A prospective test-retest quasi-experimental study was carried out with 80 breast cancer survivors. The breast cancer survivors were selected based on the following eligibility criteria: (1) breast cancer stage I, II, or IIIA, (2) 30 to 75 years old, (3) body mass index (BMI) over 25 kg/m², (4) user-level skills for app management, and (5) completed the adjuvant treatment at least 6 months before being included in the study. Eligible participants were excluded if they had mental or physical health conditions that prevented them from walking and/or participating in the assessment or if they did not sign the informed consent form. In addition, participants had to have access to a mobile device or tablet with an internet connection and an Android operating system. The research team loaned out two devices in cases where this was not possible or the operating system was incompatible with the app. All participants were recruited through the oncology unit from the University Hospital Complex of Granada, Spain, after being informed about the study and being referred by their respective oncologist. All eligible participants were contacted via telephone, screened using the inclusion and exclusion criteria, and if they were interested in participating, cited for the baseline assessment.

This study was approved by the ethics committee of the Andalusian Health Service (FIS, PI14-01627; Granada, Spain) and it was performed in accordance with the Helsinki Declaration for biomedical research (14/2017) [35]. Participants completed informed consent forms before the assessment.

The CUIDATE research group developed the Energy Balance on Cancer (BENECA) mHealth app to monitor and provide feedback to breast cancer survivors on healthy eating and PA. A description of the BENECA mHealth System [31,36] and a reliability study for the same [32] were previously published. After the baseline assessment was performed, a member of the research group downloaded the app on a patient’s mobile phone and taught them how to use it. The patient then had to prove that she understood the instructions by using the app in the presence of the researcher. Patients had to use BENECA mHealth for 8 weeks during the study. Physical activity (duration and intensity) and diet (food and drink intake) data were recorded via the app (self-recorded). Intake was recorded using a dietary record questionnaire; BENECA is structured with six consumption times. On each day, for each period, users report all food and beverages consumed. For PA, BENECA incorporated the Minnesota Leisure-time Physical Activity Questionnaire. Patients had to record intensity and duration of activities each day; BENECA only recorded those activities with a duration of at least 10 minutes. Using this information, the app sent a notification to the user of their daily energy balance, offering recommendations on diet and PA, which were based on the guidelines of the World Cancer Research Fund International, the strategies for PA and diet in patients with cancer from the American College of Sports Medicine [37], and the recommendations of the American Cancer Society [38]. Users receive a straightforward daily notification if there has been an energy imbalance; any difficulties in handling the app were resolved via calls and text messages between the researcher and patient (Multimedia Appendix 1). BENECA had been developed based on the theories of learning, Goal-Setting Theory, and Social Cognitive Theory to include techniques such as reinforcement, facilitation, self-monitoring, goal setting, feedback on performance, and reviewing goals, which have demonstrated to be promising in increasing PA in different populations [33,39]. A video tutorial was made available to the patients to review the use of the app.

Patient demographic and clinical data were obtained at the beginning of the study using a study-specific survey. Baseline data were gathered at the start of the study and again after 8 weeks of using BENECA mHealth. The outcomes measured are presented subsequently.

All analyses were performed using SPSS Statistics version 24 (IBM Corp, Armonk, NY, USA). Statistical significance was assumed when P<.05. The logging data from BENECA mHealth were obtained on request from the computer engineers responsible for the development of the app.

First, descriptive measures were used to report the data on adoption, use, attrition, and quality, as well as to report on the clinical and anthropometric variables of the participants. A Kaplan-Meier survival curve was used to visually examine the survival curve of the entire cohort to determine the attrition. In the analysis, an “app survivor” was defined as a breast cancer survivor that maintained logging practices using BENECA mHealth until at least 3 days before the last day of the experimental period. Those defined under “app death” were those who missed five consecutive daily loggings (based on a previous study [53]). A Kaplan-Meier estimator with right-censored data was used. This type of data was used because it best fit our study results. As most of the breast cancer survivors “survived” until the end of the experimental period, we do not know how long they would have continued using BENECA mHealth after this period. Then, a Cox proportional hazard model was used to examine if age, marital status, and employment had any effect on the attrition.

Second, to assess the pretest-posttest differences in the main outcomes, an analysis of paired-sample t tests was used and, when appropriate, Wilcoxon signed rank tests were conducted. Moreover, the effect size (ES) estimate was determined and interpreted using Cohen’s guidelines of 0.1=small effect, 0.3=medium effect, and 0.5=large effect.

Third, to assess differences between “users” and “nonusers” and the patients’ perception of BENECA mHealth quality, a Mann-Whitney test was used for categorizing the breast cancer survivors according to the cut-off used in the survival analysis. A simple linear regression was used to examine the influence of age on the perception of BENECA mHealth quality.

