Analyzing the Effectiveness of mHealth to Manage Diabetes Mellitus Among Adults Over 50: A Systematic Literature Review

Plain Language Summary

This systematic review analyzes studies published over the last 2.5 years in academic, peer-reviewed literature to identify mHealth-related interventions to manage Type 2 Diabetes Mellitus. Through data extraction on 30 articles from 11 countries, we identified five mHealth-related interventions. These interventions were found to improve symptoms of Type 2 Diabetes Mellitus, and many interventions show improvement over treatment as usual. Employing technologically-inclined interventions can help meet a preference of patients while still feeling confident about their efficacy. This study has implications for administrators, clinicians, and policy makers.

Introduction

Rationale

Telemedicine is defined as healing at a distance through the use of information and communication technologies.¹ Telemedicine encompasses a wide range of clinical interventions including mHealth and eHealth. The distinction between the latter and former is that the latter involves a computer as the interface while the former involves mobile technology such as a cellular phone or tablet. mHealth and eHealth are forms of telehealth. mHealth uses mobile technology to heal at a distance and eHealth uses computer based apps to do the same. These are closely related because most computer applications can be accessed through mobile devices. Mobile devices are convenient, and they are hosting medical-related applications to help patients manage many medical conditions.

It is important to note in studying alternative modalities of care that improvements that are not statistically different from treatment as usual is usually still a significant event to report. Some patients prefer technologically inclined interventions. When an mHealth intervention shows improvement that is close to treatment as usual, it means the provider can assign this intervention to the patient because it fulfils a preference and its efficacy is sound. The advantages inherent to technological interventions such as mHealth is that they allow healing at a distance. They help avoid miles driven and time expended to travel to a medical appointment. They prevent exposure to the medical environment.

Diabetes mellitus “is a heterogeneous group of disorders characterized by hyperglycemia due to an absolute or relative deficit in insulin production or action.”² This condition is associated with end organ damage, dysfunction, and failure of the retina, kidney, nervous system, heart, and blood vessels. The prevalence of this group of disorders is estimated by the International Diabetes Federation of 537 million in 2021 and predicted to rise to 643 million by 2030.³ About 90% of those with diabetes have Type 2 diabetes, which is lifestyle-related. This distinction is important because treatments for type 2 do not always work with type 1. Diabetes affects older people more than any other age group ranging from 22–33% of the population.⁴ With the aging of the population, diabetes in this age group should have priority because the prevalence will increase as the number of people entering this age bracket increases, thus causing a greater burden to the healthcare system. Diabetes doubles the risk of COVID-19 mortality to older adults.⁵

Type I diabetes is characterized by the destruction of the beta cells of the pancreas, leading to insulin deficiency in the patient.⁶ Treatment for type I diabetes includes insulin replacement therapy which meets specific glycemic targets.⁶ Type II diabetes is characterized by hyperglycemia, insulin resistance, and impaired insulin secretion.⁷ Treatment for type II diabetes includes educational measures, continuous evaluation for vascular complications, minimization of long-term risk factors, lifestyle changes, and attempts to achieve normoglycemia, often using insulin therapy.⁸ Resulting from the COVID-19 pandemic, those who suffered from the infection can suffer from type I or type II diabetes and are often seen with severe metabolic manifestations of diabetes which include diabetic ketoacidosis, hyperosmolar hyperglycemic states, and severe insulin resistance.^8,9

Telehealth has been used as an intervention to help monitor and manage diabetes for decades. It is used for telemonitoring, glucose recording, and communication with providers.¹⁰ The telehealth intervention is associated with convenience, access, and time savings for patients.¹¹ However, disadvantaged communities are often challenged with limited technological literacy, and telehealth in general is used less by people older than 65.¹² A marked increase in the use of telehealth is evidenced post pandemic, and telehealth is expected to continue to be a frequently utilized treatment modality.^13,14 Management of chronic conditions such as diabetes is increasingly being conducted via telehealth.^14–16 A descriptive systematic review was published in 2017 to explore the enablers and barriers faced by adults with diabetes using two-way information technologies to support diabetes self-management.¹⁷ A team of reviewers analyzed 48 articles and identified enablers (access to reliable technology, highly automated data entry, graphical display of data with immediate feedback, and supportive healthcare professionals and family members) and barriers (poorly designed interfaces and systems that lacked functionalities valued by patients).

