Effectiveness of mHealth diet interventions in cancer survivors: A systematic review and meta-analysis of randomized controlled trials

Objective To evaluate the effects of mobile health (mHealth) diet interventions on cancer survivors’ diet intake, weight change, waist circumference, hip circumference, and quality of life (QoL). Methods The PubMed, Embase, Web of Science, Cochrane Library, Scopus, ProQuest, China National Knowledge Infrastructure, Wanfang, and SinoMed databases were searched from their inception to September 25, 2022. Randomized controlled trials (RCTs) on the effects of mHealth diet interventions in cancer survivors were identified. Two researchers independently selected the included studies and appraised their quality. The methodological quality of the included studies was assessed using the Revised Cochrane risk-of-bias tool for RCTs (RoB2). Results A total of 15 RCTs involving 2363 cancer survivors were included. MHealth diet interventions significantly improved fruit and vegetable intake (standardized mean difference [SMD] = 0.19, 95% confidence interval [CI] [0.05, 0.33], P < 0.01), and QoL (SMD = 0.13, 95% CI [0.01, 0.26], P = 0.04) and reduced fat intake (SMD = −0.22, 95% CI [−0.34, −0.11], P < 0.01), weight (SMD = −0.35, 95% CI [−0.48, −0.22], P < 0.01), waist circumference (MD = −1.43, 95% CI [−2.33, −0.53], P < 0.01), and hip circumference (MD = −3.54, 95% CI [−4.88, −2.19], P < 0.01) in cancer survivors. No significant differences were observed in energy intake (P = 0.46) or whole grain intake (P = 0.14). Conclusions MHealth diet interventions may be an effective strategy for cancer survivors. Large-scale RCTs with rigorous study designs are needed to examine the effect of diet intervention delivered via mHealth.


Introduction
In recent years, the number of cancer survivors has substantially grown. 1 Early detection of cancer and improvements in treatments have enhanced survival rates for cancer survivors. 2 However, the diagnosis and treatment of cancer impose physical and psychological burdens on cancer survivors. 3,4 In particular, physiological changes associated with the tumor (such as malabsorption, obstruction, and diarrhea), the host response to the tumor (anorexia and altered metabolism), and the side effects of cancer treatment (nausea and loss of appetite) can lead to poor nutritional status in cancer survivors [5][6][7] . Research indicates that a healthy diet is associated with better clinical outcomes, quality of life (QoL), and overall survival in cancer survivors [8][9][10] . The goals of diet interventions include improving nutritional status, achieving and maintaining a healthy weight, and improving the QoL of cancer survivors. 11,12 Hence, it is important to provide effective diet interventions to cancer survivors to improve their prognosis.
Mobile health (mHealth) is defined as a "medical and public health practice supported by mobile devices" 13 and is widely used for diet interventions in cancer survivors. MHealth is delivered through various tools, including phone calls, text messages, websites, and applications to provide tailored information, frequent interactions, and high-quality medical services. 14,15 MHealth thus provides increased ubiquity of interventions, with the ability to reach users at nearly any time or place to deliver timely feedback. 16 Studies have shown that mHealth diet interventions can change patients' health behavior and may be efficacious in cancer survivors. 17,18 There is considerable debate regarding the value of mHealth nutritional interventions in cancer survivors. Most of the previous studies that evaluated the effectiveness of mHealth diet interventions in cancer survivors and evaluated a positive effect 19,20 ; however, others studies did not. 21,22 Thus, mHealth dietary interventions have not been conclusively shown to improve the diet intake of cancer survivors, and it remains unclear which intervention components are effective.
Previous systematic reviews focused on the ability of mHealth interventions to increase physical activity or improve psychological symptoms in cancer survivors. 23,24 Another systematic narrative review evaluated electronic health interventions delivered to cancer survivors as well as changes in diet behavior and QoL, but they did not conduct a quantitative meta-analysis. 25 The effect of mHealth diet interventions on diet intake, body composition, and QoL has not been examined. Therefore, this meta-analysis aimed to (1) use available evidence to quantitatively determine the effects of mHealth diet interventions, including diet intake, weight change, waist circumference, hip circumference, and QoL, in cancer survivors, and (2) provide a scientific basis for the clinical application of mHealth diet interventions in cancer survivors.

