We conducted a single-arm feasibility trial of STRONG, a 12-week digital intervention to improve postoperative nutrition. Guided by the Obesity-Related Behavioral Intervention Trials model, the goal of the single-arm study was to identify potential technical issues with digital health delivery, assess the optimal length of intervention delivery, and gather participant feedback on acceptability to inform intervention refinement prior to larger testing in a randomized trial [23]. The intervention was developed based on the Theoretical Domains Framework, a theory used to understand and address multilevel behavior change (ie, patient and clinician behavior) in health care settings [24]. Participants received biweekly MNT (6 sessions) that included nutritional counseling with a registered dietitian and continuous remote monitoring of participants’ dietary needs by the dietitian. In addition, participants logged daily food intake using a Fitbit device (Inspire 2) and completed 5 study assessments related to patient malnutrition, nutrition-related symptoms, and quality-of-life outcomes (at baseline and 4, 8, 12, and 16 weeks after study enrollment). Participants provided feedback on the intervention’s acceptability and usability (at week 12).

Individuals who met the following criteria were eligible to participate in the study: (1) older than 18 years, (2) diagnosed with primary GI cancer, (3) diagnosed with PD, (4) underwent curative-intent CRS-HIPEC at Moffitt Cancer Center (Moffitt; with cytoreduction completeness score of 0‐1), (5) transitioned to a postoperative oral diet, (6) were able to speak and read English, and (7) provided informed consent. Individuals were excluded from the study if they met any of the following criteria: (1) had documented or observable psychiatric or neurological condition that would inhibit with study participation, (2) were undergoing treatment for another primary cancer, and (3) received postoperative parenteral or enteral nutrition.

Potential participants were identified through a collaboration between Moffitt’s GI clinic staff and the study coordinators. In addition, we screened the patients’ electronic health records (EHRs) to determine their eligibility. Eligible participants were contacted by phone, unencrypted email, videoconference, or in-person meetings to introduce them to the study and determine their interest in participating. The participants who provided informed consent were given study materials and equipment during patient visits or by mail, including (1) a welcome packet and checklist describing the study components, instructions on using and caring for the Fitbit tracker and tablet, and brief instructions on estimating food portion size; (2) a Fitbit tracker; and (3) a study loaned tablet to log daily food intake with the Fitbit application already downloaded and synced. Participants also had the choice of downloading the Fitbit application on a personal device if preferred. Within 3 to 5 days of the participant receiving the Fitbit tracker, one of the study coordinators (RH, OS, and SD-C) contacted the participant to confirm that they were able to use the device. Participants received an introduction to the Fitbit tracker before undergoing CRS-HIPEC. Recruitment occurred from December 2022 to July 2023.

Dietitians reviewed the participants’ food intake and nutritional assessments and conducted 6 biweekly telehealth or in-person counseling sessions with them. During these visits, the dietitian established individualized dietary plans that included a calorie goal, discussed challenges to dietary intake, and made recommendations for improving nutrition. If a participant did not record food intake for 5 days or more, a study coordinator contacted the participant to discuss barriers to using the Fitbit tracker and to encourage continued tracking. Study assessments were completed on REDCap (Research Electronic Data Capture; Vanderbilt University), a web-based software platform [25,26], on a paper survey, or in person using a tablet during clinic visits.

Descriptive statistics were computed to describe the study sample and assess whether prespecified benchmarks for feasibility, acceptability, and usability were met at the end of the intervention. Given the small sample size, continuous variables were summarized using median and IQR, and categorical variables were summarized using frequency and percentage. Changes in patient outcomes at baseline and the end of the study period (week 16) were assessed using linear regressions for continuous outcomes and ordered logistic regressions for ordinal outcomes. Models included participant fixed effects to obtain within-participant estimates, and SEs were robust and clustered by the participant. We adhered to the CONSORT (Consolidated Standards of Reporting Trials) guidelines for pilot and feasibility studies and study reporting (Checklist 1) [42]. All analyses were performed in Stata (version 18; StataCorp).

The trial (ClinicalTrials.gov NCT05649969) was conducted at a single site, Moffitt, a National Cancer Institute (NCI)–designated comprehensive cancer center. The study was approved by Moffitt’s Institutional Review Board of Record, Advarra (protocol Pro00066098). Informed consent was obtained from all participants. To protect participants’ confidentiality, deidentified information and pseudonym IDs (eg, Participant 1) were entered into participants’ Fitbit profiles. The study staff maintained a separate password-protected file behind Moffitt’s firewall linking participant IDs to patient identifiers (eg, name and medical record number). Paper questionnaires were stored in a locked file cabinet in an office with a locked door. Only the study team had access to participant research data, and only trained staff with appropriate approvals had access to patient medical records. Participants were compensated with a US $25 gift card for completing each of the 5 study assessments; participants who completed all 5 assessments received an additional US $25 gift card.

Among the patients (n=42) screened for eligibility, 25 patients were deemed ineligible (Figure 1). Of the 17 patients approached, 10 patients consented to participate in the study. All 10 participants completed at least 1 assessment, and no participants were lost to follow-up.

