Study design

The HOMEFOOD study was a 6-month, randomized controlled, assessor blinded intervention trial conducted in older adults (age 66–95 years) recruited in the Reykjavik capital area, Iceland between January 2019, and July 2020. The primary aim was to investigate the effects of intense nutritional therapy, including free access to energy- and protein-dense foods delivered to subjects recently discharged from hospital. The primary outcomes of this trial were changes in body weight and physical function (Short Physical Performance Battery (SPPB)). Body weight loss and poor physical function are both important predictors of negative health outcomes in older adults [2, 17]. These two variables were chosen to be the primary outcomes, as a nutrition intervention with focus on increasing energy- and protein intake is likely to affect body weight and physical function [4, 18], considering the low energy intake previously reported in elderly discharged patients [19]. Secondary outcomes included other anthropometric measurements, nutritional status, muscular strength, dietary intake, exercise, and reported food-related digestion issues, such as diarrhoea, nausea, constipation, or stomach pain.

Reporting, approval, and funding

This study was conducted and reported according to the Consolidated Standards of Reporting Trials guidelines for Randomized Trials of Nonpharmacologic Treatments (CONSORT) [20]. The study was approved by the Ethics Committee for Health Research of the National University Hospital of Iceland and data protection registry (24/2018) in August 2018 and performed in accordance with the Declaration of Helsinki [21]. The study was registered at ClinicalTrials.gov (NCT03995303).

Screening and recruitment

Potential participants (N = 1003) were screened by a clinical nutritionist in collaboration with attending nurses at the Landspitalinn University Hospital of Iceland. Eligible patients were discharging home to independent living from the hospital, aged 65 years or older, and assessed as being at risk for malnutrition (score ≥ 3) according to the validated Icelandic Nutrition Screening Tool [22], and had given their written informed consent. Excluded were those with known dietary allergies/being on a special diet, severe chronic kidney disease (glomerular filtration rate < 30 mL/min/1.73 m2), in active cancer treatment, receiving tubal feeding, not being able to communicate with the research team, cognitive function ≤20 according to the Mini Mental State Examination (MMSE) [23], and not having access to a functioning kitchen at home (i.e., refrigerator, oven, or microwave oven). Of the 1003 screened potential participants, n = 897 were ineligible for participation in the study. They were ineligible as they were too sick to participate, had been discharged, deceased, scored <20 on the MMSE, had been admitted to a nursing home, were not community dwelling, relying on tubal feeding, were <65 years of age, were not living in the capital area, or had declined participation (Fig. 1).

Fig. 1: Flow chart.

Flow chart of assessment, recruitment, allocation, follow up, and analysis process.

Randomization

The participants were randomly allocated to either the intervention or the control group by using a random number generated as implemented by the Statistical Package for the Social Sciences (SPSS, version 26.0, SPSS, Chicago, IL, USA).

Intervention group

The participant received nutrition therapy from the clinical nutritionist consisting of five home visits (1 day after discharge and one-, three-, six- and twelve weeks later) and three telephone calls in between the home visits. The nutrition therapy was implemented following the principles of Nutrition Care Process, which entails the following: nutritional assessment, diagnosis, intervention, monitoring, and evaluation of the nutrition therapy [16]. During the dietary counselling sessions, family members, relatives, friends, or home-care workers were invited to join as well. At the initial visit after discharge, the participant was educated about the importance of adequate energy and protein intake [24]. Nutrition-related problems were identified during the interviews, and suggestions given to resolve them. In addition to the dietetic counselling, participants received free supplemental energy- and protein-rich foods (1 hot meal/day and 2 in-between-meals/day; Supplementary Table 1) delivered once a week. During the first home delivery, study staff educated the participants on how to store the meals, how to open the packages and how to heat the meals.

Control group

At discharge, the control group received a booklet on good nutrition during aging published by The Icelandic Medical Directorate [24] and were encouraged to order MOW without any further dietary counselling during the study period, reflecting current standard care in Iceland for older adults discharged from hospital.

Participant characteristics

Background variables, e.g., age, sex, education, living arrangements, alcohol use and smoking habits, were assessed using questionnaires. Additional variables were collected from the Icelandic electronic hospital registry SAGA (TM software 3.1.39.9), e.g., height, number of diagnoses according to the International Statistical Classification of Diseases and Related Health Problems 10th Revision (ICD-10), and number of different medications.

Outcomes assessed

All primary and secondary outcome measurements were conducted at baseline (at the hospital) and at endpoint (at the participants’ homes). These measurements were conducted in a predefined order and questions on food or diet were asked only at the very end of each assessment. As the outcome assessors (who did not deliver the intervention) were unaware whether a participant was in the control or intervention group, measurements of anthropometrics, physical function, muscular strength, and nutritional status were blinded.