Methods
The methods adhered to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement. 26

Inclusion criteria
Studies that met the following criteria were included: (1) Population: Patients who were cancer survivors (any type or stage) and aged above 18 years. (2) Intervention: mHealth diet interventions using phone calls, text messages, websites, or mobile applications to improve cancer survivors' diet intake. (3) Control groups that received the usual care, waitlist, and printed materials. (4) Outcomes: The outcome includes diet intake, weight change, waist circumference, hip circumference, or QoL. (5) Study: The research design was a randomized controlled trial (RCT).

Exclusion criteria
We excluded studies that (1) were research protocols, conference abstracts, comments, or case reports; (2) provided mHealth interventions (sending messages through mHealth technology, such as phone, emails, or newsletters) to the control group; (3) without adequate data for outcome analysis; or (4) were published in languages other than English or Chinese.

Search strategy
A comprehensive search of the literature was performed in nine electronic databases, including PubMed, Embase, Web of Science, Cochrane Library, Scopus, ProQuest, China National Knowledge Infrastructure, Wanfang, and SinoMed databases from inception until September 25, 2022. We also reviewed the references of the selected publications. The search strategies involved a combination of Medical Subject Headings (MeSH) terms and free text terms. The search strategy was as follows: (neoplasm OR tumor OR cancer) AND (mHealth OR mobile applications OR telephone) AND (diet OR diets OR dietary intake) AND (randomized controlled trial OR randomised controlled trial). The list of search strategies for the nine databases are presented in Supplementary Table S1.

Screening and data extraction
All records were imported to Endnote X9, and duplicates were removed. Two reviewers (GYB and JXH) independently screened the titles and abstracts of identified records. The full texts of selected publications were assessed in detail according to the inclusion criteria by two independent reviewers (GYB and JXH) in the Endnote X9 library. Full texts that did not meet the inclusion criteria were excluded. In cases of disagreement, a third reviewer was consulted (PJS), and a consensus was reached through discussion. If data on the outcome were missing, we contacted the authors by email to request these data.
Researchers extracted data using Microsoft Excel software. The extracted details included the following: (1) general information (eg, the first author's name, publication year, study design, and the country where the research was carried out); (2) characteristics of the cancer patient population (eg, participant demographic characteristics, diagnoses, and sample size); (3) intervention characteristics (eg, intervention type, interveners, mHealth diet intervention, frequency, duration, and details); and (4) outcome indicators and measurement tools.

Quality assessment
Risk of bias of the included RCTs was assessed by two reviewers (GYB and JXH) using the Revised Cochrane risk-of-bias tool for randomized trials (RoB2). 27 Five domains were evaluated, including the risk of bias from the randomization process, deviations from intended interventions, missing outcome data, measurement of the outcome, and selection of the reported result. Bias in selection of the reported result is a type of reporting bias. Each study was classified in each domain was divided into low risk of bias, high risk of bias or some concerns (Fig. 2). When there were disagreements, the third reviewer (PJS) was consulted, and a consensus was reached through discussion.

Data analysis
The meta-analysis was conducted using RevMan 5.4.1 software. Stata version 17.0 was used to perform funnel chart analyses and Egger's regression test. P < 0.05 was taken to indicate significant publication bias. 28 Standardized mean differences (SMDs) or mean differences (MDs) were calculated. MDs were used to calculate results obtained using the same measurement method. If different scales were used to evaluate the same outcome, SMDs were used. The effect size was calculated using the 95% confidence interval (CI). The heterogeneity was determined by the Q test and I 2 statistics. In addition, for homogeneous data sets, P > 0.1 and I 2 < 50% were used as the test thresholds. When these two statistical conditions were met, a fixed-effects model was used for the meta-analysis because the pooled effect sizes were relatively homogenous. If one of the above conditions was not met, a random-effects model and sensitivity analysis were applied to control heterogeneity. 29,30 P < 0.05 was considered statistically significant.