A systematic review was published in 2021 that explored the effectiveness of technology-based psychosocial interventions on diabetes distress and health-relevant outcomes.¹⁸ This group of reviewers analyzed 20 randomized control trials to conclude that technology-based psychosocial interventions improved diabetes distress, self-efficacy and HbA1c levels with significant and small effect sizes.

Objectives

The purpose of this review is to analyze the effectiveness of mHealth to manage Type 2 diabetes mellitus self-care among adults greater than 50 years old from literature published over the last 2.5 years in peer-reviewed academic journals.

Material and Methods

Eligibility Criteria

The eligibility criteria for this review is older adults (over the age of 50), research articles published in the last 2.5 years in peer-reviewed, academic journals, using mixed method, quantitative, and qualitative, methods to analyze the relationship between mHealth as an intervention to manage the self-care of diabetes mellitus. Two and a half years was chosen due to the plethora of articles available on this topic, and to focus on research efforts since the beginning of the pandemic. In addition, this time frame produces articles incident to COVID-19, a period when telehealth tools were used ubiquitously in medicine due to the inability to meet face-to-face. In order to avoid confounding results, other systematic reviews were not included in the analysis because systematic reviews already reported on results from studies that may also be counted in our analysis. Articles should report results of studies. Grey literature was only considered if the articles reported results.

Information Sources

Four research databases were queried: PubMed (MEDLINE), CINAHL Ultimate, Web of Science, and Science Direct. They were searched between July 1st and July 20th 2022.

Search Strategy

We created a Boolean search string to combine key terms listed in the Medical Subject Headings (MeSH) of the US Library of Medicine. We used the same search strategy in all databases. The search string was (mhealth OR ehealth OR telemedicine OR smartphone) AND (diabetes OR “diabetes control”) AND (‘older adult’ OR elderly). We used similar filter strategies, because not all databases have the same tools. MEDLINE was excluded from all databases except PubMed to eliminate duplicates.

Selection Process

In accordance with the Kruse Protocol, we searched key terms in all databases, filtered results, and screened abstracts for applicability.¹⁹ At least two reviewers screen each abstract and analyzed each study. Three consensus meetings were held to determine which articles would be analyzed, which data-extraction items were significant measurements of effectiveness, and what observations should become themes. Results were reported in accordance with the PRISMA 2020 standard.²⁰

Data Collection Process

We used a standardized Excel spreadsheet to extract data, and we collected additional data at each step of the process. This spreadsheet was standardized in the Kruse Protocol and has developed over a decade of use to collect data useful for clinicians, administrators, and policy makers.¹⁹ We used a series of three consensus meetings to screen abstracts, identify articles for analysis, and identify themes through narrative analysis.²¹

Data Items

In accordance with the Kruse Protocol, we collected the following fields of data at each step: Participants, experimental intervention (mHealth), results compared with a control, medical outcomes commensurate with the intervention, study design, sample size, bias identified in the study, effect size, country of origin, statistics used, patient satisfaction, quality associated with the intervention (effectiveness), barriers to adoption, strength of evidence and quality of evidence.

Study Risk of Bias Assessment

We observed bias and assessed the quality of each study using the John’s Hopkins Nursing Evidence Based Practice tool (JHNEBP).²² We considered the instances of bias in how to interpret the results because bias can limit external validity.²³

Effect Measures

Because we chose to accept mixed methods and qualitative studies, we were unable to standardize summary measures, as would be performed in a meta-analysis. We reported measures of effect in tables for those studies in which it was reported. The odds ratio was the preferred measure of effect, but we also collected measures reported as Cohen’s d. These were tabulated and reported.

