Introduction

Nutrition and physical activity guidelines proposed by the American Cancer Society recommend breast cancer survivors consume five or more servings of fruits and vegetables per day and spend at least 150 min per week in moderate-to-vigorous physical activity (MVPA)1,2,3. Breast cancer recurrence and survival are associated with higher levels of systemic inflammation, greater abdominal adiposity, and reduced physical function, which are all affected by nutrition and physical activity4,5. Despite the evidence demonstrating the benefits of these lifestyle behaviors on improving cancer outcomes, most breast cancer survivors do not meet the nutrition and physical activity recommendations6,7. Reasons for not meeting these guidelines include a lack of knowledge and/or social support, low self-efficacy, poor self-regulation, and structural barriers for sustainable change8,9,10,11.

Behavioral health interventions for cancer survivors with high frequency in-person contact are associated with large changes in behaviors, but this mode of delivery is costly and burdensome12. Interventions that do not involve face-to-face contact, such as those that are telephone-based or web-based, cost less and can reach more people12. Furthermore, evidence indicates technology-based interventions can be a useful and effective strategy for improving health behaviors13,14 and are acceptable to cancer survivors15. Technologies include eHealth (defined as electronic delivery of health care information) and mHealth, a subcomponent of eHealth involving mobile technologies, such as smartphones and wearable devices. Trials of technology-based health interventions for cancer survivors have not only shown improvements in behavior, but that these interventions are safe and feasible to implement13,14,16,17. However, data are limited on the optimal dose, efficacy, and acceptability of multiple delivery modes, and the long-term effectiveness of scalable technology-based interventions12.

To help address this gap, we developed and tested the feasibility and preliminary efficacy of the Cook and Move for Your Life study, which offered two scalable doses of a remotely delivered, eHealth behavioral nutrition and physical activity intervention for breast cancer survivors, with the goal of improving diet quality and increasing physical activity. As previously described18, the intervention was originally designed as an in-person intervention but due to the COVID-19 pandemic, was changed to a remotely delivered format. If proven effective, this intervention has potential to be cost-effective and to have high population-level reach to support breast cancer survivors in making positive changes to their diet and physical activity behaviors. The study’s primary objective evaluated the feasibility of delivering the online intervention. Secondary objectives compared the effects of two intervention doses on changes in daily intake of fruits and vegetables and minutes per week in MVPA. Exploratory outcomes examined changes in diet quality, weight and body mass index, different physical activity intensities, and levels of physical functioning and anxiety.

Results

Sample power and accrual

From December 2019 to January 2021, a total of 341 breast cancer patients were screened, of which 74 eligible women completed baseline data collection (Fig. 1)18. Using permuted block randomization, 38 were randomized to the high dose intervention and 36 were randomized to the low dose intervention. With a minimum adherence of 60% attending the online session, a power calculation indicated that 50 participants (25 per study arm) would provide 81% power to detect a difference of 0.8 standard deviation (SD) units between dose arms. Thus, our enrollment of 36 and 38 women across the study arms provided ample power to test our primary outcomes.

Fig. 1: CONSORT diagram for the Cook and Move for Your Life randomized controlled pilot study.
figure 1

Flow chart describing the number of women recruited, ineligible, randomized, and enrolled in the Cook and Move for Your Life study.

Full size image

Baseline characteristics of the study sample

Baseline characteristics are presented in Table 1. Women were a mean (SD) age of 58.4 (10.1) years and had a mean (SD) body mass index of 29.1 (7.3) kg/m2. Most women were White (87%), had at least a college degree (65%), were employed (69%), and married (72%). Approximately 90% of women reported being extremely or very comfortable with the internet and text messaging (Table 1). Most (73%) women reported having no major chronic conditions. At baseline, women were an average (SD) of 4.6 (2.3) years since their breast cancer diagnosis and had histories of breast surgery (95%), chemotherapy (50%), radiation therapy (77%), and endocrine therapy (73%). Characteristics at baseline were similar across the two dose interventions.