Study selection
As shown in Fig. 1, a total of 2648 articles were initially retrieved. After duplicates were removed, a total of 2214 articles were screened at different stages. First, 2180 articles were removed after by reviewing their titles and abstracts; and further reviews were performed of the full texts of 34 articles. A total of 22 studies were excluded after full-text screening, and the main reasons for exclusion are shown in Fig. 1. In addition, we identified six relevant cited references and finally included three additional studies. In total, 15 articles were included in this study.

Study characteristics
After screening, a total of 15 studies involving 2363 participants were included (Table 1). All included articles were published between May 2009 and April 2021. These studies mainly involved patients with breast cancer, prostate cancer, or bladder cancer. The locations of the studies were the United States (10 studies), Australia (two studies), the Netherlands (one study), the United Kingdom (one study), and South Korea (one study). The mean age (standard deviation) was 61.24 years (17.08). The sample size of the studies ranged from 21 to 641 participants. All of the included studies compared mHealth interventions with usual care, wait-lists, or printed materials. Six studies delivered mHealth interventions by phone calls 19,22,31,33,34,41 , four studies by a website, 20,21,32,35 one study by a mobile app, 36 one study by emails, 37 one study by a combination of phone calls and a website, 38 one study by a combination of phone calls and Skype calls, 42 and another study by a combination of text messages, emails, and Facebook. 39 Nearly half of the interventionists were dietitians, 19,31,38 cancer survivors, 21 doctors in nutrition, 37 trained counselors, 40 health professionals and cancer survivors, 22 doctoral students in clinical psychology and medical students, 41 or fully automated expert systems without human involvement. 32 Interventions included healthier food choices, decreased energy intake, and reduced dietary component intake. The length of the interventions of the included studies ranged from 4 weeks to 24 months. The frequency of the interventions ranged from daily to 22 times over 24 months. The most common frequency and length of interventions were daily and 6 months, respectively.

Assessing risk of bias
The RoB2 was used to assess the risk of bias in RCTs. Four studies stated that random assignment was performed but did not elaborate at length. We did not obtain adequate information about the blinding procedures for participants, interventionists, or assessors form 13 studies. In the remaining studies, one study blinded outcome assessors, and another blinded outcome assessors and interventionists, but participants were not blinded to their group allocation. Three studies were identified as having a potential risk of bias because of missing outcome data. Twelve studies have some concerns due to missing protocols. In our study, one study was categorized as having a low risk of bias, 11 studies had some risk of bias, and three studies had a high risk of bias (Fig. 2).

Effects of mHealth diet interventions on fruit and vegetable intake
Six studies, 19,21,32,33,36,38 involving 1456 patients evaluated fruit and vegetable intake. Fruit and vegetable intake was measured by using 120-item food frequency questionnaire, Block Food Frequency Questionnaire, 8 items of the Dutch Standard Questionnaire on Food Consumption, interactive NDSR software, The National Health Interview Survey 2000, 3 separate 24-h dietary recalls. A random effect model was used because of high heterogeneity (I 2 ¼ 61%, P ¼ 0.03; Fig. 5a). To reduce this significant heterogeneity, a sensitivity analysis was conducted by sequentially excluding individual papers. After one 33 study was removed, no significant heterogeneity was observed among the remaining five studies (I 2 ¼ 0%, P ¼ 0.47; Fig. 5b). The fixed-effects model showed that there was a positive effect of mHealth diet interventions on fruit and vegetable intake (SMD ¼ 0.19, 95% CI [0.05, 0.33], P < 0.01; Fig. 5b).