Synthesis Methods

A thematic analysis was performed after data extraction was completed.²¹ This procedure helps makes sense of the data collected through data extraction. Although this technique is often used in qualitative research, its use in quantitative research is well established in the literature.^24–26

Reporting Bias Assessment

The overall ratings of quality from the JHNEBP provide us with an assessment of the applicability of the cumulative evidence by identifying the strength and quality of evidence. Each reviewer recorded observations of bias on the standardized spreadsheet.

Additional Analyses and Certainty Assessment

We performed a narrative analysis of the observations to convert them into themes.²¹ We calculated frequency of occurrence and reported these in affinity matrices. The frequency does not imply importance: only the probability of occurrence in the group of articles for analysis. The reported frequency also provides confidence in the data analyzed.

Results

Study Selection

Figure 1 illustrates the article selection process, to include the inclusion and exclusion criteria from four databases plus one targeted article search. A kappa statistic was calculated based on the level of agreement between authors, (k=0.99, near perfect agreement).^27,28 The initial search yielded 113,698 results. A series of filters (full text, human subjects, English language, peer-reviewed, academic journals, and exclude reviews), abstract screening, and a brief full article examination reduced the final group to 30. About 42% of those rejected were due to the age of the participants. These 30 studies generated from 11 different countries, but most were from the US (14/30, 47%), China (4/30, 13%), the UK (2/30, 7%), Korea (2/30, 7%), and Bangladesh (2/30, 7%). Eastern and Western medicines are different in their approach, but they were included to provide a comprehensive overview of interventions and their level of success.

Figure 1 PRISMA diagram—article selection process. Notes: PRISMA figure adapted from Page MJ, McKenzie JE, Bossuyt PM et al. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ. 2021;372:n71. Creative Commons.20

Study Characteristics

Following the PRISMA (2020) checklist, Table 1 was created to summarize the characteristics of each article analyzed (participants, intervention, comparison, outcomes, study design, PICOS). Of the 30 articles analyzed, over the 2.5 year period, 100% were older adults over 50, 100% implemented a form of mHealth for the intervention (mHealth app 37%, mHealth Short Message Service (SMS) 27%, mHealth plus remote monitoring 17%, mHealth plus telemedicine 13%, mHealth app plus SMS 7%). Of the 30 studies, 15 (50%) were randomized control trials (RCT), 6 (20%) were qualitative, 3 (10%) were observational and 3 were quasi-experimental, 2 (7%) were true experiments, and 1 was a focus group. Slightly less than half originated in the United States (14/30, 47%). Studies are listed in chronological order and alphabetically: 2020 (n=19),^29–47 2021 (n=8),^48–55 2022 (n=3).^56–58 For each study, we extracted the following data fields: participants, intervention, comparison (to control or other group), medical outcomes, study design, sample size, bias within study, country of origin, statistics used, patient satisfaction, quality associated with intervention, barriers to adoption, strength of evidence, and quality of evidence. This was performed as a way to summarize study characteristics in a manner established in the literature. Results varied across studies. Twenty-five out of 30 (83%) articles found an improvement in at least one area studied, (eg, HbA1C), but the improvement was not always statistically significant. While mHealth apps were effective at helping diabetes patients manage their disease, some apps required manual entry of data, which discouraged users from using it. The intervention of mHealth SMS showed consistently improvement across multiple focus areas: BMI, weight loss, exercise, diet, disease awareness, HbA1C, and healthy behaviors. Similar results were also observed with mHealth plus telemedicine coaching. Additional explanation of the results will be provided below.

