Evaluating the Effectiveness of Mobile Health in Breast Cancer Care: A Systematic Review

Abstract Breast cancer is affecting millions of people worldwide. If not appropriately handled, the side effects of different modalities of cancer treatment can negatively impact patients’ quality of life and cause treatment interruptions. In recent years, mobile health (mHealth) interventions have shown promising opportunities to support breast cancer care. Numerous studies implemented mobile health interventions aiming to support patients with breast cancer, for example, through physical activity promotion or educational content. Nonetheless, current literature reveals that real-world evidence for the actual benefits remains unclear. In this systematic review, we focus on analyzing the methodology used in recent studies to determine the effects of mHealth applications and wearable devices on the outcome of patients with breast cancer. We followed the PRISMA guideline for the selection, analysis, and reporting of relevant studies found in the databases of Medline, Scopus, Web of Science, and Cochrane Library. A total of 276 unique records were identified, and 20 studies met the inclusion criteria. Study quality was assessed with the Effective Public Health Practice Project (EPHPP) Quality Assessment Tool for Quantitative Studies. While many of the studies used standardized questionnaires as patient-reported outcome measures, there was minimal use of objective measurements, such as activity sensors. Adoption, drop-out rates, and usage behavior of users of the mobile health intervention were often not reported. Future work should clearly define the focus and desired outcome of mHealth interventions and select outcome measures accordingly. Greater transparency facilitates the interpretation of results and conclusions about the real-world evidence of mobile health in breast cancer care.


Introduction
Breast cancer is the most commonly diagnosed cancer worldwide. 1This corresponds to more than 2 million individuals receiving a breast cancer diagnosis annually.To date, early detection and therapy remain challenging, especially due to the differences in healthcare systems and socioeconomic resources. 2The variety of subgroups for breast cancer requires a wide range of therapeutic measures, including surgery, radiation, chemotherapy, endocrine therapy, and targeted therapies. 3,4These therapies can lead to a range of adverse effects that can impact the quality of life (QoL) of patients, 5,6 such as nausea, fatigue, hair loss, and insomnia, among others. 6rthermore, the diagnosis and treatment of cancer can be emotionally and mentally taxing.Many patients experience anxiety, depression, and other psychological difficulties. 7,8n the past years, there has been increasing interest in the use of mobile health (mHealth) applications, [9][10][11][12] wearable devices, 13 and telemedicine 14 to support breast cancer management.Ample reviews aimed to summarize the evidence for the implementation of mHealth in breast cancer care.Suchodolska et al 15 analyzed the effects of chemotherapyrelated symptom reporting and management through mobile applications.While highlighting the feasibility and potential benefits, the authors acknowledge the limited significance of the results in the included studies.Within the systematic e848 The Oncologist, 2023, Vol. 28, No. 10 review of Jongerius et al, 10 an overview of the measured impact of mHealth applications in empirical studies from 29 studies in breast cancer care was generated.The authors conclude that the evidence of mHealth interventions for patients with breast cancer is inconsistent and recommend further exploration, especially in the area of psychological interventions.This corresponds to the findings of Cruz et al, 16 where the great potential is equally emphasized, but the actual benefit remains unclear.
All these systematic reviews have attempted to identify the current evidence regarding the use of mHealth interventions in breast cancer care.All of them come to a similar conclusion: a positive impact could exist, but the evidence remains limited in terms of its quality, consistency, and generalizability.Despite the large number of studies in the last years, it is unclear why evidence for the effectiveness of mHealth applications for breast cancer is still contradictory.
To address this knowledge gap, the objective of this systematic review is to analyze the methods used in recent studies for the evaluation of the effectiveness of mHealth interventions, such as smartphone applications and wearable sensors in breast cancer care.Through a detailed investigation of the study designs and methodologies, this review aims to identify barriers of current research to clear evidence.By acknowledging these limitations and identifying reasons that impede comparisons of different mHealth studies, we can work toward developing more standardized approaches for evaluating these interventions.This enables us to draw meaningful conclusions about the impact of mHealth in breast cancer care and facilitate the development of evidence-based interventions that can improve patient outcomes.