Study participants had a median age of 57.5 (IQR 54.0‐69.0) years; 8 (80%) participants were female, and 10 (100%) were non-Hispanic White (Table 1). In total, 6 (60%) participants were married, 1 (10%) had never been married, and 3 (30%) were divorced. All participants spoke English as their primary language and resided in an urban area. The median ADI of participants was in the 30th percentile nationally (IQR 24-32), with a higher ADI indicating a more disadvantaged neighborhood. In total, 4 (40%) participants received health insurance through employment, 4 (40%) received public insurance, 1 (10%) had direct-purchase insurance, and 1 (10%) did not provide insurance-type information. The highest education attained by participants was some college for 3 (30%) participants, a bachelor’s degree for 3 (30%) participants, or a graduate or professional degree for 4 (40%) participants. Participants’ annual household income ranged from US $35,000 to ≥US $100,000.

In total, 8 (80%) participants never smoked tobacco, and 2 (20%) were former smokers (Table 1). The median BMI of participants was 28 (IQR 27.3-29.5), and participants had a median Charlson Comorbidity Index of 7 (IQR 7-9). A total of 6 (60%) participants were diagnosed with appendiceal cancer, 2 (20%) with colorectal cancer, and 2 (20%) with peritoneal mesothelioma. The median peritoneal cancer index was 21.5 (IQR 13.0-25.5). Participants were all fully active or able to carry out light or sedentary work, with a median Eastern Cooperative Oncology Group performance status of 0 (IQR 0-0). No participants at baseline were assessed to be at low malnutrition risk, 2 (20%) were at medium risk, and 7 (70%) were at high risk.

Among the eligible patients (n=17) who were approached to participate in the study, 10 (59%) consented to participate (Table 2). In total, 9 (90%) participants completed the baseline assessment and completed 4 of 5 assessments. All participants completed the assessment at the end of the intervention (week 12), and 9 (90%) participants completed the follow-up assessment at week 16. A total of 7 (70%) participants attended at least 4 of 6 dietitian appointments, and 6 (60%) participants logged food intake for at least 63 of the 90 days (median logged 76, IQR 15‐87 days). Adherence to logging food intake decreased slightly over the span of the intervention, from 7 (70%) participants meeting the benchmark number of days logging food intake in the first 30 days (median logged 29, IQR 15‐30 days) to 5 (50%) participants meeting the benchmark in the last 30 days (median logged 15, IQR 0‐27 days; not shown in table).

Among the 10 participants who completed the week 12 assessment, 8 (80%) rated the intervention as acceptable (benchmark score >12), with a median score of 18 (IQR 16‐20; Table 3). In total, 5 (50%) participants rated the Fitbit tracker and application as usable for logging food intake (benchmark score >68), with a median score of 68.8 (IQR 54.4‐90). All participants (100%) were satisfied with the dietitian services (benchmark score >12), with a median score of 23.5 (IQR 17.8‐24.0). One participant reflected on the high acceptability of the dietitian services and said, “The program added a lot of value. It helped with my recovery, especially with getting the right nutrition. It has been really great!” Another participant conveyed the value of the digital nature of the STRONG intervention, expressing “The ZOOM meetings were wonderful because I am over an hour from the hospital. The [nutritionist] was also a great encourager and contributed to my healing process.”

Compared to baseline, average PG-SGA malnutrition scores saw a decrease of 6 points (P=.01), with a corresponding reduction in patients with high malnutrition risk (P=.03; Table 4). Functional Assessment of Cancer Therapy—General scores increased by an average of 20.5 points (P=.002), and Functional Assessment of Anorexia/Cachexia Treatment—Anorexia/Cachexia Scale scores increased by an average of 7.4 points (P=.03), indicating an improvement in participants’ health-related quality of life. There was no change in participants’ average BMI or weight, suggesting a stabilization of weight loss. This is supported by a 5.6 percentage point increase in the average 1-month weight change (P=.04), in which participants saw a slight weight gain compared to the previous month at week 16.

The goal of this trial was to evaluate the feasibility and acceptability of the STRONG intervention for individuals with GI cancer and PD undergoing CRS-HIPEC. Our findings demonstrated that the STRONG intervention was feasible to be implemented with high participant recruitment, adherence, and retention to the intervention. Participants rated the intervention favorably and found the dietitian services to be both acceptable and usable. This rating is consistent with previous pilot studies assessing the implementation of mobile phone–based nutritional intervention for individuals with cancer [43,44]. Patient outcomes, including malnutrition risk, health-related quality of life, and 1-month weight change, saw marked improvement. Given that this study was a single-arm intervention without a comparison group, we were unable to attribute the changes in patient outcomes to the intervention alone without considering the effects of cancer treatment and disease progression. The successful implementation of STRONG in this study, positive feedback from participants, and promising improvements in patient outcomes suggest that a future, fully powered trial with a comparison group is warranted. Our team is currently evaluating potential improvements to STRONG and assessing alternative food logging approaches in preparation for a randomized controlled trial of STRONG.