Table 1 PICOS

Risk and Reporting of Bias in Studies

Reviewers used the JHNEBP quality assessment tool identified the following the strength and quality of evidence. Of the 30 articles analyzed, 18/30 (60%) were strength I (the highest rating) because they were either RCTs or true experiments. 10/30 (33%) were type III (qualitative or observational) and 2/30 (7%) were type II (quasi-experimental). Most studies used large samples (average 1022) reported consistent results with definitive conclusions, and used adequate controls, so the quality of evidence was strong, or Type A (23/30, 77%) and the remainder used smaller but adequate samples (7/30, 23%). Studies of type C or strength below III were not used for analysis. Reviewers also made note of bias within and among the studies. The most common bias was selection bias (27/30, 90%) because most studies used convenience samples in one city or country. The second most common was sample bias (8/30, 27%) because the sample was a majority of one gender or race. Selection and sample bias affect the external and internal validity of studies, respectively.

Results of Individual Studies

Reviewers independently analyzed each article and recorded independent observations commensurate with the objective statement. A thematic analysis was conducted as part of sense making.²¹ Each observation that occurred more than once became a theme. Themes were used to summarize the observations, but they did not always match the observation completely. A summary of the themes is tabulated in Table 2. Articles are listed in chronological order. Appendices 1A and B show an observation-to-theme match. Appendix 1c tabulates the additional data extracted from the literature. Additional analysis of the findings is provided below.

Table 2 Summary of Analysis, Sorted Chronologically by Author

Results of Syntheses, Additional Analysis and Certainty of Evidence

The thematic analysis made sense of the data, and themes were tabulated into affinity matrices for analysis. While frequency does not imply importance, it does provide a probability of occurrence in the group of articles for analysis.

When mHealth apps were utilized as the intervention, the most common patient satisfaction theme was that patients were satisfied with the intervention, however, many were discouraged when the app required manual input of data. When the mHealth apps were coupled with wearable sensors, and the data from the sensors were automatically entered into the app, the observation about manual entry went away. These interventions showed improvement in at least one area of observation, but the results were not as strong as other interventions. When mHealth SMS was utilized as the intervention, patients noted that the messages helped them improve their self-care habits. The results, medical outcomes, and quality observations were strong with this intervention. However, when the mHealth SMS was coupled with telemedicine visits to offer coaching, the results were the strongest. Patients developed good self-care behavior, improved in weight loss, diet, exercise, blood glucose levels, cholesterol, blood pressure, and foot care. This intervention showed the strongest results.

Three sections of themes are similar, but they were created for different audiences. The section “results” is generic, and it includes results of studies (compared to a control group, where appropriate) that are both administrative, clinical, and patient-centered. The section “medical outcomes” is designed for the practitioner. The section “Quality” is designed for the administrator. While there are some repeats in these three sections, they are tailored for the specific audiences and some verbiage changes for them.

Results of Studies, Compared with the Control Group

Table 3 tabulates the Result themes and individual observations. Reviewers made some general observations. In 25/30 (83%) articles, an improvement in at least one area was observed, but the improvement was not always statistically significant. This accounted for 25/107 (23%) of the observed themes or observations.^{29,30,32–37,39,41–43,45–51,53–58} Improved self-management behaviors appeared in 14/107 (13%) of the occurrences.^{29,32–35,42,43,45,47,49,51,54,55,57} Patients who used mHealth as interventions in their care demonstrated improved blood glucose level (BGL)/HbA1C, included fasting blood glucose (FBG). This theme appeared in 15/107 (14%) of the occurrences.^{29,32,37–39,41,43,45,47,48,50,51,54,56,58} The theme “improved exercise” appeared in 10/107 (9%) of the occurrences. The themes “improved diet”^{36–39,47,48} and “improved weight loss”^{34,39,41,42,58} each appeared in 6/107 (6%) of the occurrences. The theme “increased disease awareness” appeared in 4/107 (4%) of the occurrences. Improved BMI appeared in 3/107 (3%) occurrences. The themes “increased communication”,^43,45 “improved blood pressure”,^56,58 “improved cholesterol”,^56,58 “improved treatment adherence”,^36,39 and “app identified patients at risk”,^31,44 each occurred in 2/107 (2%) occurrences. The following observations could not be categorized into themes: “increased access to support”, “improved cardiopulmonary endurance”, and “improved foot care.”^33,48,53