Effects of mHealth diet interventions on whole grain intake
Two studies, 35,40 involving 85 patients, evaluated whole grain intake. Whole grain intake was measured by using National Cancer Institute's Automated Self-administered Dietary Assessment Tool (ASA24) and NDS-R. The results showed that mHealth diet interventions had no significant effect on whole grain intake compared with the control conditions (MD ¼ 0.34, 95% CI [À0.12, 0.80], P ¼ 0.14; Fig. 6). A random effect model was used because of the high heterogeneity (I 2 ¼ 56%, P ¼ 0.13; Fig. 6).

Effects of mHealth diet interventions on QoL
Six articles, 22

Publication bias
Visual inspection of the funnel plot and the Egger test (P ¼ 0.564, P ¼ 0.948, P ¼ 0.592 ) identified no publication bias (Fig. 11, Fig. 12,  Fig. 13). Since the number of included studies for other outcomes was too small (n < 5), funnel plots were not constructed.

Discussion
To the best of our knowledge, this is the first systematic review and meta-analysis of the effects of mHealth diet interventions on cancer survivors. The results indicated that there was significant improvement in vegetable and fruit intake and QoL, reductions in fat intake, weight, waist circumference, and hip circumference, and no significant changes in energy intake or whole grain intake in cancer survivors after mHealth diet interventions.
The results show that the mHealth diet intervention increased fruit and vegetable intake and reduced fat intake in cancer survivors. Nutrition guidelines emphasize the importance of personalized nutrition counseling to help cancer survivors develop healthy eating habits including increased fruit and vegetable intake and reduced fat intake. 11,43 Our findings were consistent with the recommendations of these guidelines. A variety of studies have tested the effects of nutritional interventions, including nutrition clinics, nutrition education, and printed materials on developing healthy eating habits in cancer survivors, but these studies have observed only small benefits. 44,45 Our study found that mHealth diet interventions had an obvious effect on cancer survivors' healthy eating habits. The superiority of mHealth diet intervention may be due to mHealth enabling long-term and individualized intervention. Patients could set specific, achievable goals and receive relevant information, such as individually tailored dietary advice, individual guidance to achieve diet goals, and tailored progress reports. The mHealth diet intervention also increased cancer survivors' interactions with the medical staff. Therefore, cancer survivors were more likely to change their diet behaviors and develop healthy eating habits. mHealth interventions had no effect on energy intake in cancer survivors. This finding was consistent with the results of other studies. 46,47 Further well-designed RCTs with larger sample sizes are necessary to evaluate the effects of mHealth diet interventions on energy intake in cancer survivors.
MHealth diet interventions had no effect on whole grain intake in cancer survivors. The two studies included in this analysis both concluded that mHealth diet interventions could improve whole grain intake in cancer survivors, but after merging these studies in our metaanalysis, we obtained the opposite result. This was due to the use of a random-effects model to account for the high heterogeneity in data due to the far better outcomes of Van Blarigan's study compared to Parsons's. However, we could obtain a positive result by using a fixed-effects model.  In our opinion, mHealth diet interventions likely benefit whole grain intake in cancer survivors, since the two studies were well designed. However, further prospective studies are needed to confirm this finding.
The results of our meta-analysis suggested that mHealth diet interventions can significantly reduce weight, waist circumference, and hip circumference. This may be due to the individualized health diet intervention. At present, targeted diet interventions focus on either weight loss or encouraging weight maintenance in cancer survivors. 48 The most common topics in mHealth nutritional interventions were related to increasing fruit and vegetable intake and whole grain intake. 49 Increased fruit and vegetable intake were similarly associated with reduced weight. 50,51 Diet intervention combined with oral liquid nutritional supplements has an effect on encouraging weight maintenance in cancer survivors who experience weight loss or even cachexia. 52,53 MHealth diet interventions can help cancer survivors change their diet behaviors and develop healthy eating habits. Therefore, mHealth diet intervention merits clinical use and should be regarded as a crucial component of comprehensive nutrition support for cancer survivors who experience weight loss or even cachexia. Reductions in waist circumference and hip circumference are related to weight loss. 54,55 In addition, some studies have suggested that nutritional interventions should be combined with exercise to improve weight change in cancer survivors. 42,48 However, in this study, we explored only the effects of mHealth diet interventions on cancer survivors.
In addition, mHealth diet interventions improved QoL in cancer survivors. There are possible reasons for this increase QoL. First, mHealth diet interventions increased interactions between health professionals and cancer survivors. Perceived increases in contact with health care providers is an important determinant of QoL of cancer survivors. 56 Second, cancer survivors must be aware in many situations to adhere to a healthy diet. 57 Changes in diet behaviors have a positive effect on nutritional intake, which may improve the body composition and symptom experience of cancer survivors. 58,59 A study by Jones revealed that cancer survivors are particularly concerned with their nutritional status. 60 Finally, diet significantly affects the mood of survivors. Healthy diet behaviors lead to more positive emotions. 61 Therefore, nutrition delivered via mHealth interventions is well accepted by cancer survivors allowing them to improve their QoL.
Based on the findings of this study, we recommend using individualized nutritional interventions delivered via phone calls, text messages, and mobile apps could improve the fruit and vegetable intake, and QoL, of cancer survivors. The most recommended frequency and length of mHealth diet interventions were weekly and 6 months, respectively. MHealth interventions that combine nutrition with exercise are expected to benefit weight change in cancer survivors. Cancer survivors, academicians, and health professionals could benefit from the results of this study. Health professionals could obtain the methods for mHealth diet interventions. MHealth diet interventions can provide remote access to individual nutritional interventions to cancer survivors that change their diet behaviors. This research can guide future researchers to design interventions that benefit cancer survivors with more economical methods and improved outcomes.

