Older people represent a growing population in Western countries due to sociodemographic developments. As a consequence, the number of people who have therapeutic or nursing needs is also increasing sharply [1]. Therefore, geriatric rehabilitation is becoming increasingly important in the medical care of older people. Patients undergoing geriatric rehabilitation are characterized by a reduced functional status due to disabling impairments. These are often combined with the presence of multimorbidity and geriatric syndromes such as sarcopenia or frailty [2-4]. The aim of geriatric rehabilitation is to improve functional resources in order to enable patients to return to their home environments [5].

Nutritional status is a well-known factor that is interrelated with physical function, frailty, and sarcopenia, which are associated with negative effects on functional dependence in patients undergoing geriatric rehabilitation [6-8]. However, malnutrition and the risk of malnutrition are widespread conditions among these patients, with a prevalence of 5% to 20% for malnutrition and 40% to 54% for the risk of malnutrition [8]. Because there is profound evidence, malnutrition is treated as part of the structured rehabilitation program within nutritional therapy [9]. The focus here is on optimizing energy and protein intake and on weight stabilization to generate a positive effect on muscle building and muscle maintenance [10]. The amount of energy and protein intake consumed is determined by the nutritional therapist using the method of a food record. Individual nutritional recommendations for the patient are then derived and communicated on this basis. Due to financial and personnel resources, nutritional therapy is often limited in time during rehabilitation treatment, even if the literature indicates that close-meshed nutritional therapy might achieve greater success [11].

eHealth applications, such as computer-based food records, can be used to provide support by intensifying the nutrition therapy process. The time-consuming processes of recording and analyzing dietary intake data are no longer necessary for the nutrition therapist. Maintaining a computer-based food record regularly enables the diet to be monitored more consistently and proactively. Deviations from the nutrition target can be recognized earlier, and reintervention can be initiated by the nutrition therapist. In addition, maintaining the computer-based food record can increase the patient’s awareness of their diet and therefore might lead to higher motivation in achieving nutritional goals. In the context of geriatric rehabilitation, advanced age, limited experience with technology, reduced visual acuity, changes in haptics, cognitive limitations, and fears in dealing with technology are characteristic challenges when using eHealth systems in this target group that must be addressed in the technical development process.

eHealth interventions in geriatric rehabilitation have already been shown to have the potential to improve rehabilitation outcomes. In a meta-analysis, the evidence for feasibility, usability, and effectiveness in the treatment of patients undergoing geriatric rehabilitation was demonstrated. However, the eHealth interventions focused predominantly on activity and physical function. No nutritional interventions were reported [12]. Only Happe et al [13,14] demonstrated in 2022 and 2023 the feasibility and usability of a tablet-based e-coach to improve nutrition and physical activity in patients undergoing geriatric rehabilitation. As far as we know, there is no research data on e-food record methods in the group of patients undergoing geriatric rehabilitation published to date.

Therefore, the primary objective of this pilot feasibility study was to verify whether it is fundamentally possible for patients undergoing geriatric rehabilitation with low technical experience and simultaneous high postacute or chronic stress to independently use a tablet computer−based food record (e-food record) over 3 days that has been specially adapted to their needs. Secondary and exploratory objectives included the assessment of usability as well as the analysis of the recorded dietary data. In addition, the recorded food items, food portions, and nutritional intake data of the e-food record are compared to the recorded data of a 24-hour dietary recall, a commonly used reference method in clinical practice. Finally, the study aimed to identify potential areas of improvement to inform further development of the e-food record.

Patients aged 70 years and older who were consecutively admitted to a geriatric rehabilitation ward of a rehabilitation center in the northwest of Germany were eligible for the present prospective pilot feasibility study. Exclusion criteria were (1) immobility in the sense of confinement to bed, (2) the inability to understand the study content and the course of the study, (3) difficulties in communication in the German language, and (4) participation in another study. Inpatient geriatric rehabilitation in Germany is a structured multidisciplinary program with the goal of regaining independence and functionality in older persons with reduced functional conditions due to acute or chronic medical conditions [15]. Daily therapy is administered for a period of usually 20 days [16], including medical therapy, nursing care, physiotherapy, occupational therapy, neuropsychology and social services, nutritional therapy, and speech therapy [15]. Patients undergoing geriatric rehabilitation represent a highly vulnerable and clinically burdened population, often characterized by multimorbidity, functional limitations, and increased physical and psychological stress following acute or chronic health events. Furthermore, familiarity with and affinity for digital technologies are often limited in this age group. Data for the feasibility study were collected between July and December 2018. All participants took part in the study during the geriatric rehabilitation program. Sex and age were requested from the patients and documented. Due to ethical and data protection regulations, no data were collected from patients who declined participation. Therefore, no systematic information on reasons for nonparticipation is available.