Table 3 Results Themes and Observations

Medical Outcomes Commensurate with the Intervention

Table 4 tabulates the medical outcome themes and observations. Like in the results themes, reviewers made some general observations: the use of mHealth interventions showed improvements in at least one clinical area in 24/30 articles (60%),^{29,31,32,34–37,39,41–43,45–51,53–58} and there was an improvement in self-management in 15/30 articles (50%).^{29,32–35,39,42,43,45,47,49,51,54,55,57} The theme “improved blood glucose/HbA1C” appeared 15/100 (15%) of the occurrences.^{29,32,37–39,41,43,45,47,48,50,51,54,56,58} The theme “improved physical functioning” occurred 9/100 (9%) occurrences.^{32,36–39,42,47,48,50} Improved diet^{36–39,47,48} and increased disease awareness^{31,32,36,42,46,47} each appeared 6/100 (9%) occurrences. Improved weight loss occurred 5/100 (5%) of the occurrences.^{34,39,41,42,58} Improved BMI occurred 3/100 (3%) of the occurrences.^41,53,56 The themes “improved cholesterol”,^56,58 “improved smoking cessation”,^39,47 and “improved blood pressure”,^56,58 each occurred 2/100 (2%) of the occurrences. Finally, three observations could not be categorized into themes: “improved medication adherence”, “improved foot care”, and “improved cardiopulmonary endurance.”^39,48,53

Table 4 Medical Outcomes Themes and Observations

Quality Themes and Observations Commensurate with the Intervention

Table 5 tabulates the Quality themes and observations. Of these observations, the “increase in self-management skills” is the most significant, appearing 21/80 (26%) of the occurrences.^{29,32–39,43–46,50,51,54,55,57,58} Improved blood glucose/HbA1C appeared in 15/80 (19%) of the occurrences.^{29,32,37–39,41,43,45,47,48,50,51,54,56,58} Improved exercise^{32,36–39,42,47,48,50} appeared 9/100 (9%), and improved weight management^{34,39,41,42,53,56,58} appeared 8/100 (8%) of the occurrences. Improved diet^{36–39,47,48} appeared 6/100 (6%) and increased education of disease^{32,33,36,42,47} occurred 5/100 (5%) of the occurrences. The following themes each appeared 2/100 (2%) of the occurrences: “increased medication adherence”,^36,39 “improved smoking cessation”,^39,47 “improved blood pressure”,^56,58 “improved cholesterol”,^56,58 and “increased connectedness.” The following observations could not be categorized into a theme: “improved foot care”,⁴⁸ “increased monitoring”, “greater screening for at-risk patients”, and “increase in cardiopulmonary endurance.”^30,31,48,53

Table 5 Quality Themes and Observations

The most commonly identified theme for patient satisfaction was that patients/participants were satisfied with the functionality of mHealth. This theme appeared 26/31 (84%) of the occurrences.^{29,32,33,35–39,41–58} Four other observations were identified, but they could not be fit into themes: “Activity on app increased for those recently diagnosed with diabetes”, “users were dissatisfied with usefulness”, “travel savings were greatly appreciated because the mHealth app prevented them from driving into the clinic”, and “users requested larger fonts and simplified data visualizations.”^30,34,40,52

Table 6 tabulates the themes and observations associated with barriers to adoption of mHealth interventions for the management of diabetes. The most common themes identified were associated with the necessity to acquire a mobile device in order to use it as an intervention. This includes the cost of acquisition and the training of the users. Teaching users occurred the most often, 24/76 (32%).^{29,31,34,37–48,50–58} Cost of device appeared 22/76 (29%) of the occurrences.^{35,36,38,39,41–48,51,52,54–58} Must own a text-enabled cell phone^{18,35–39,42,43,45,46,48} or a smart cell phone each occurred 11/76 (14%).^{41,47,50–58} Manual entry of data discourages use of the app appeared 2/76 (3%) of the occurrences.^30,49 The following observations could not be categorized into a theme: “Disinterest of users”, “inconvenience of data entry”, “display and presentation issues”, “multiple devices complicates monitoring”, and nursing activities increase as a result of the intervention.”^32,40,49,52