Table 1 Baseline characteristics of Cook and Move for Your Life sample by randomization dose arm
Full size table

Feasibility by intervention dose arm

The nutrition and physical activity group sessions were well attended by both intervention groups – 84% of high dose participants attended eight or more sessions, while 94% of low dose participants attended their single session (Table 2). Out of 23 possible text message responses requested, women in the low dose intervention were more engaged (mean (SD) number of texts: 18.2 (6.0) vs. 14.9 (6.4); P = 0.03). However, women in the high dose arm were more likely to have accessed the study website (84.2% vs. 66.6%, P = 0.08). Retention was high across the dose arms, with 92.1% and 97.1% of women in the high and low dose groups (P = 0.99), respectively, completing the 6-month questionnaires. Percent of women who returned the accelerometer for 6-month data collection was 84.2% and 72.2%, for high and low dose groups (P = 0.21), respectively. Women in the high dose group were more likely than women in the low dose group to report that the nutrition and physical activity group sessions were very or somewhat helpful (90.6% vs. 51.7%, P = 0.001). Although women in the low dose arm were more engaged with text messaging, women from both arms reported similar acceptability for helpfulness of texts (Table 2).

Table 2 Feasibility measures of Cook and Move for Your Life trial by randomization dose arm
Full size table

Diet and physical activity change by intervention dose arm

The mean (SD) fruit and vegetable intake at baseline for women in the high and low dose arms was 5.2 (2.0) and 4.5 (2.7) servings per day, respectively. Over the 6-month study period, relative to women in the low dose arm, women in the high dose arm consumed 1.5 (95% CI: 0.4, 2.6) more servings of fruits and vegetables per day. At baseline, the mean (SD) minutes per week spent in MVPA was 19.5 (18.7) for women in the high dose arm and 25.9 (25.4) for women in the low dose arm. At the 6-month follow-up, women in the high dose arm spent 1.7 (95% CI: −7.0, 10.5) more minutes per day in MVPA compared to women in the low dose arm, although this difference was not statistically significant (Table 3). With regards to other changes in dietary intake observed between arms, women in the high dose relative to the low dose arm, reported consuming 1.1 (95% CI: 0.3, 1.9) more servings of vegetables per day, 170 (95% CI: −321, −18) fewer kilocalories per day, and 11 (95% CI: −20, −2) fewer grams of total fat per day. No statistically significant differences between dose groups were observed for other measures of physical activity, body mass index, weight, physical functioning, or anxiety levels (Table 3).

Table 3 Adjusted mean difference in the outcomes associated with being in the high dose intervention arm relative to the low dose intervention arm
Full size table

Discussion

This randomized controlled pilot study evaluated the feasibility and preliminary efficacy of two doses of a remotely delivered eHealth intervention for women with a history of breast cancer. Our results indicated good adherence, retention, and acceptability for both the high and low intervention doses. However, women who received the high dose intervention showed more improvements in dietary intake, including increased daily servings of fruits and vegetables, and reductions in total caloric intake, total fat, and saturated fat. The intervention did not effectively improve physical activity levels. Remotely delivered behavioral interventions can be more cost-effective, less burdensome for both study staff and participants, and have the potential for high-population reach compared to in-person interventions. These pilot findings support the need to test the efficacy of a large-scale remotely delivered eHealth nutrition intervention for women with breast cancer.

The use of eHealth interventions to support improvements in dietary intake and physical activity of cancer survivors has become more frequent in recent years12,13,14,19,20,21. However, knowledge gaps still exist, such as determining the ideal exposure dose for long-term maintenance of behavior change. Since our intervention implemented two different doses, we offer some insight on this gap. For the high dose intervention, a maximum of 12 sessions were offered via two access points – in real-time or through recordings. Among the 38 women in this arm, eight accessed the sessions solely in real-time and two only viewed the recordings, indicating that most women participated in the sessions using both points of access. Indeed, no one participated in all 12 sessions using only one type of access point, but four did so using a combination. The data also indicated that several women watched some sessions repeatedly, sessions 10 through 12 were not viewed via recording by anyone in the high dose group, and no one from the low dose group viewed the archived recording of the single session. More investigation will be done to understand how these details affected retention, acceptability, and health behavior change. Ultimately, however, these findings suggest that ensuring the intervention is flexible so participants can engage on their own schedule improves adherence. Further, receiving only a single session (per the low dose intervention) was not a sufficient exposure dose – adherence was high, but acceptability of the sessions was low, and no diet or physical activity improvements were observed among women in this dose arm at follow-up.

