“Putting Quality Food on the Tray”: Assessing Patients’ Expectations and Experiences of Meals Served in Sixteen Ontario Hospitals

Abstract

Background: Malnutrition is prevalent among hospital patients and occurs in up to 45% of Canadian medical and surgical patients. Hospital malnutrition is associated with detrimental outcomes for patients, such as further morbidity and mortality and increases hospital-associated costs. Low food intake is a primary contributor to the development or worsening of malnutrition and may be influenced by factors such as poor appetite, illness, and perceptions of poor food quality and disliking food served. Currently there are no standards to assess patients’ experiences with meals in Ontario hospitals and tools used to assess patient satisfaction are limited in their approach with respect to assessing both patients’ expectations and ratings of a served meal and have not always demonstrated good measures of validity or reliability.

Purposes: 1) Assess the internal consistency reliability of the Hospital Food Experience Questionnaire (HFEQ) in addition to construct validity with the overall rating of meal quality at a meal and predictive validity with food intake. 2) Measure patients’ hospital food and food-related expectations in addition to ratings of a single meal served in hospital. 3) Determine patient and hospital characteristics associated with three measures of meal quality from the HFEQ (i.e. a single overall meal quality rating, HFEQ score and short form HFEQ (HFEQ-sv) score). 4) Asses meal intake and specific foods served and consumed. 5) Determine the independent effect of three measures of meal quality in predicting overall food intake at a meal while considering selected patient and hospital characteristics.

Methods and Findings: The multi-site study collected data from sixteen Ontario hospitals and 1,087 patients. Data was collected at the hospital, unit, and patient-level. The original HFEQ included 23 questions assessed using a 5-point Likert scale. Food (n = 6) and food-related (n = 10) expectations were assessed by “not important” (1) and “very important” (5), while meal ratings (n = 7) were assessed by “very poor” (1) and “very good” (5). The HFEQ was completed at a single meal where overall hospital expectations and ratings of the meal served were assessed. Overall food intake and intake of specific food groups were also assessed. Three studies resulted from this thesis work.

1) Internal consistency reliability was assessed using Cronbach’s alpha and principal components analysis (PCA). The three subscales of the HFEQ (food expectations, food-related expectations and meal ratings) and the entire HFEQ demonstrated good internal reliability (0.80-0.91) and all but one of the HFEQ questions (the importance of food served being healthy) clustered together in PCA to reveal the following factors: Meal Ratings, Food Traits, Food-Related Traits, Meeting Patients’ Dietary and Accessibility Needs, and Food Familiarity and Source. Four ordinal logistic regressions were conducted with the three subscales and entire HFEQ with overall meal quality rating. The three subscales and overall HFEQ demonstrated construct validity with overall meal quality (p < .050). Specifically the expectations of taste, local food provision, easy to open packaging and easy to eat foods in addition to meal ratings of taste, appearance, texture, temperature and combination of food served were significantly associated with overall meal quality (p < .050). A 5x2 chi-square revealed that overall meal quality rating was significantly associated with food intake at a single meal, where patients with lower overall meal quality ratings experienced low food intake. Cross validation with all 22 items of the HFEQ (relevant items identified with PCA) and overall meal intake was conducted to attempt to shorten the HFEQ, however only the expectation of food choice and meal taste ratings were significantly associated with food intake (P < .050). The shortened HFEQ-sv was determined using the 10 items identified in convergent validity analyses and overall meal quality (n = 11). A final binary logistic regression was conducted with these 11 items, which revealed that the HFEQ-sv was significantly associated with food intake, with the importance of food choice, and meal ratings of texture and taste being significantly associated with food intake (p < .050).

2) Food expectations most frequently rated as “very important” included: taste, freshness, and healthiness (73.75%, 70.45%, and 64.60%, respectively). Food-related expectations most frequently were rated as “very important” included: meeting patients’ dietary needs, appropriate temperatures, easy to eat foods and receiving a sufficient amount of food (69.54%, 67.45%, 62.89, and 61.07%, respectively). Median sensory meal ratings were all scored at 4 (i.e. “good”). Overall meal quality was rated as “very good” by 28.92% of patients, while meal temperature and taste received the most “very good” ratings (34.56%, and 30.09%, respectively). Average HFEQ and HFEQ-sv scores were 90.60 (SD 10.83) and 44.22 (SD 6.55), respectively.

3) Three regressions (1 ordinal, 2 linear) were conducted to test the association between patient and hospital characteristics and measures of meal quality (i.e. overall meal quality rating, HFEQ and HFEQ-sv scores). Age was significantly associated with all three measures of meal quality, while gender was only significantly associated with HFEQ and HFEQ-sv scores (P < .050). Older and female patients were significantly more likely to rate meal quality more favourably. No other patient characteristics were significantly associated with any of the three measures of meal quality. Hospital characteristics associated with meal quality varied depending on the meal quality measure used as the dependent variable. Only hospital size was significantly associated with overall meal quality rating. Foodservice model and proportion of foodservice budget allocated to local food were significantly associated with HFEQ score. Average daily food cost per patient was only significantly associated with HFEQ-sv score.

4) Proportions assessed overall meal intake and intake of specific food items. Approximately 29% of patients consumed ≤50% while 42% consumed all of their meal. Beverages (i.e. tea/coffee, milk and tea), soup, and pudding/Jell-O were items frequently consumed by all patients, even those experiencing low overall meal intake.

5) Binary logistic regressions with food intake as the outcome were conducted considering hospital characteristics with each of the three meal quality measures, and selected patient characteristics stratified by gender. No hospital or patient characteristics were significantly associated with food intake when any measure of meal quality was considered, however meal quality ratings were significantly associated with food intake (ps < .050). AIC and max-rescaled R2 was determined for each model to assess model fit and explained variance in food intake. The models where HFEQ-sv was used as the meal quality measure demonstrated the best compromise between model fit and explained variance when patient characteristics and hospital characteristics were considered, suggesting that this version of the HFEQ may be most appropriate to assess meal quality while considering patient and hospital characteristics.

Conclusion: The HFEQ demonstrated good internal reliability and convergent validity with overall meal quality rating, and both the HFEQ and HFEQ-sv demonstrated predictive validity with food intake at a single meal. Patients generally rated food expectations and food-related attributes highly, suggesting that patients have high expectations of meals served in hospital. Ratings of meals served did not always meet these high expectations. Patient age and gender were significantly associated with perceptions of meal quality, while hospital characteristics associated with meal quality were dependent on which measure of meal quality was used. Approximately 29% of patients consumed ≤50% of their meal. Patients experiencing low food intake were more likely to consume soft, or fluid foods. When patient and hospital characteristics were considered, only perceptions of meal quality were significantly associated with food intake, where higher scores of meal quality were associated with increased odds of patients consuming their entire meal. The HFEQ-sv demonstrated the best compromise between model fit and explained variance in food intake when the three measures of meal quality (i.e. overall meal quality rating, HFEQ and HFEQ-sv scores) were considered. The HFEQ or HFEQ-sv should be implemented in practice or used in future research to assess perceptions of meal quality and aim to improve the meal experience and support subsequent food intake for patients in hospital.