Search Strategy
This systematic review followed the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-analyzes) guidelines. 17A review protocol was drafted for internal use, and registration was not performed.The search was performed using the databases of Medline (Pubmed), Scopus, Web of Science, and Cochrane Library.To identify relevant records, the following search string was used; "(app OR application) AND (smartphone OR mobile OR wearable) AND (breast OR mammary) AND (cancer OR tumor)."In addition, results were filtered according to the language they were written in (only English) and the year (after January 1, 2018) they were published.All results were imported into a literature management tool to screen the records.

Eligibility Criteria and Study Selection
The definition of the eligibility criteria was performed based on the PICO Statement. 18The following study characteristics were defined to be necessary for the inclusion of a study into this review: (1) the study population was composed of patients diagnosed with breast cancer; this includes patients during active treatment as well as during the follow up.(2) A mobile application or a wearable device was used by the patients themselves as an interventional support tool.(3) The outcome of interest should measure an impact on the patients with quantitative methods.(4) The article was published in English.(5) The article was published in a peer-reviewed journal after January 1, 2018.This time span was considered to represent the most recent and state-of-the-art research on mHealth interventions for breast cancer care to increase the relevance and applicability of the findings to current clinical practice.
All publications falling under one of the following aspects were excluded: (1) studies using an application only as a platform to provide questionnaires or similar, and (2) publications such as reviews, commentaries, conference abstracts, pilot studies, conference abstracts, and qualitative studies.The search process of this systematic review was concluded in January 2022.Two reviewers independently screened all found records.First, based on the title and abstract, afterward, a full text assessment was performed for the remaining publications.Disagreements about inclusion or exclusion were discussed among the reviewers after the screening.

Data Extraction and Synthesis
The study characteristics were extracted by 2 reviewers independently into a custom structured excel sheet.The following items were collected: authors, publication year, journal, country, study design, sample size, mobile app or wearable, duration, outcome measures, and main results.Disagreements about the extracted data were discussed among the reviewers.Of every record, the author, year, journal, study design, population, sample size, country of conduction, outcome measures, mobile app or wearable features, main findings, and limitations were summarized.Based on the complexity of interventions and tools used for the outcome measures, every data collection method used in the study was categorized into one of the following topics: (1) quality of life; (2) symptom burden; (3) psychological side effects and symptoms; (4) physical activity and healthy lifestyle; (5) self-efficacy, self-perception, and social support; and (6) patient-relevant structural and procedural improvement.This topic includes positive care effects that can be indirectly related to the patient's health, eg, improvement of health literacy, patient empowerment, reduction of therapy-related effort, and burden on patients and their families.

Quality Assessment
To assess the methodological quality of the studies included in this review, the Effective Public Health Practice Project (EPHPP) Quality Assessment Tool for Quantitative Studies were used, as it enables a comparison of different study designs. 20,21Based on selection bias, study design, confounders, blinding, data collection methods, withdrawals, and dropouts, the quality is categorized as low, moderate, or strong.This classification was performed by 2 reviewers independently based on the supporting material of the tool for each publication.Disagreements were resolved with a third reviewer and consensus-based discussion.

Study Selection
In total, 533 records were identified in the extended bibliographic search.After removing the duplicates, the title The Oncologist, 2023, Vol. 28, No. 10 e849 and abstract of 276 records were screened.Sixty records met the inclusion criteria after screening the title and abstract and were eligible for the full-text assessment.Within this step, 12 records were excluded, as they were only usability or feasibility studies, 21 did not include the evaluation of an intervention, 3 only used text messages and no application, 2 described preliminary results, and 1 publication contained a concept description solely.Finally, 20 records met all inclusion criteria and were included in the synthesis of the results.The PRISMA flow diagram of the search and screening process is shown in Fig. 1.An overview of the eligible studies and their main characteristics is presented in Table 1.

Characteristics of Included Studies and Participants
The studies included in this systematic review were widely spread over the world.As shown in Table 1, most studies (n = 12, 60%) were conducted in Asia.Twenty-five percent were conducted in Europe and 15% in North America.Within the identified studies, 18 mobile applications and 1 wearable activity tracker (WAT) were used.
As shown in Fig. 2, all but one were interventional studies.One of the records used an observational design. 40An overview of the intervention period, sample size, and population of each study is shown in Table 2.The number of participants recruited for the studies varied in a range from 37 23 to 4475 40 participants.The intervention period ranged from 3 weeks 30 to 6 months. 22,31In three publications, 27,39,40 the duration of the intervention was only reported within a certain time span without further explanations about the actual duration of the intervention.

