of the DEAP summative scales. First, a regression analysis determined those DEAP variables that predicted the homelikeness and functionality summative scales to demonstrate if and how these two constructs are similar and different. Bivariate analysis determined those DEAP variables more highly associated with these scales; those that had a p-value <0.25 were included in the initial multivariate model. Final multivariate models resulted from inclusion of these variables and backwards elimination using a p-value of <0.05 to determine the order for removal and retention of variables. The final model for each scale was achieved when all variables had a p-value of <0.05; potential interactions were also assessed. When the multivariate model for functionality was assessed, the DEAP variable “respecting and responding to resident’s opinions” was found to interact with the variable “residents are able to see the dining area from their bedroom”. Both of these variables were eliminated from the multivariate model, as the first variable had missing data for 12/82 dining rooms and concerns about accuracy of reporting by home staff on this question, while the second variable was not included due to low prevalence (3/82). The number of exits was also eliminated from the multivariate model due to the inability to differentiate between one exit and open concept dining rooms; further, this variable was also highly skewed. Collinearity tests were performed using the tolerance values and Cooks d to gather information on the existing relationships between each of the variables that remained in the final model. As tolerance values were >0.2 in all models, it was determined that multicollinearity was not present. Upon conducting Cooks d, outliers were detected and removed; however, this did not alter the interpretation of the model.
The second analysis to determine construct validity for the DEAP summary scales of homelikeness and functionality was based on their association with theoretically relevant resident and home/unit level measures. Descriptive statistics were computed for the staff PDC, MTS scales, M-RCC, MNA-SF, CPS, DRS and resident energy and protein intake. A Spearman rho correlation was computed for each instrument with the homelikeness and functionality scales, with p < 0.05 indicating statistical significance. To determine the association between CPS and the homelikeness and functionality scales, the CPS score was dichotomized into none to mild cognitive impairment (scores 0–2) and moderate to severe cognitive impairment (scores 3–6). Using a Student t-test, it was determined if homelikeness and functionality varied by cognitive status. All analyses were performed using SAS University (version 9.4).