In addition to the sessions, the eHealth intervention comprised various components delivered over the 6-month intervention period, including weekly motivational text messaging, biweekly electronic newsletters, physical activity self-monitoring using a study-provided Fitbit device, and access to a study website. Study participants from both dose intervention groups had high measures of adherence, study completion, and acceptability of these various study components. These results are consistent with several systematic reviews of eHealth diet and physical activity interventions in cancer survivors12,13,14,19,20,21. For example, a 2017 systematic review of 11 (six telephone-based, five eHealth-based) remotely delivered behavioral weight loss interventions for female cancer survivors reported 64–75% adherence to self-defined program goals and 70–100% rates of study completion13. A more recent (2022) review of 23 diet and physical activity eHealth interventions for cancer survivors reported eight studies using wearable devices with a median wear time of 87%. Additionally, data from four studies using text messaging responses showed a median text response rate of 73%21. Moreover, Wang et al. indicated that five studies reported the effectiveness of wearable devices in promoting physical activity engagement and four studies found text messages helped maintain motivation for physical activity21. Finally, a systematic review of 27 diet and physical activity eHealth intervention studies for cancer survivors reported that 22 of these studies had good retention, defined as at least 80% of assessments completed for interventions no more than 6-months in duration12.

Examination of secondary and exploratory dietary outcomes by intervention dose revealed improvements in dietary intake only among women who received the high dose intervention. However, we observed no change in physical activity or weight measures in either intervention dose. The existing literature indicates that the effectiveness of eHealth interventions to change diet and/or physical activity in cancer survivors is mixed, with some studies improving both behaviors, some affecting only one behavior, and some reporting no statistically significant changes in either diet or physical activity12,13,21. There is some consensus that eHealth interventions focusing on one specific behavior, such as fruit and vegetable intake, spending more minutes in MVPA, or weight loss may be more effective than interventions designed to target more than one behavior change simultaneously. Self-reported time spent in MVPA was about twice that of MVPA levels identified from accelerometer and prior studies have shown people overestimate their physical activity when it is self-reported compared to when it is measured objectively22, including in cancer survivors23,24. This overestimation might have affected the participant’s motivation and need to be more physically active. Of note, an inclusion criterion for study participation was self-reported daily consumption of 5 servings of fruits and vegetables or less or spending less than 150 min per week in MVPA. Despite this, the average consumption of fruit and vegetable intake at baseline for women in the high and low dose group was 5.2 and 4.5 daily servings, respectively. Regarding physical activity levels at baseline, the average minutes per week in MVPA at baseline was 137 and 181 for those in the high and low dose arms, respectively. Thus, women in our study were relatively close to meeting the recommended nutrition and physical activity guidelines proposed by the American Cancer Society at study enrollment. Although women in the high dose intervention significantly improved their fruit and vegetable intake by 1.5 servings per day relative to women in the low dose intervention, by the end of the study, women from both dose arms were, on average, meeting the nationally recommended guidelines for fruit and vegetable consumption and minutes spent in MVPA.

This trial has multiple strengths. The intervention was based on both Social Cognitive Theory and Self-Determination Theory. Lifestyle behavioral interventions based upon theoretical frameworks have been shown to have stronger effects than those that are not theory-based. It is novel to have tested two dose levels of an intervention. The low dose intervention was designed to be like clinical usual care that may offer a one-time class on lifestyle behaviors for cancer survivors with some low-touch eHealth supports (text messaging and a website). The high dose intervention was designed to provide didactic and experiential education. Another strength is the 6-month duration of the intervention as most trials of this nature are limited to 12 weeks21. Lastly, the COVID-19 pandemic forced the implementation of a remotely delivered intervention, which ultimately is a more scalable delivery method.

Despite these strengths, this pilot study has limitations. Most of the women enrolled in the study were White with high measures of socioeconomic status based on average level of education and annual household incomes. This limits the generalizability of our results. Furthermore, whether similar results would be observed in more culturally diverse samples or in women who are non-English speaking or with low socio-economic status remains largely unknown. Wang et al. also recognized a lack of racially and economically diverse study samples across the 23 studies reviewed and recommended future studies test eHealth interventions in larger and more diverse populations21. The implementation of multiple eHealth components makes it challenging to distinguish which components were effective at influencing positive behavior changes. However, recent technological advances, such as automation of text messaging and remote downloading of data, allow the inclusion of multiple modes of exposure without increasing costs or burden to study staff or participants12. Moreover, the systematic reviews evaluated in this paper suggested that incorporating multiple eHealth components, such as online education programs, text messaging, and study-specific websites, is a common and likely approach for future behavioral eHealth interventions12,13,21. Finally, a third of the participants did not provide accelerometer data at follow-up and it is possible these women might have had lower levels of physical activity. This would cause an overestimation of physical activity levels in our findings.