Limitations
There were several limitations of this meta-analysis. First, all included studies were published in English. Due to language restrictions, some articles may have been omitted. Second, this meta-analysis mainly focused on the effectiveness of mHealth interventions on diet intake; thus, one limitation of this meta-analysis is that other behaviors that may also contribute to weight change (eg, activity) were not included. Third, this study was not registered in the International Prospective Register of Systematic Reviews (PROSPERO). Fourth, we did not perform subgroup     analysis and thus could not determine the most effective mHealth diet interventions. These limitations mean that our results should be interpreted with caution.

Conclusions
Our study provides preliminary data for the future development and application of mHealth diet interventions in cancer survivors. This systematic review and meta-analysis showed that mHealth diet interventions improve fruit and vegetable intake, and QoL and significantly reduce fat intake, weight, waist circumference, and hip circumference in cancer survivors. However, there was insufficient evidence regarding the effects of these interventions on energy intake or whole grain intake. This study raises awareness of mHealth diet interventions and encourages health professionals to implement them to improve the diet of cancer survivors. Junsheng PENG: Writing -Review and Editing, Project Administration, Funding Acquisition. All authors had full access to all the data in the study, and the corresponding author had final responsibility for the decision to submit for publication. The corresponding author attests that all listed authors meet authorship criteria and that no others meeting the criteria have been omitted.

Declaration of competing interest
The authors declare no conflict of interest.

Funding
Research Fund of the Sixth Affiliated Hospital of Sun Yat-sen University (Grant No. P20200217202309876). The funders had no role in considering the study design or in the collection, analysis. Interpretation of data, writing of the report, or decision to submit the article for publication. They do not receive compensation from pharmaceutical companies or other than those listed above, and the authors declare no conflict of interest.

Ethics statement
Not required.

Data availability statement
Data availability is not applicable to this article as no new data were created or analyzed in this study.