Main Features of the mHealth Applications
Fig. 3 shows the distribution of the different features for all 18 applications and the wearable categorized according to Mendiola et al. 19 The categories of Usability, Cost, and Social Media were omitted.The usability could not be evaluated due to missing access to most of the applications, and a usability analysis was not part of the studies.Because most apps were developed for research purposes and were, therefore, presumably made available to the participants free of charge, and cost was not analyzed.None of the analyzed concepts included a social media feature.Only 5 of the identified 19 mobile apps were available on the app stores of Google and Apple (Vida, 22 WalkOn, 23 Msemptom, 34 Pillsy, 35 Headspace 37 ).None of them were specifically advertised for patients with breast cancer.While Pillsy and Msemptom are targeted at patients suffering from a disease or cancer in general, the other 3 apps are also designed for a broad audience to promote a healthy lifestyle.Within 3 studies, 36,39,40 the application's name was not mentioned, so it remains unclear whether these applications were published.

Evaluation of User Behavior
Seven studies 23,24,26,27,29,33,34 did not report whether any usage behavior was logged.Cairo et al 22 only mentioned that in a

e850
The Oncologist, 2023, Vol. 28, No. 10 Table 1.Overview of the studies included in this systematic review and their main characteristics.follow-up consultation 6 months after the intervention, no participant used the app anymore.Visser et al, 38 as well as Wyatt et al, 39 analyzed self-reported usage of the application.The login frequency was reported in 3 of the studies. 28,37,41In the study of Rosen et al, 37 the authors additionally had a look at the behavior within the app (performed lessons, date, and time of actions).Zhu et al 41 also logged information about the duration of time spent in the app.Park et al 36 mentioned that login frequency and the duration of usage were tracked but did not report data about it.Two records 25,40 evaluated whether the app was used.They reported that 23% of the participants never logged into the app.In the mobile game, 30 the time the users spent playing was reported.Lozano-Lozano et al 32 reported adoption, usage, and attrition rates in detail.This included the number of participants initially agreeing to use the application, actual active users, and dropouts over the whole study duration.In one study, 42 usage frequency was indirectly measured through the medication tracked through the application.Finally, in the study incorporating a WAT, 31 compliance was assessed by counting all days the WAT recorded more than zero steps.

Quality Assessment
The summarized results of the quality assessment of the studies can be found in Fig. 4. The result for the individual studies is listed in Table 1.Many studies were particularly weak in terms of the selection of study participants and the blinding.In contrast, most studies chose data collection methods that were shown to be both valid and reliable.Two of the studies 39,42 used custom scales.Further details about the used outcome measures, their aims, and information about their validity and reliability will be analyzed in the following chapters.

Outcome Measures
The core aspect of the studies are the outcome measures used to show the impact of the mHealth application on patients with breast cancer.An overview of the findings is represented in Fig. 5.

Quality of Life
In 11 of the identified studies, 24,25,27,29,30,[32][33][34]36,38,41 one of the outcome measures were quality of life (QoL). In total 6 different questionnaires were applied.Seven of the studies used the validated QoL of Cancer Patients questionnaires of the European Organization for Research and Treatment of Cancer (EORTC).Five of them 27,29,33,34,38 used both the QLQ-C30, for patients with cancer in general, 43 and the BR-23, specifically developed for patients with breast cancer.44 Two of the studies 25,32 only used the QLQ-C30.While 5 of the studies 27,29,32,33,38 reported the general QoL with these modular questionnaires, 2 25,34 only reported specific subscales.
Three of the studies 24,36,41 made use of validated questionnaires from the Facit Group. 45Zhu et al 41 used the FACT-B, 46 a questionnaire specifically designed for measuring the health-related QoL in patients with breast cancer.Park et al 36 incorporated the FACT-G 47 in their evaluation, a questionnaire for patients with cancer in general.The FACT-ES was applied in the study of Çinar et al, 24 with its special focus on patients receiving endocrine treatments. 48To evaluate their mobile game, Kim et al 30 used the WHOQOL-BREF 49