In summary, we found that an online intervention delivered twice monthly over 6 months to early-stage breast cancer survivors successfully improved daily servings of fruits and vegetables and decreased caloric intake and total fat consumption. However, the intervention did not yield significant improvements in physical activity. Given the substantial number of cancer survivors in the US (18+ million), scalable interventions are crucial to promote healthy lifestyle behaviors. In our future trials, we will explore scalable, culturally tailored strategies to enhance diet quality and manage weight effectively in women with breast cancer, with specific focus on survivors from underrepresented populations.

Methods

Study design

The Cook and Move for Your Life study design and protocol has been previously described18. Briefly, Cook and Move for Your Life was a phase two, randomized controlled pilot trial that aimed to test the feasibility and preliminary efficacy of two scalable doses of a remotely delivered nutrition and physical activity intervention for breast cancer survivors. The low dose arm included one real-time 90-min online group session within the first month of enrollment. The high dose arm offered twelve real-time 90-min online sessions delivered twice monthly over six months. All participants received eHealth communications, including interactive text messages, electronic newsletters, a Fitbit device, a Fitbit Aria WIFI-enabled scale, and access to study-specific password protected and HIPAA compliant webpages on the Fred Hutchinson Cancer Center’s Cook for Your Life website (cookforyourlife.org). Informed consent from all study participants was obtained electronically via REDCap, a HIPAA-compliant platform. The study and protocol complied with all ethical regulations, including the Declaration of Helsinki, and was approved by the Institutional Review Board at the Fred Hutchinson Cancer Center (NCT04200482).

Study population and recruitment

Women were recruited from Seattle Cancer Care Alliance (now known as Fred Hutchinson Cancer Center) patient database registries or direct physician referral from the breast oncology clinic (Fig. 1). Potential participants were mailed a recruitment letter signed by their physician, then called by phone to gauge interest and screen for eligibility. Women were eligible to participate if they were at least 18 years old, not pregnant, English-speaking, and a non-smoker who had a history of early-stage breast cancer (American Joint Committee on Cancer Stage 0-III with no recurrent or metastatic disease); were at least 60 days post treatment (current endocrine therapy use permitted) and physically functional based on the Eastern Cooperative Oncology Group scores (0-1); had no uncontrolled diabetes or hypertension; self-reported consuming less than five servings of fruits and vegetables per day and/or spending less than 150 min per week in MVPA based on screener questionnaires25,26, and had internet access via a smartphone or computer18.

Data collection

Upon receiving their consent, women were mailed the following materials for baseline data collection: 1) an Actigraph GT3X+ accelerometer (Actigraph LLC., Pensacola, FL); 2) a stool and blood collection kit; 3) a Fitbit Aria WIFI-enabled scale (Fitbit Inc., San Francisco, CA); and 4) pre-paid mailers to return the materials back to the study center. To objectively measure physical activity, including time spent in MVPA, participants were asked to wear the accelerometer for seven consecutive days. To evaluate changes in the gut microbiota and systemic inflammation, participants were asked to collect fresh stool and blood samples via the instructions provided. Each Fitbit Aria WIFI-enabled scale was linked to every participant allowing us to objectively capture their weights. In addition, web links were sent to participants for completion of online, REDCap-administered questionnaires. Finally, participants completed three (2 weekday, 1 weekend day) 24-hour dietary recalls administered by study staff over the telephone. These procedures were repeated at the 6-month follow-up, which also included an exit interview.

eHealth intervention

Upon submitting completed baseline data, a permuted block randomization implemented by our study’s statistician was used to allocate participants to receive the low dose or high dose intervention. The Cook and Move for Your Life intervention was based upon Social Cognitive Theory and Self-Determination Theory theoretical frameworks to target behavioral constructs focused on making positive behavior changes18. The low dose intervention included one 90-min nutrition and physical activity group session delivered in the first month of enrollment. The high dose intervention included twelve 90-min nutrition and physical activity education group sessions delivered twice monthly over the 6-month intervention period. The group sessions were from a tailored curriculum focused on improving diet and physical activity knowledge and self-efficacy, addressing barriers to health behavior change, and building skills around cooking, eating, grocery shopping, and engaging in various forms of exercise intensities. Specific topic areas for each session by dose arm have been described18. All sessions comprised diet and exercise education, hands-on experiential learning for skills building, and a question-and-answer session. Education and experiential learning components were taught by a registered dietitian, trained chef, culinary educator, and a certified exercise physiologist. All sessions were recorded and made accessible for viewing on the study website according to the participant’s study arm.

