Development and evaluation of a decision support system to prevent and treat disease-related malnutrition

Background: About 30% of patients in hospitals are malnourished or at risk of malnutrition. Malnutrition is associated with increased morbidity, longer convalescence, prolonged length of hospital stay, higher readmission rates and premature death. Several barriers are associated with the current practice of nutritional care and treatment in hospitals and the methods are perceived to be cumbersome. Efficient systems and tools to follow up and monitor nutritional care and treatment for the large group of malnourished patients are currently lacking and little is known about the effects and implementation of such systems in clinical practice. Aims: The aim of this PhD thesis was to develop and evaluate a decision support system to prevent and treat disease-related malnutrition in hospitalized patients. We also aimed to explore the readiness and potential barriers to and facilitators of use of such a system and to study the effects of this system in a clinical hospital setting. Methods: A combination of quantitative and qualitative methods was used to investigate the aims. The MyFood decision support system was developed with four main functions: 1) patient registration; 2) dietary recording; 3) evaluation of intake compared with nutritional requirements; and 4) report to nurses, including recommendations for nutritional treatment and a nutrition care plan. To validate the dietary recording function in the MyFood system, 32 hospitalized patients were included and told to record their nutritional intake in the MyFood app for 2 days. Their recordings were compared with digital photographs of the meals combined with partial weighing of meal components. A qualitative study was performed to explore the current practice with nutritional care and treatment in the hospital departments, and barriers and facilitators perceived by health-care professionals for the use of the MyFood system as part of their clinical practice. Four focus groups were conducted with 20 nurses, plus individual interviews with 3 middle managers, 2 physicians and 2 registered dietitians. The Consolidated Framework for Implementation Research (CFIR) was used to develop the interview guide and analyse the results. To investigate the effects and implementation of the MyFood system in a clinical hospital setting, a randomized controlled trial (RCT) was conducted among 100 patients. The patients assigned to the intervention group were told to use the MyFood system during their hospital stay and the nurses were encouraged to follow up the patients with the system. The control group followed routine care. The patients’ body weight was measured and their body composition estimated twice each week. The Nutritional Risk Screening (NRS 2002) and the Patient-Generated Subjective Global Assessment Short Form (PG-SGA-SF) were filled in weekly by the researchers and patients, respectively. Data on nutritional treatment, nutritional documentation and the use of nutrition care plans were gathered from the electronic patient record. Data on length of stay were obtained from the hospital administration system. Results: The MyFood decision support system was developed with an interface consisting of an app for tablet computers and a webserver. The dietary recording function in the MyFood app was found satisfactory in its estimate of the consumption of energy, protein and liquids for the majority of patients. About 70% of the patients had 80% or higher agreement between the estimated intake of energy, protein and liquids based on the MyFood app and the reference method. With regard to the intake of food and beverages, the agreement between the methods varied according to food group. Most of the patients experienced the MyFood app as easy to use and navigate, and reported to become more aware of their nutritional requirements after 2 days’ use. With regard to the current situation with nutritional care and treatment at the hospital departments, the health-care professionals expressed tension for change. The practice deviated from the guidelines for malnutrition in several areas. The MyFood system was perceived as more precise, trustworthy, motivational and fun to use compared with current practice. The use of MyFood was perceived to lead to earlier implementation of nutritional treatment and some thought it would be a time-saver. Potential barriers to the use of MyFood in clinical practice were patients from other cultural backgrounds eating types of food other than the hospital food, patients not speaking Norwegian, hygienic aspects over the use of tablet computers, concerns about the time used to follow up the system and the lack of automatic data transfer to the electronic patient record. In the RCT, the patients allocated to the MyFood group did not differ with regard to change in body weight or body composition during their hospital stay when compared to the control group. Nutritional treatment was documented in the electronic patient records for 81% of the patients in the MyFood group and 57% in the control group (P = 0.019). In the MyFood group, 70% of the patients received a nutrition care plan, whereas the corresponding proportion in the control group was 16% (P = 0.011). Documentation of nutritional intake compared with patient requirements for energy, protein and liquids was present for 84% of the patients in the MyFood group and 4% in the control group (P