Symptom Burden
In 8 22,25,28,30,34,36,38,41 of the studies, the effect of the intervention on symptom burden was analyzed.Five studies 22,25,34,36,41 used validated questionnaires to measure symptom burden.Three studies 28,30,38 did not use validated questionnaires, and 2 of them 28,30 analyzed symptoms through patient-reported outcomes (PRO) in the application.In 2 of the studies, 25,34 the Memorial Symptom Assessment Scale (MSAS) 50 was used to assess comprehensive information on common cancerrelated symptoms.Fjell et al 25 additionally assessed the number of reported symptoms through their mobile app.To evaluate specific symptoms during endocrine therapy, Park et al 36 used the subscale for menopausal symptoms of the FACT-ES. 48he MD Anderson Symptom Inventory (MDASI) was used in the study by Zhu et al 41 to capture a variety of symptoms and their severity as experienced by the patients.Visser et al 38  The Oncologist, 2023, Vol. 28, No. 10 e853 used the SCL-90, a Dutch questionnaire.Cairo et al 22 solely assessed the subjective cancer-related fatigue and made use of the Visual Analog Scale-Fatigue (VAS-F). 51wo studies 28,30 did not use questionnaires.Instead, the number of side effects that were reported through the mobile app was counted in one of the studies. 28Kim et al collected the number of adverse events according to the Common Terminology Criteria of Adverse Events (CTCAE). 30

Psychological Side Effects and Symptoms
Eight of the studies [22][23][24]26,28,30,38,41 investigated the influence of their mHealth intervention on psychological side effects and symptoms. The most promient evaluated aspects in this area were anxiety and 26,28,30,41 other aspects included distress, 23,24 mindfulness, 37 and fear of recurrence.38 Two of the studies 26,30 used the Spielberger StateTrait Anxiety Inventory (STAI) to analyze the effects of their intervention on anxiety.Kim et al 30 additionally applied the Beck Depression Inventory (BDI) to determine the severity of depressive symptoms.Park et al 35 used the Center for Epidemiologic Studies Depression Scale (CES-D) for assessing depressive symptoms.To evaluate both anxiety and depression, the Hospital Anxiety and Depression Scale (HADS) was used in 2 of the publications.28,41 In 2 of the studies, 23,24 distress among participants was measured using the Distress Thermometer of the National Comprehensive Cancer Network (NCCN-DT). Cairo et al22 applied the Patient Health Questionnaire-2 (PHQ-2) 52 to capture the frequency of depressed mood and anhedonia of the study participants. One stud 38 assessed the fear of recurrence of patients using the Cancer Worry Scale (CWS).
Finally, Rosen et al 37 used the Mindful Attention and Awareness Scale (MAAS) to capture the self-assessed dispositional mindfulness of participants in their study.