Table 6 Themes and Observations for Barriers to the Adoption of mHealth Interventions

Discussion

Summary of Evidence

This research found that 83% of the articles analyzed reported a positive outcome in at least one area of focus.^{29,30,32–37,39,41–43,45–51,53–58} mHealth SMS helps patients develop health habits and increase their level of self-care.^{35,36,38,42,43,46–49,59} The most significant finding of this research is that mHealth SMS coupled with telemedicine for coaching yields positive results for BMI, weight loss, exercise, foot care, diet, and overall self-care.^40,54,56,58

Healthcare providers can leverage mHealth SMS services to help diabetes patients create health habits and improve self-care for older adults. Coupled with telemedicine visits, provider teams can coach diabetes patients into long-term, self-care habits and decrease mortality incident to the disease. mHealth SMS coupled with telemedicine not only serves as a preventative for future diabetes-related ailments, but it also increases the health and longevity of patients who have suffered with diabetes for years.

Administrators can view the expense of mHealth SMS and telemedicine interventions as both preventative and ongoing maintenance of diabetes. These measures greatly increase the self-management of diabetes patients. Screening tools can be employed to identify at-risk patients and begin to change their habits early.

Developers should note the continued observations about manual entry, small fonts, and complicated data visualizations. Manual entry of data discourages users from using the app.⁴⁹ mHealth applications should fully interface with the wearable sensors and transmit data to the provider team and electronic health record. Fonts must be large enough for older adults to comfortably view, and data visualizations should be simple enough to provide immediate feedback and, if necessary, compel action.⁵²

Future research should explore mHealth combinations more thoroughly. Does mHealth apps plus telemedicine coaching yield as strong of results? Do mHealth apps, SMS, and telemedicine yield strong results? Also, cohort studies would be helpful to determine if the change in self-management is transitory or long term. This study analyzed Type 2 diabetes, instead of Type 1. Future research should focus on Type 1 diabetes. Historically, Type 1 diabetes has only comprised less than 10% of the diabetes population, but COVID-19 has caused a large surge in this population. Type 1 diabetes has always been a significant disease to study, and COVID-19 has raised its importance.

Limitations

A limitation of this review is the short time-frame chosen for publication of articles. We chose 2.5 years due to the plethora of results that occurred in our initial search using 5 or 10 years. Analyzing additional articles may have yielded new results or additional iterations of the same results. However, technology advances rapidly, and more recent articles will reflect current technology and capabilities of the same. Additionally, our search string focused on MeSH terms (with the exception of “smartphone”), which might have omitted studies not indexed with MeSH hierarchy. The term “smartphone” was used to control for this issue, however, we may still have not identified all studies.

Including grey literature may have also yielded different results. Our team chose to omit grey literature because opinions are not always based on study results. We chose to include research studies with strong designs. As a result of that decision, half of the articles analyzed in this review were RCTs.

Conclusion

The broad implication of this study is that mHealth SMS helps educate persons with diabetes and when coupled with telemedicine coaching, it yields very strong clinical and administrative results. These particular interventions are not costly, do not adversely affect patient satisfaction, and may result in fewer hospitalizations for diabetes related ailments. Most people own text-enabled phones, but not as many own smart phones. Unless mHealth apps fully interface with remote sensors and transmit data to care teams, patients will lose patience with any manual-entry and will abandon it. SMS, however, provides quick, simple education and reminders to help develop health habits and encourage self-care.

Other Information

This review is conducted in accordance with the Kruse Protocol for writing a systematic review.¹⁹ It is reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analysis.²⁰ This review is registered with PROSPERO: registration number CRD42021266568.