To encourage study adherence and retention, all participants, regardless of dose arm assignment, received six months of eHealth communications comprising of weekly motivational texts, biweekly electronic newsletters, a Fitbit activity tracker device for self-monitoring physical activity, and study-specific access to the FHCC-sponsored Cook for Your Life study website. Over the intervention period, 96 text messages were sent to participants, of which 23 requested a response. On alternating weeks, an automated text message was sent asking women to set a nutrition or physical activity goal for that week, which was followed up the next week with another text message asking if the goal was achieved. Alongside eHealth communications and the Fitbit device, all participants also received kitchenware (e.g., chef’s knife, measuring cups, cutting board) and an exercise band (Dyna Band, Bucks, UK) by mail, to promote healthy eating and activity and to enhance study adherence and retention.

At the end of the study, all women randomized to the low dose arm were given online access to all the recorded group sessions from the high dose intervention. Access to the CMFYL study website was provided to all participants for 12 months after the study ended.

Measures

Feasibility was assessed based on the accrual rate, adherence, retention, and acceptability. The accrual rate was defined as the number of eligible women enrolled. The recruitment goal was 90 women, with an expected enrollment and randomization of 75 women, which accounted for a 17% withdrawal and baseline data incompletion rate based on our previous trials with breast cancer survivors. Adherence was defined as the proportion of women who attended the online group sessions, responded to text messages, wore the Fitbit device, and accessed the study website. High adherence was characterized as 60–70% participation, a threshold based on our experience with similar lifestyle interventions in breast cancer survivors27,28. Thus, high adherence to the intervention was attending eight or more of the 12 online sessions among high dose participants and the single session among low dose recipients. Attendance to the sessions was defined as either participating in the real-time group session or viewing the recorded session on the study website, objectively captured via Google Analytics data (Google LLC, Mountain View, CA). Since eHealth communications were the same for both doses, high adherence for text messaging was receiving 14 or more text message responses and, for Fitbit, was wearing the device for at least 70% of the days over the 6-month study period. Retention was defined as the proportion of women with 6-month follow-up data from the online questionnaires, 24-h dietary recalls, accelerometers, and biospecimens, with a prespecified goal of collecting data from 75% or more of the sample. Acceptability was measured from responses to select questions about the intervention asked in the exit interviews.

Dietary intake was calculated using the Nutrition Data System for Research (versions 2016–2019, developed by Nutrition Coordinating Center, University of Minnesota, Minneapolis, MN). Physical activity data was collected by the Actigraph GT3X+ accelerometers, which captured movement across three axes at a frequency of 30 Hz and processed in one-minute epochs using ActiLife V6.13.3 software (Actigraph LLC, Pensacola, FL). Non-wear time was classified using the Choi algorithm for triaxial accelerometers29. Valid wear time was defined as having at least 10 h per day of wear for four or more days29,30. Only participants who met this wear time criteria were included in the physical activity analysis. MVPA was defined from the ActiLife software.

Demographic, socioeconomic, and clinical characteristics were self-reported at baseline via online questionnaires. Participants also completed measures of self-efficacy and food preference, comfort with technology, physical functioning, and anxiety31,32, and self-reported their height and weight. Weight was objectively captured via the WIFI-enabled Fitbit Aria scale. Body mass index was calculated from height and weight. Cancer stage, years since diagnosis, and treatment history were abstracted from participants’ electronic health records.

Statistical analysis

Participant characteristics and feasibility measures were compared by dose arm using T-tests and Chi-square tests. Normality in the distributions of the outcomes were examined using univariate statistics and histograms; all outcomes were reasonably normal. Using an intent-to-treat approach, differences in 6-month study outcomes between dose arms were estimated from linear regression models as a function of randomization assignment and baseline value of the relevant outcome. A two-sided p-value of <0.05 was considered statistically significant and all analysis was programmed using SAS v9.4 (SAS Institute, Cary, NC).