Our data contained a few missing values (5%, 4/80 of the total number of cases), but these can be considered random and inconsequential [54]. Hence, no multiple imputation method was necessary (casewise deletion was used).

The baseline demographic and clinical characteristics of the participants (mean age 51.80, SD 8.64 years) are presented in Table 1. Of the 80 breast cancer survivors, 50 (62%) were married, 31 (38%) had a higher education, and 40 (50%) were diagnosed with stage II breast cancer, followed in frequency by stage IIIA (28/80, 35%). All participants received instructions on how to use BENECA mHealth to monitor energy intake and expenditure. Four participants were unable to be assessed postintervention (dropouts); three were not assessed due to changes in their health status unrelated to the study, and one decided to discontinue.

The mean MARS quality score for the app was 3.71 (SD 0.47) out of 5, and the NPS was positive (6.58 in range of −100 to 100), consisting of 24% (19/80) detractors, 30% (24/80) promoters, and 46% (37/80) passives. On average, the best-rated MARS category was app-specific change (mean 4.30, SD 0.37), followed by information (mean 4.22, SD 0.51), app subjective quality (mean 3.73, SD 0.46), and functionality (mean 3.71, SD 0.52). The worst-rated section was aesthetics, with a mean of 3.25 (SD 0.63). The specific scores for each section of the MARS are illustrated in Figure 3. The participants were divided according to the cut-off used in the survival analysis. It shows how the participants who used BENECA until the end of the experimental period scored higher and were statistically significant in all sections (P<.001). Linear regression showed that the older the patient, the lower the app quality score (beta=−0.29, t₇₅=−2.64, P=.01).

In summary, seven barriers and five facilitators were quoted five times or more when the participants were interviewed. Among the barriers, the most common was “BENECA does not have some food items” followed by “difficulty at the time of introducing the intake.” Among the facilitators, the most common was “BENECA provides relevant information to the patient” followed by “patient considers it important to know BENECA’s feedback on energy balance.” Table 3 summarizes the barriers and facilitators mentioned.

According to our initial hypothesis, after using BENECA mHealth for 8 weeks, the app was considered feasible by the breast cancer survivors in terms of use, adoption, and satisfaction, although the app quality score did not make it one of the best-rated apps. BENECA mHealth was associated with changes in the QoL of breast cancer survivors, as well as their self-perception of effectiveness and motivation for engaging in PA in their daily life.

The adoption rate in this study was 69%, and the usage rate was 73% to 76%. These results can be explained by the technical characteristics of BENECA mHealth and its functionality, such as user-friendliness, the use of internationally accepted measures, and the visual feedback. The results of this study are comparable with those obtained by Melissant et al [13] for a supportive care app for breast cancer survivor, which had an adoption rate of 75% and usage rate of 75% to 84%. Another study of a lifestyle intervention with a mobile app for endometrial and breast cancer survivors recorded a 75% usage rate [55]. However, Duman-Lubberding et al [11] obtained an adoption rate of 64% and a usage rate of 75% to 91% for a similar app for head and neck cancer survivors. The somewhat lower rate of use in our study for the latter may be due to how these data were obtained (ie, by the number of log-ins—objective measure—instead of the self-reported data of those studies—subjective measure). With regard to “app survival,” we found that in a study by Springer et al [53] to test an mHealth app targeting healthy eating behavior in the general population, they obtained a survival rate less than 60% using the Kaplan-Meier survival curve. The higher survival rate in our study (over 70%) can be explained by the type of population studied. In general, patients with some type of pathology will be more predisposed to be involved in this type of study than the general population [56]. In addition, experiencing cancer treatment may be a stimulus to use the app, as patients may feel the increased need to learn more about the treatment.

Taking into account the barriers perceived by the participants in the use of the app, the barriers reported by BENECA mHealth were in line with a recently published review on the adherence to online psychological interventions [57] as well as with those in a study by Melissant et al [13] with the Oncokompas app to monitor the QoL of breast cancer survivors (eg, “Oncokompas is too extensive”). The reported mean satisfaction score of the quality of BENECA mHealth, although it may seem not very high, is in line with a recently published study on the quality of 18 mobile apps for pain management using the same MARS quality scale [41]. In addition, the low scores in some sections can be explained by the barriers reported by the patients, such as the difficulties in inputting the intake that makes it very extensive to fill in the app. This barrier was also reported in another feasibility study on head and neck cancer patients [11]. Considering that the minimum score to be considered a best-rated app based on the MARS scale is 3.73 (according to a previous study [41]), BENECA mHealth can be regarded as an app with average ratings. BENECA is currently being improved in an attempt to address the reported barriers.