The study was approved by the medical ethics committee of the Carl von Ossietzky Universität Oldenburg (ethical approval code 2018-029). We conducted the study in accordance with the Declaration of Helsinki. The study was a priori registered in the German Clinical Trials Register (DRKS-ID: DRKS 00014370).

Informed consent was obtained from all participants prior to participation. Pseudonymized data were collected and stored on secure, password-protected systems accessible only to the research team. Results are reported in aggregated or anonymized form. Participants did not receive any financial compensation, incentives, or other benefits for their participation in this study.

The statistical analysis was conducted using SPSS Statistics version 29 (IBM SPSS Inc). Descriptive statistics are presented as mean values and SD, median and IQR, or as total numbers and percentages. Due to the pilot characteristics of the study, the statistical analysis refers to an explorative analysis, and no interference statistics were performed.

Twenty-five out of 224 (11%) patients who were eligible for inclusion were enrolled in the study. Seventeen out of 25 (68 %) patients completed the study and were included in the analysis of usability and technology commitment, while 8 patients dropped out of the study prematurely. Detailed information about the reasons for dropout is shown in Figure 2. During the study, a technical malfunction occurred in which the day selection did not automatically jump to the current day, contrary to what the study participants were told. As a result, the entries could no longer be validly assigned to the individual protocol days. Due to this, 9 out of 17 datasets were included in the analysis of the nutritional data (group A), while 8 datasets could not be analyzed (group B).

The mean age of all participants (n=17) who completed the study was 79.5 (SD 3.7) years. Six (35.3%) of them were female. The mean age of the participants who completed the 3-day application test (group A, n=9) was 78.2 (SD 2.9) years. Five out of 9 (55.6%) participants were female. The average age of the participants whose data were included for the analysis of the food records (group B, n=8) was 81.0 (SD 4.0) years, and 1 (12.5%) participant was female.

The SUS was rated 76.0 (SD 11.3) points on average by all participants. It was higher in participants of group A with a mean score of 79.4 (SD 10.0) points compared to 72.2 (SD 11.4) points in participants of group B.

The mean technology commitment of all participants was 42.4 (SD 6.8) points. For group A, the mean technology commitment was 42.9 (SD 7.3) points compared to group B with 41.9 (SD 6.1) points. The single-item responses of the Likert scale (1‐5 points) were rated on average with 3.5 (SD 1.8) points (n=17). Detailed information on the values of the SUS and technology commitment per participant are represented in Figure 3.

This pilot study primarily investigated whether an e-food record is a feasible method for documenting nutritional data in older adults undergoing geriatric rehabilitation. The results of this study demonstrate that the e-food record was successfully used by older adults in geriatric rehabilitation to record their food and beverage intake over a period of 3 days. Secondary and exploratory findings are related to usability and the comparison with a 24-hour recall. The e-food record’s usability score (mean 76.0, SD 11.3 points) indicates good acceptability and ease of use. However, the e-food record and the 24-hour recall determined different results in terms of recorded nutrient intake and food portion sizes, indicating possible limitations. These differences should be interpreted with caution, as both methods are subject to measurement errors. While the 24-hour recall may be affected by recall bias and memory limitations, the e-food record may be influenced by input errors or difficulties in the estimation of portion size. These findings suggest that the e-food record is a promising tool for recording food and beverage intake in older adults, but it currently also has some limitations.