Malnutrition screening, counseling, and related interventions remain underused in cancer care [45,46]. Nutritional counseling has been shown to improve the nutritional status, quality of life, and survival for individuals with GI cancer [47,48]. Digital nutritional interventions show clear benefits over traditional MNT, including efficiency, accessibility (eg, reduced transportation barriers), and the ability to remotely monitor patients outside of a traditional clinic visit [22]. Digital tools can also help individuals maintain adherence to nutritional interventions [14]. In our study, despite undergoing a complex surgical procedure, participants were able to adhere to the dietitian visits and food logging in the postoperative period. We hypothesize that this may be facilitated by strong support from the clinic team. Members of the surgical and dietitian teams encouraged patients to participate in the intervention and periodically checked on participants to monitor their progress through the intervention. This study is innovative in that individualized and remote monitoring of dietary needs bridges the gap between the clinical need for close, in-person patient follow-up and the substantial barriers for this patient population to access nutritional support.

Prior studies have shown that digital nutritional interventions are feasible and effective for achieving weight loss among survivors of cancer [49,50]. However, there has been limited research on the use of digital health interventions for individuals with malnutrition to maintain weight or to prevent weight loss. To the best of our knowledge, this is the first digital nutritional intervention conducted among individuals receiving CRS-HIPEC, who have increased risk for malnutrition and face unique barriers to accessing and using traditional MNT [7,20,21].

One study that assessed food-intake tracking with a Fitbit device among individuals with colorectal cancer undergoing surgery found decreased acceptability of the intervention in the postoperative period due to the complexity of the Fitbit application [51]. In our study, we also found declining adherence to tracking food intake over time. The benchmark for tracking food intake was not achieved and was driven primarily by 3 participants who logged 0, 3, and 12 days over the course of the intervention. The decline in adherence to tracking food intake was driven primarily by 2 participants who logged 12 and 24 days over the first 30 days of the intervention, followed by no additional days logged over the rest of the intervention period. It was unclear why these participants were disengaged with tracking food intake, as they did not provide any qualitative feedback.

Further research is needed to investigate why the usability of Fitbit may be low (eg, differences in digital literacy) and what strategies could be used to improve usability. One potential explanation is the choice of the usability measure. Our study team used a generic SUS that was not targeted to mobile health specifically [52]. In future studies, we plan to include mobile health–specific measures of usability to see how that may affect usability ratings. Another potential explanation is that study participants did not have sufficient education on how to use the Fitbit. Since this pilot study, our team has added teach-back sessions to the Fitbit training. Additionally, our team has created a paper version of the food log as an alternative for patients who cannot manage electronic food logging even after training. A third hypothesis is that participants may only need to food log for a certain amount of time before learning enough about their dietary patterns to manage their nutrition. Further study is needed on the optimal time needed for food logging for malnutrition self-management.

Future digital health interventions should assess the eHealth literacy of participants and make efforts to address participant concerns about the potential complexity of digital interventions [53,54]. One approach may be to target participants with low eHealth literacy with additional technical support and resources to ensure equal and inclusive participation in the intervention [53]. Another approach to improve adherence to food tracking may be to adopt a tracking system designed for patients with cancer (eg, the Automated Self-Administered 24-Hour Dietary Recall developed in collaboration with the NCI) [55]. Recent developments in image processing technology can also be leveraged to reduce the burden on participants in tracking food intake [56]. For example, artificial intelligence–enabled applications such as MyFitnessPal, Fastic, and Noom enable food tracking through capturing photos of the foods via a smartphone [56]. However, the feasibility of these technologies for individuals with cancer has yet to be fully explored.

Our study should be considered in light of the following limitations. First, this is a single-arm feasibility study conducted at an NCI-designated comprehensive cancer center. Therefore, the study findings may not be generalizable to other settings. Second, the study sample is small and consisted only of non-Hispanic White participants, limiting generalizability to other patient populations. There is a critical need to test the intervention in a more diverse population, across a wider range of settings, and on a larger sample. Additionally, our study focused on extending nutritional support in the postoperative setting, which has been previously recommended for individuals receiving CRS-HIPEC [57,58]. Further research is needed to test nutritional interventions for this population prior to surgery, which may improve CRS-HIPEC tolerance and reduce the likelihood of nutrition-related, postoperative complications [57,58]. Finally, this study was a single-arm trial without a comparison group. Findings on changes in patient outcomes over time cannot be solely attributed to the effect of the intervention without considering the effects of cancer treatment and disease progression. A larger, fully powered randomized controlled trial is needed to rigorously evaluate the impact of STRONG on patient outcomes.

Our study demonstrated that STRONG, a digital health intervention aimed at improving nutritional management for individuals with GI cancer and PD receiving CRS-HIPEC, is feasible, acceptable, and usable. Future studies are needed to establish the effectiveness of the STRONG intervention and to evaluate its implementation in more diverse patient populations and settings.