Our literature search was conducted in accordance with the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines [25]. We searched for published literature in electronic literature databases PubMed/Medline, The Cochrane Library, EMBASE, CINAHL Plus, Web of Science, Global Health, Scopus, and Library & Information Science Source for unpublished studies, using ClinicalTrials.gov.

A research librarian performed searches for English-language articles on human adults in scientific journals (aged ≥18 years), which combined terms for (1) minority population terms, (2) mHealth intervention terms, and (3) chronic condition terms for type 2 diabetes mellitus and obesity. The last search was conducted on September 23, 2020. Searches were limited by study design to controlled trials. Multimedia Appendix 1 provides full details regarding the search terms. Studies were included if they met the eligibility criteria (Table 1). Selected studies were either randomized controlled trials or nonrandomized controlled trials, and study participants had to be either African American or Hispanic individuals in the United States. Studies were also included if at least 30% of participants were African American adults, Hispanic adults, or a mixture of both minority groups. The intervention was any mobile or web-based health intervention including cellular phone calls, text messaging (SMS or MMS), and web-based applications or downloadable mobile apps that targeted management for diabetes and obesity.

Studies with any of the aforementioned interventions in combination with other types of activities such as keeping a journal of meals or extra group or one-on-one coaching were also included. Studies that focused on patients with diabetes or overweight with other health conditions as a group such as stroke, pregnancy, or depression were excluded from this study. There was no limit on the publication time, duration of the intervention, or participants’ age. For this review, primary outcomes of interest were objective measures related to obesity and diabetes, including BMI, weight change, waist circumference, or hemoglobin A_1c, which is a standard measure of glycemic control in patients with diabetes.

Two reviewers (CE and MH) extracted the data, and each reviewer independently extracting data from all studies. Extracted elements included the following: publication date, authors, study aims and objectives, study design, number of participants in the intervention and control arms, components and duration of the intervention, and results of the study. Outcome measures were extracted at all timepoints for studies that included multiple assessments. Reviewers checked each other’s extractions for accuracy and completeness. Discrepancies were discussed until arrival at an agreement and when necessary, a third reviewer was involved.

Each included study was assessed for risk of bias, using the Cochrane tool for risk of bias in randomized trials [26]. The following domains were assessed for risk of bias: selection bias (including the method of randomization and allocation concealment), detection bias, attrition bias, and reporting bias. Studies were rated as low risk, high risk, or unclear risk. Unclear risk of bias was assigned to a study if there was uncertainty or lack of information.

We summarized our findings for each question in tabular and narrative format. We did not conduct a meta-analysis owing to the limited number of studies that met the inclusion criteria. The overall strength of the evidence (SOE) was assessed as high, moderate, low, or insufficient on the basis of the overall risk of bias of included studies, the consistency of findings across studies, directness, precision, and other limitations [27].

The combined search strategies yielded 2358 electronic citations, which were screened to assess for eligibility (Figure 1). In total, 200 studies were found to be potentially eligible, and their full texts were obtained for further assessment. Of these, 7 studies met the eligibility criteria. The top 4 reasons for rejecting other studies were primarily unavailability of results or incompleteness of studies, incorrect patient population, incorrect intervention, and incorrect setting.

The primary aim of 5 studies was to evaluate the effectiveness of an mHealth intervention for weight loss or diabetes, while 2 studies included an evaluation of the feasibility and acceptability of the mHealth intervention as their primary aim. All included studies were randomized control trials [1,2,12,28-31]. Included studies focused on diabetes control (2 trials) [12,28], obesity and weight loss (4 trials) [2,29-31], or both diabetes and obesity (1 trial) [1].

The total number of participants in all studies was 942, with sample sizes ranging from 18 to 371. Three studies had all African American or Hispanic participants, and the other 3 studies had greater than 30% of participants who identified as African American or Hispanic minorities. Study durations ranged from 3 months to 12 months. Mean participant ages across studies ranged from 24 to 53 years. BMI of the participants ranged from 31.5 to 38.0. Mean hemoglobin A_1c values for the diabetes studies ranged from 9.02% to 10.2%.

All studies had mobile phones and SMS text messaging as the main mHealth device and medium of communication (Table 2). SMS text messages were sent to participants anywhere from 2-3 times a day to 3-4 times a week. Mobile technology was applied in several ways including medication reminders, prompts for blood glucose check, provision of motivational messages, education on healthy eating patterns, eating cues, and links to other resources and social support.

mHealth technology was also used to check on adherence to or progress on set physical activity and dietary goals.

In addition to SMS text messages, 3 studies that focused on obesity had additional activities as part of the intervention. One study included all elements of the Women, Infants, and Children (WIC) program [30], another included 15 minutes biweekly calls from health counselors [29], while the other had a face-to-face 1-hour session at the end of the study, which focused on problem-solving, progress assessment, and behavior change [31].

Obesity trials reported mean weight loss and weight change from baseline, percentage weight loss, and change in BMI. One study reported the proportion of participants who achieved a ≥5% weight loss and a ≥10% weight loss from baseline. The diabetes trial reported hemoglobin A_1c, weight, and BMI of participants at 3 months and 6 months.

With regard to the risk of bias of studies included, a randomization sequence generation process was reported by all studies. Bias for allocation concealment and selective outcome reporting were determined to be low for all studies. Only two studies provided information on blinding of personnel and outcome assessors [2,30]. One trial was determined to have a high risk of bias for incomplete data (up to 63.7% missing at 3 months and 41.1% at 6 months after the start of the intervention) [2,29] and another study was unable to assess differences in provider contact time between intervention and control groups [29].