With respect to the primary objective of feasibility, this study demonstrated that the target group of patients undergoing geriatric rehabilitation was able to use an e-food record independently during their rehabilitation treatment. A systematic review by Kraaijkamp et al [12] showed that eHealth is feasible, usable, and effective in geriatric rehabilitation. However, none of the 40 included studies were related to nutritional content. A nutrition and exercise e-coach was developed iteratively by Happe et al [13,25] and tested successfully in a piloted application study with a follow-up of 9 weeks among older people during and after geriatric rehabilitation [14]. However, this study was designed for usability testing and user behavior analysis, and the data of the e-food record were not analyzed. A study on community-dwelling older adults with a mean age of 72.2 (range 65-89 y) demonstrated that an e-food record is comparable to a paper-based food record method [26]. During the study, 8 out of 25 participants withdrew prematurely from the study. These results are comparable to the study of Happe et al [14] mentioned above. Here, 8 out of 30 participants dropped out of the study prematurely. In this study, we observed a large difference between the number of patients who were in principle eligible for participation and the number of patients who actually participated (25 out of 224 patients undergoing geriatric rehabilitation).

In contrast, findings related to usability and data comparability should be interpreted as secondary and exploratory, as these aspects were not the primary focus of the study. The usability was rated with a mean score of 76.0 (SD 11.3) points, which reflects a high usability of the e-food record. A score of 82 points is described with the adjective “good” in the evaluation by Bangor et al [23] and is near to the score of 85 from which excellent usability begins. The SUS score is slightly lower than that for a tablet computer−based e-coach to improve nutrition and physical activity in patients undergoing geriatric rehabilitation developed by Happe et al [14] with 78.6 (SD 18.1) points on average. This study was conducted as a 3-day application test that did not contain any increased benefits for the participants (eg, feedback on dietary behavior or dietary recommendations). A higher benefit for the participants and thus a higher willingness to participate could be achieved in future studies by providing feedback on nutritional intake, integrating the e-food record into the nutrition therapy, or extending the study period beyond the rehabilitation program.

The demonstrated feasibility of independent use of the e-food record provides an important basis for future research in the field of nutritional assessment in geriatric rehabilitation. The e-food record shows potential as a tool for documenting nutritional data in geriatric rehabilitation, which should be emphasized in future research. In order to exploit the full potential of the e-food record in the future, it should be integrated into daily treatment, ideally as part of a randomized and controlled intervention study. A comparison between the e-food record and the recorded nutritional values should be conducted using established reference methods and a larger number of participants to further investigate the validity and reliability of the e-food record in adequately powered studies. In addition, subsequent studies should investigate which factors contribute to the successful use of the e-food record and which barriers exist to use an e-food record among patients undergoing geriatric rehabilitation. Relevant aspects for further development and optimization of the e-food record were identified in this study by comparing the e-food record with the practical method of the 24-hour recall. The data in Figure 8 indicate that the minority of food and beverage entries (<30 % in 13 out of 15 food categories) are consistent between the e-food record and the 24-hour recall. To estimate portions more precisely, images of food items with the corresponding portions could be displayed on the user interface of the e-food record. In literature, there are also attempts to estimate portions using photo documentation and image recognition [27-29]. Furthermore, the data of Figure 8 indicate a relatively large discrepancy between the foods recorded via the e-food record and the 24-hour recall. A confirmation prompt could be added to the end of a meal entry, for example, to detect incorrect entries by the patients and give them the opportunity to correct them. The results represented in Multimedia Appendix 2 indicate that certain foods such as condensed milk, water, black tea, coffee, cookies, and gravy were partly reported only via the 24-hour recall. This might be due to the fact that the 24-hour recall was queried using a predefined list of foods in food or meal groups that possibly counteracted the forgetting of foods and beverages consumed. A structured query of frequently forgotten or often related foods or beverages such as coffee and condensed milk, bread and spreadable fat, or salad and dressing could prevent forgetting of foods and beverages. The results of Multimedia Appendix 3 indicate that participants may have rated the fat content of foods such as yogurt milk drinks, curd cheese, cold cuts, and fish differently in the e-food record and 24-hour recall. A barcode scanner function or letter recognition of the packaging may increase input accuracy. Within the use of the e-food record, it was possible to select both individual foods and entire dishes. Entire dishes could be omitted. They could be replaced by individual foods. If entire dishes are used as an input option, ingredient queries may increase accuracy of, for example, a stew. However, it is important to balance the benefits of additional functionality against the potential increase in complexity to ensure ease of use for patients undergoing geriatric rehabilitation. The exploratory findings of this study, particularly regarding usability and discrepancies with the 24-hour recall, provide important directions for further development and hypothesis generation.