Anthropometrics

Body weight was measured in light underwear/clothing on a calibrated scale (model no. 708, Seca, Hamburg, Germany) and height was taken from the Icelandic electronic hospital registry SAGA (TM software 3.1.39.9). Body mass index (BMI) was calculated from the height and weight (kg/m2). Participants were categorized into three BMI categories: low BMI < 23 kg/m2, middle BMI 23–30 kg/m2, or high BMI ≥ 30 kg/m2 [24]. Body composition was measured using a hand-held bioelectrical impedance analysis device (BIA, Omron HBF-306C, Kyoto, Japan) [25]. Calf circumference was measured in a seated position. The tape was wrapped around the right calf and moved up and down to locate the maximum circumference in a plane perpendicular to the long axis of the calf [26]. Midarm circumference was also measured in a seated position and was taken on the left upper arm, at the mid-point between the tip of the shoulder and the tip of the elbow (olecranon process and the acromion) [27].

Physical function

Physical function was assessed using the SPPB, which evaluates lower extremity function assessing (1) usual-paced gait speed over a four-meter-course, (2) standing balance, and (3) time to rise from a chair five times. For each test, a score of 0 to 4 is assigned using cut points [28]. The three test scores are summed, yielding a range from 0 to 12. As SPPB testing in this study was performed at the participants’ homes, it was shortened for practical reasons, and thus did not include the gait speed part. The possible score therefore ranged from 0 to 8. Additionally, participants were asked the question “Do you have difficulties walking?” (Yes vs. no).

Muscle strength

Handgrip strength was measured in a seated position with a hydraulic hand dynamometer (Baseline® Baseline Evaluations Corporation) set on position two and the maximal grip strength of two trials was registered as the subject’s grip force in kilograms using their dominant hand [27].

Dietary intake

Dietary intake was assessed using a 24-hour-dietary-recall interview (24-HR) to obtain estimates of intakes of fluids, energy, and energy-giving nutrients [29,30,31,32,33,34]. The results from the 24-HR were entered into the nutrition calculation program ICEFOOD originally developed for the National Survey of Icelandic Diet 2002 and continuously updated for consequent National Surveys of Icelandic Diet (2011 and currently ongoing) [35]. ICEFOOD relies on the Icelandic database of the chemical composition of food (ISGEM within the Icelandic Medical Directorate of Health) and on a database within the Medical Directorate containing information on several hundred recipes of common dishes and ready-to-eat meals on the Icelandic market [35, 36]. Additional food-related questions and frequency of intakes of hot meals, major food groups, and liquids were assessed at endpoint using a simple food frequency questionnaire [37].

Nutritional status and food-related adverse events

Nutritional status was assessed using the Icelandic Nutrition Screening Tool as recommended by the Icelandic Medical Directorate of Health [24]. This validated questionnaire [22] consists of 13 questions which are scored and summed, yielding a range from 0 to 30. A clinical nutritionist also assessed whether any food- related digestion issues, such as diarrhoea, nausea, constipation, or stomach pain, were experienced during the intervention.

Sample size considerations

Sample size calculations based on our previous studies on body weight change [19, 38] suggest that the number of participants n = 44 in each group was estimated to be sufficient to detect a body weight difference of 1.8 ± 3.0 kg between groups as significant. The corresponding numbers for SPPB were n = 45 in each group, detecting a significance by 1 as significant (assuming SD = 1.7) [39]. The recruitment of >50 participants in each group allowed around 10% drop out to still retain sufficient statistical power.

Statistical analysis

Data were analysed using statistical software (SPSS, version 26.0, SPSS, Chicago, IL, USA). Data were checked for normality using the Kolmogorov–Smirnov test. Data are presented as mean ± standard deviation (SD). Differences between groups at baseline were calculated using independent samples’ t-test (normally distributed variables) or Mann–Whitney U-test (not normally distributed variables). We used intention-to-treat analysis.

Despite randomization, sex distribution was slightly uneven between treatment and control groups. As a result, we corrected for sex in all multivariate statistical endpoint analyses. Unadjusted analyses are also presented for comparison as supplemental material (Supplementary Table 2). Differences in anthropometrics and physical outcomes (continuous variables) between the groups at endpoint were assessed using linear mixed models in SPSS. Results are shown as parameter estimates, in which B describes the estimated and adjusted differences in the outcome variables between groups.

Differences in the abilities to perform physical tasks (single items from SPPB and “Do you have difficulties walking?” all categorical variables, yes vs. no) between the groups at endpoint were assessed using a logistic regression model, in which we corrected for the corresponding baseline values and sex.

Subgroup analysis was performed by comparing body weight changes between intervention and control in subgroups of males vs. females, married/cohabitating vs. single/divorced/alone, and low BMI group vs. middle BMI group vs. high BMI group. The effects of the intervention within subgroups were investigated using an independent samples’ t-test (for two variable subgroups) or ANOVA including LSD post hoc test (for three variable subgroups). We tested for interaction between subgroups and intervention using a general linear model.

Endpoint calculations represent per-protocol analysis with those dropping out of the intervention included in the baseline, but not in endpoint assessment. The level of significance was set at P < 0.05.