Overall, 3 studies reported significant differences in weight loss between participants in the intervention group and those in the usual care group [2,29,30] (Table 3). Improvement in weight was noted at 3 months, 14 weeks, 6 months, and 12 months. Two studies found no significant difference in weight loss between groups [1,31].

For the studies that focused on diabetes, all reported an improvement in glycemic control in participants in intervention groups compared to those in control groups [1,12,28], but only one recorded a significant improvement at 6 months [1].

Overall, some evidence supports the efficacy of mHealth and web-based interventions for weight loss among obese African American and Hispanic adults (moderate SOE; Table 4).

We identified 7 trials that reported the benefit of mHealth interventions directed toward obesity and diabetes among African American and Hispanic adults. We found an association between receiving the mHealth intervention and weight loss compared to those in control groups, which suggests that mHealth and web-based interventions are effective for minority patients who are overweight. Of the three studies that addressed diabetes, only one reported an improvement in hemoglobin A_1c among those who received mHealth interventions compared to their control group counterparts, suggesting that the evidence may be unclear for the role of mHealth interventions in diabetes control among patients in ethnic minorities. This systematic review shows that mHealth and web-based interventions for 6 months to a year can be effective for weight loss in African American and Hispanic adults.

Our findings in support of mHealth interventions for weight loss among patients in ethnic minorities are similar to those of other studies [32-34]. A systematic review by Ryan et al [32] reported that mHealth intervention for weight loss is beneficial, while the systematic review and meta-analysis by Sorgente et al [33] was more specific in suggesting that web-based interventions are more effective than usual care or minimal intervention, but the evidence is conflicting with regard to the benefit of web-based interventions compared to non–web-based interventions. In our systematic review, of the 3 studies that reported significant weight loss in the intervention group, one of them had a control group that received usual care, while control group participants in the other 2 studies had comparable non-mHealth interventions. Beleigoli et al [34] suggest that web-based interventions only lead to a short-term weight loss but provide no evidence for weight loss over a long period of time. The longest study duration in our review was 12 months. Studies beyond 12 months, which met our inclusion criteria, were not found, and it is unclear if participants will remain in an intervention of this kind for longer than a period of a year. Changes in diet and activity patterns during holiday periods, the challenge or burden of adherence to self-monitoring over a long period of time, and the need for coaching in addition to self-monitoring have been suggested as reasons for the difficulty in maintaining the long-term effects of mHealth interventions for obesity and weight loss [35,36].

We also observed an improvement in hemoglobin A_1c with mHealth interventions compared to those with usual care. Our findings are similar to those of other studies that have shown improved glycemic control with mHealth interventions in the general population [12,14,36], except that only one study in this review showed a significant difference at 6 months. Another systematic review showed an improvement in 42% of studies in the review [37]. One major finding noted in the diabetes studies in this review was that the study with significant results incorporated culturally appropriate messages as part of their intervention [1]. Even with the improvements noted, it is unclear if outcomes would be significantly better if mHealth and web-based interventions have improved usability. African American and Hispanic adults have been found to be less adherent to interventions with technology [38,39]. Better outcomes have been noted in studies where the intervention was developed to be more appealing to the target minority population [40]. The lack of cultural consistency and application has been noted as a major challenge with mHealth and web-based interventions, which can influence results and outcomes [37,41]. Furthermore, some studies have shown that diabetes mHealth programs do not always meet health literacy guidelines, and they are not as user-friendly [42]. The design of some mHealth apps makes it challenging for some groups of people. For instance, Hispanic patients with diabetes who use apps are more likely to use certain functionalities such as medication and blood glucose diaries and less likely to use diaries developed for insulin and hemoglobin A_1c [43].

Another important finding from this study is the limited number of studies with African American and Hispanic adults in mHealth interventions for obesity and diabetes. The underrepresentation of participants from minority groups in research studies has been well-documented [44-46] and is reflective in this review. Several reasons have been cited for this, including barriers related to misconceptions or lack of information, mistrust, concerns about outcomes, stigmatization, insurance coverage, and immigration status in the United States [46-48]. More studies focused on mHealth interventions for African American and Hispanic adults with diabetes and obesity will augment the current evidence on the role of mHealth interventions for chronic disease management. This may also contribute to reducing the disparities gap that currently exist in the health care system.

The limited number of studies with primarily African American and Hispanic participants is a limitation of this study, which has also been observed in several other studies [21]. Another limitation is our study inclusion criteria. This review focused on African American and Hispanic participants and only included studies with ≥30% of the participants of interest. The team considered a set of less rigid inclusion criteria but decided against it since greater the proportion of African American or Hispanic participants, greater the ability to make conclusions specifically about the effectiveness of mHealth interventions on these racial groups in these studies. While all included studies involved mobile phones and SMS text messages, they differed in the content of the intervention, sample sizes, and study duration. The absence of studies with data beyond 12 months is also a limitation. Obesity and diabetes are chronic conditions that require long-term management and lifestyle modification. Furthermore, although some studies that met our inclusion criteria showed significant changes, it is unclear whether these translate to clinical significance.

Although the overall strength of the evidence was moderate for weight loss and low for diabetes, our findings suggest that mHealth and web-based interventions may provide a promising approach. mHealth technology can be used to meet different health needs such as the delivery of motivational messages, health education, prompts or reminders, and personal coaching [49]. Further studies are required to fully understand the role of mHealth interventions in chronic disease management in minority patients. These findings are significant because minority patients bear a heavy burden of diabetes and obesity in the United States. mHealth and web-based interventions have the potential to assist medical providers as they provide care to patients with chronic disease and improve health disparities. Further research to better understand how mHealth and web-based interventions can help overcome challenges or barriers to chronic disease management and will be very useful.