This study was aimed at serving as a pilot feasibility test and was, therefore, designed as a short-term observation with only a limited number of participants. This could limit the generalizability of the results. In addition, the study was conducted in a single rehabilitation center, which may limit the transferability of the results to other settings or patient populations.

Only a small number of eligible patients participated in the study, and even a smaller number contributed analyzable dietary data. This may introduce substantial selection bias and limited generalizability, as participants who agreed to participate may differ systematically from nonparticipants, for example, in terms of motivation, health status, or technology affinity. Due to ethical constraints, no data on nonparticipation were collected, although the characteristics of this highly burdened population undergoing geriatric rehabilitation likely contributed to the low participation rate. As a result, the findings are based on a highly selected subgroup, and their generalizability to the broader population of patients undergoing geriatric rehabilitation is strongly limited.

A technical malfunction occurred during the study, as the automatic navigation to the current day’s page did not function as intended. This may have affected user interactions and data quality. Feasibility was operationalized primarily based on the dropout rate. However, more detailed indicators of feasibility, such as the level of assistance required during use or the frequency of user errors, were not systematically recorded. This limits a more comprehensive understanding of independent use.

Furthermore, the cognitive status was not systematically recorded within the study. Cognitive impairment may have influenced both the ability to use the e-food record and the accuracy of the 24-hour recall, for example, due to recall bias or difficulties in reporting. However, patients with inability to understand the study content and procedures were excluded from participation, which may partially reduce the impact of severe cognitive impairment in the study sample.

Differences observed between the e-food record and the 24-hour recall could be interpreted with caution, as both methods are subject to different sources of measurement error. The comparison between the two methods was exploratory and not intended as a validation study. Given the small sample size and exploratory design, no firm conclusions regarding the validity or accuracy of either method can be drawn. In addition, analyses related to usability and comparison with the 24-hour recall were exploratory and not powered for definitive conclusions. These results should therefore be interpreted with caution and require confirmation in larger studies.

Effective nutrition therapy is crucial for achieving an optimal nutritional status in patients undergoing geriatric rehabilitation with a reduced nutritional status, as malnutrition is directly linked to physical functionality and rehabilitation outcomes [8]. The e-food record method offers several advantages, including direct access to nutritional information for nutrition therapists. This contributes to the implementation of early and targeted interventions. Traditional food records require manual analysis by nutrition therapists and are often limited to 1-day recalls. With the e-food record method, the patients record their nutritional intake themselves. This reduces personnel resources, which in turn can be used for personalized advice.

By utilizing consumption data, the e-food record can improve nutritional status by providing more intensive and targeted therapy. To date, the detailed data on food and beverage intake over several days are rarely available in clinical practice, although it can enhance counseling. Additionally, nutrition therapists can use the e-food record to monitor the progress of nutrition therapy and introduce new interventions when necessary. This was previously more difficult due to time and financial constraints. This can lead to improved patient outcomes, such as enhanced physical function, reduced hospitalization rates, and better quality of life. Furthermore, an e-food record can facilitate communication between health care professionals, patients, and family members, promoting a more collaborative approach to nutrition therapy. In addition, the use of the e-food record is transferable to patient groups with similar requirements for the use of technology, such as patients in acute geriatric, older patients in neurology, or older people in short-term care facilities. However, given that findings beyond feasibility are exploratory, further validation is required before routine implementation in clinical practice.

This pilot feasibility study of an e-food record in the population of inpatients undergoing geriatric rehabilitation is, to our knowledge, the first of its kind and demonstrated, in general, that patients undergoing geriatric rehabilitation were able to independently record their food and beverage intake using the e-food record over 3 days. The results suggest that app-based food recording can be an integral part of nutrition therapy in the future to improve the medical care of patients undergoing geriatric rehabilitation. However, the findings of this study show that the e-food record and the 24-hour recall provide different results in terms of nutrient intake and food portion sizes. Further research with studies involving larger numbers of participants is needed to draw conclusions regarding data accuracy or validity of the e-food record.