The benefits of PA for cancer patients have been amply demonstrated [58], although a recent meta-analysis (2013-2018) of distance-based PA behavioral change interventions for cancer survivors concluded that the effects of interventions on PA were small [18]. In addition, although efficacy cannot be discussed in a study such as this, according to the literature, a difference of 8 points between assessments of QoL measured with the EORT QLQ-C30 is the minimum clinically significant difference required to discuss the clinical relevance of the findings [59]. The QoL findings in this study reinforce these preceding conclusions and are consistent with the results of the EAF scale and those observed via accelerometry. Changes are observed in the participants with more motivation to do PA, and it seems that using BENECA mHealth is associated with changes that lead to a positive feedback chain that improves physical and emotional functioning. The significant differences in cognitive functioning can be explained by the actual use of the mobile device, as there is evidence of the cognitive benefits of using electronic devices [60]. Our findings are in agreement with those reported by Pope et al [20], who used a mobile app and social media for 10 weeks to improve the QoL of breast cancer survivors, with a sample size much smaller than ours. However, they differ from the conclusions of McCarroll et al [55], who assessed the effectiveness of a public mobile app (LoseIT) for dietetic intervention for 4 weeks in breast and endometrial cancer survivors. They did not find significant changes in the QoL of the patients. It is possible that the experimental period of 4 weeks and lack of stratification of the type of cancer could explain these differences, despite the use of a powerful questionnaire to assess QoL. Lastly, we only found statistically significant differences in the MVPA of the data obtained via accelerometry, although we observed an improvement in other variables after the use of BENECA mHealth. These results are consistent with those of a clinical trial published in 2018 that used smartwatches and social media PA behavioral change over a 10-week intervention to determine the health outcomes for breast cancer survivors, in which no significant differences in the accelerometry variables were observed [61].

Finally, one of the main challenges not only with cancer patients but with the general population is the maintenance and reduction of body weight [5,62]. Different studies of lifestyle interventions have shown beneficial results, such as the one by von Gurenigen et al [63] in which they evaluated the effectiveness of a face-to-face intervention on diets in obese patients with endometrial cancer, achieving a reduction of approximately 5%. Similarly, McCarroll et al [55] achieved a reduction of approximately 6% from baseline weight. The literature indicates that a weight reduction of 5% is sufficient to reduce medical comorbidities [62]. In our study, an average weight loss of approximately 2% was achieved, which is below the recommendations. This may be because BENECA mHealth is not really a lifestyle intervention mobile app, but rather one that tries to incite behavioral change in users by monitoring their energy balance and making them aware of it. Therefore, we believe that the results obtained can be considered a first step, although future research should corroborate these results. The internal architecture of BENECA mHealth can also be extrapolated to suit patients with other types of cancer.

It is important to recognize some of the limitations of this study. The main one is its design. It is a nonrandomized, single-arm exploratory study; therefore, the results should be taken with caution. The ideal study would have been a randomized controlled trial (RCT); nevertheless, it was mandatory to develop a feasibility study for this sensitive population before carrying out an RCT. Moreover, due to the nature of the design of this study, the reported results must be confirmed in a larger RCT because the observed changes may not be attributable to the intervention. Secondly, BENECA was only developed for the Android operating system, but we are currently working on the next version of the BENECA app to solve this limitation. Thirdly, BENECA was designed to monitor energy balance and then propose recommendations based on international guidelines of clinical practice, systematic reviews, and meta-analysis to ensure the recommendations can be generalized. However, we believe that it is a good starting point, especially for very sedentary people. Finally, the generalization of results is limited due to the design of the study, the use of restrictive inclusion and exclusion criteria and the recruitment strategy (the participants were referred by their oncologists), which may involve a bias of the threat of regression to the mean. In addition, another added difficulty could refer to the use of the app by older people in southern Spain, who may not even have mobile phones adapted to the app. Therefore, future studies should be conducted with a larger sample size; a controlled and randomized clinical trial design comparing the use of BENECA with, for example, a face-to-face intervention; and including biomarker measurements such as those for inflammation or development/recurrence of breast cancer.

Despite these limitations, this study also has strengths. These include the wide range of ages of the participants, which makes it possible to generalize the results; the use of energy balance as a means of changing behavior, which has not been studied much; its ease of use; it has high adherence; and it has no adverse effect on the prior validation of BENECA mHealth [32], which guarantees its reliability.

BENECA mHealth can be considered feasible in a real clinical context and has been associated with behavioral changes in the lifestyles of breast cancer survivors, but it needs to be enhanced to improve user satisfaction with use and functionality. Having assumed that BENECA is usable and applicable in a real clinical context, as well as having the first data of its applicability and clinical efficacy, the next step will be to confirm these results through a larger study with a control group. In addition, efforts should focus on overcoming the barriers reported by the participants and developing a new version of BENECA mHealth in which these improvements will be implemented. Finally, future research could focus on its generalization for application to other oncological processes. This study highlights the importance of the use of mobile apps based on energy balance and how the QoL of breast cancer survivors can be improved via monitoring. The results of this study could garner support for the use of this type of strategy in the projected 29.5 million cancer patients in 2040 [64].