Physical Activity and Healthy Lifestyle
In 5 of the studies, 22,23,[31][32][33] physical activity and a healthy lifestyle were investigated.Two of them 23,31 made use of activity sensors built into the smartphone or wearable device as an indicator of physical activity.Two of the studies 32,33 included functional tests and biomarkers.Three studies 22,32,33 analyzed effects on the weight or body composition of participants.One study investigated the nutrition of participants. 22ong et al 31 included a WAT to accompany the mobile application.To assess physical activity, the authors analyzed the steps recorded through this tracker and additionally used the Korean version of the Global Physical Activity Questionnaire (GPAQ) of the World Health Organization. 53,54Chung et al 23 included a step counter into their application.As an outcome measure for physical activity, the authors compared the total weekly steps of patients.The Oncologist, 2023, Vol. 28, No.
In the study of Cairo et al, 22 the Godin-Shephard-Leisure-Time-Physical-Activity Questionnaire was used.Furthermore, body mass index (BMI) and weight were measured at baseline and postintervention as an indicator of lifestyle behavior.A nutrition assessment using the Rate Your Plate questionnaire, which assesses eating patterns, was also conducted.
Lozano-Lozano et al used a diverse set of outcome measures in their 2 studies. 32,33In their earlier study, 32 accelerometer data of the smartphone, the weight, and body composition were analyzed.Complementary patients' motivation for physical activity was investigated using the validated Spanish self-efficacy scale for physical activity (EAF). 55In the follow-up study, they focused on upper-limb functionality and body composition. 33The assessment of existing disabilities was performed with a Spanish version of the Disabilities of the Arm, Shoulder, and Hand (DASH) questionnaire.The active range of motion (AROM) of the shoulder was measured using a goniometer, and a digital handgrip was used to measure upper-body muscular strength.
Self-Efficacy, Self-Perception, and Social Support Self-efficacy was assessed in 2 different studies 36,41 with 2 different questionnaires.One study 26 investigated self-esteem, and 2 studies 38,41 evaluated the social support of patients.One of the studies 39 evaluated the confidence of patients in making decisions with a custom scale.Park al 36 assessed the self-efficacy of their participants with the Self-Efficacy Scale for Self-Management for Breast Cancer (SESSM-B). 56The Stanford Inventory of Cancer Patient Adjustment (SICPA) was used in a Chinese version in one of the studies 41 to investigate the self-efficacy of participants.Ghanbari et al 26 made use of the Rosenberg Self-Esteem Scale (RSES) to investigate self-esteem or an individual's overall sense of self-worth and personal value. 57mpowerment in patients with breast cancer was assessed with a Dutch version of the Cancer Empowerment Questionnaire (CEQ) in one of the studies.38 Zhu et al 41 analyzed the social support of participants with the Chinese version of the Multidimensional Scale of Perceived Social Support (MSPSS).The confidence in making decisions was measured in one of the studies.39 Participants rated whether they prefer to make their own decisions or more doctor-guided decisions.In addition, participants ranked their confidence on a scale of 1 to 10.

Patient-Relevant Structural and Procedural Improvement
As outcome measures for patient-relevant structural and procedural improvements, the studies investigated therapy and medication adherence, 30,35,38,40 the number of visits to the doctor or hospital, 27 health literacy, 28 and medication self-efficacy. 35hree of the identified studies 30,35,38 investigated the adherence to the medication, and one 40 to the therapy in general.Medication adherence was measured using the Korean Version of the Medication Adherence Rating Scale (K-MARS) 58 in the study of Kim et al 30 Visser et al 38 used the 5-item Medication Adherence Report Scale (MARS-5). 59The adherence to the therapy of the participants in the study of Yu et al 40 was assessed through medical staff, usually the follow-up specialist or a nurse using predefined criteria.One of the studies 27 evaluated the self-reported number of doctor visits and hospitalizations of patients during therapy.
One 28 of the included studies investigated the health literacy of patients with the Japanese Health Literacy Scale (HLS-14). 60In the work of Park et al, 35 medication self-efficacy was measured with a self-developed scale on which no details were provided.

Discussion
Our findings show that current study designs are heterogeneous.This makes it difficult to derive direct comparisons and generate overarching evidence and conclusions.To support patients and healthcare providers in the adoption of mHealth applications, future developments should be guided through clear and consistent scientific guidelines or frameworks.A thorough understanding of the effectiveness of mHealth interventions is a prerequisite for the development in the field.At the same time, this will reduce unnecessary or even detrimental interventions.

Targeted Outcomes of Specific Features Should Be Defined and Reported
The mHealth interventions analyzed in this review have a broad range of features (Fig. 3).This variety prohibits the assessment of how the effectiveness of these apps is related to their specific features.Some features may support positive outcomes, while others may not have an impact or may even cause negative outcomes. 61Our review reveals that descriptions of features, content, and interfaces were often limited.This hinders the understanding of how the different features contribute to the effectiveness of apps.It makes it challenging to compare the results across studies and, thereby, determine which features are most important for promoting positive outcomes and designing more effective interventions in the future.Open-source development of mHealth interventions may support the interpretation and reproducibility of results.

Higher Transparency of Adoption, Adherence, and User Behavior Is Necessary
Thirty-five percent of the studies did not report any information about the user behavior.Without such information, the relation between the app and the effectiveness of the intervention cannot be adequately compared.This is especially important in light of short usage durations, the limited number of participants as well as, in some cases, wide inclusion criteria.This results in a study population with patients of many different stages and levels of severity and, thus needs, being included in a single study.3][64] Transparent reporting of acceptance, dropout, and engagement is essential to interpret the findings, as they may entail a bias in the final sample.

Selection Biases Are Common
The quality assessment of the included studies revealed that, in general, the methodological quality was good.However, a common issue found among the studies was the presence of selection bias, mainly due to their single-center nature.
Furthermore, study participants were aware of the research questions in most of the studies.In contrast to these limitations in study quality, the data collection methods used were, in most cases, deemed to be valid and reliable.Finally, the reported information by studies was insufficient to determine the actual risk of bias.

Heterogenous Outcome Measures Hinder Comparability
Our detailed analysis showed a large number of different outcome measures for single parameters.While the authors mostly chose validated measures, a lack of consistency in the measurement of outcomes makes it more difficult to draw meaningful conclusions about the effectiveness of mHealth applications.The development of a set of appropriate, standardized outcome measures that can be used consistently across studies reduces variability and improves comparability in the future.

PRO Measures Should Be Complemented by Objective Measures
In all studies, questionnaires were the most commonly used method to assess PROs.In addition, researchers made use of reporting symptoms via their respective mobile apps.Previous studies have shown a significant improvement between such patient-reported symptoms and the overall survival rate of patients. 65,66As the ESMO Clinical Practice Guidelines highlight, using electronic systems to collect PRO measures can be beneficial for both patients and caregivers. 67Health interventions have a great potential to streamline e856 The Oncologist, 2023, Vol. 28, No. 10   the implementation of PROs clinical by enabling remote data collection, longitudinal monitoring, and individual feedback systems.However, objective measurements such as physical activity tracking through smartphone sensors have been found to provide valuable insights that might not be captured through a questionnaire. 68Furthermore, expanding the scope of objective measurements to include physiological parameters such as heart rate or blood pressure can also lead to a deeper understanding of patient health and new opportunities in tailoring therapies.Future studies should explore and compare the effectiveness of different measurement methods through patient-centered strategies.This ensures the assessment of multidimensional parameters such as quality of life.In addition, the development of new objective measures needs to be considered to improve data quality and enhance the potential impact of mHealth interventions for breast cancer care.

Limitations and Strengths
Some limitations apply to our work.Only a limited number of databases were searched and studies falling into the scope of this review might have been missed.Due to the strong heterogeneity of the included studies, a meta-analysis was not possible.Another limitation is the short time span the publications were chosen from.Due to large advances in the field of mHealth interventions in recent years, especially pushed forward through a global pandemic, we chose this time frame to be suitable.
The strength of this review is the systematic search process for eligible studies with strict criteria and the data extraction and analysis that was independently performed by 2 researchers.This limited the risk of errors during the synthesis process and quality assessment.

Conclusion
The use of mHealth technologies in breast cancer care has gained increasing attention in recent years.Such technologies can improve patient outcomes, empower affected individuals, and finally reduce the burden of treatment on both patients and healthcare providers.However, the methodology and results of effectiveness evaluations of such interventions are still inconsistent.Consequently, a particular precaution is necessary due to the limited evidence.With this systematic review, we were able to pinpoint underlying factors that hinder the comparability of current mHealth applications applied in breast cancer care and provide guidelines to overcome these barriers in future work: Especially, the wide variety of outcome measures used to analyze single parameters (eg, QoL, symptom burden), the not-targeted app features and the lack of transparency in reporting applications and user behavior reveal the need for improvement to drive efficient and effective development.Therefore, for future studies, we recommend a clear focus on specific features and outcome parameters, the comprehensive provision of information on the functionality and content of mHealth applications, and the development of gold standards for the evaluation of effectiveness in the respective areas of mHealth applications in breast cancer care.

Figure 1 .
Figure 1.PRISMA flowchart representing the identification, screening, and inclusion phase of this systematic review.

Figure 2 .
Figure 2. Overview of the different study designs and their interventional or observational character.

Figure 3 .
Figure 3. Overview of the respective features within the mobile applications and wearables, clustered according to Mendiola et al.19

Figure 4 .Figure 5 .
Figure 4. Overview of the quality ratings of the subcategories of the quality assessment with the (EPHPP) Quality Assessment Tool for Quantitative Studies.

Table 2 .
Summary of the sample sizes, intervention period, and populations included into the studies analyzed in this review in alphabetical order.