Descriptive Analysis of the Systematic Reviews
General Characteristics of Reviewed Papers
The 23 systematic reviews included (Multimedia Appendix 2) were published between 2009 and 2016 in 16 journals. The systematic reviews involved 371 studies. After verifying the sample size of each study, we found that at least 79,665 participants were included. In these reviews, systematic literature searches were performed from 1950 to April 2015 (see Multimedia Appendix 2).
Of the reviews, 17 included studies that were performed in low- and middle-income countries, 13 included studies in multiple settings, 6 specified particular settings, and 2 did not describe the setting. mHealth modalities described were mainly apps for chronic diseases, but also for disease management, treatment adherence, and changes in health behavior. Nine studies performed a meta-analysis [13-21].
Objective
The main objective of the reviews was to analyze the impact or effectiveness of mHealth interventions on chronic and noncommunicable diseases. Other focuses were to analyze mHealth in supporting chronic diseases management, health behavior change, attendance at appointments, disease and rehabilitation management, and the use of mHealth strategies by health workers.
Intervention
Different devices were used, including mobile phones, smartphones, personal digital assistants, MP3, phone plus app, medical device connected to phone by cord or wirelessly, and many others.
The most frequent intervention was SMS for reminders, education, motivation, or prevention. Sensors and point-of-care diagnosis, data collection, provider-provider communication, patient-provider communication, decision support, client education, provider work planning, training, protocol-based treatment, voicemail, videos, immediate physician feedback from a central location, cloud-based interactive voice response, disease management calls, disease monitoring, automated email to clinicians, treatment adherence, and phone counseling were also used.
These interventions were performed for smoking cessation, to increase physical activity, chronic disease management, chemotherapy-related symptoms monitoring, sexual health behavior safety, alcohol consumption reduction, medication adherence, appointment attendance, stress management and anxiety reduction, vaccination timeliness, prenatal support, reduction in emergency referrals or adverse events, health information access, cardiopulmonary resuscitation skills, patient satisfaction, and social functioning.
Hall et al [22] categorized 12 common applications: (1) client education and behavior change, (2) sensors and point-of-care diagnostics, (3) registries and vital events tracking, (4) data collection and reporting, (5) electronic health records, (6) electronic decision support: information, protocols, algorithms, checklists, (7) provider-provider communication: user groups and consultation, (8) provider work planning, (9) provider training and education, (10) human resource management, (11) supply chain management, and (12) financial transactions and incentives.
Multiple interventions were used on significantly varied targets, and the duration of follow-up varied from a few minutes to up to 24 months.
Control Group
The control group care was not clear in some reviews [16,23,24], but others were very specific.
Outcomes
The primary outcomes assessed were clinical outcomes (eg, frequency of hypoglycemic events, symptoms, deaths), surrogate outcomes (eg, glycated hemoglobin [HbA1c], blood pressure, lipid profile, cardiovascular disease risk profiles, lung function tests results, nebulizer use, weight, body mass index [BMI]), behavioral or lifestyle changes (eg, sexual behavior, smoking cessation, increase in physical activity), and processes of care (eg, attendance rates, compliance with medication taking, data management, communication performance, time to diagnosis, time to treatment, changes in professionals’ workload).
The secondary outcomes were cost, patient satisfaction, and potential harms and adverse effects.
Quality of Included Studies
Multimedia Appendix 3 shows the results of the quality assessment of the 23 systematic reviews. Ten were rated low quality (AMSTAR score 0-4), seven moderate quality (score 5-8), and six were high quality (score 9-11).
Regarding bias risk assessment, 10 reviews did not explicitly report on study quality assessment. Two specified that the risk of bias was mostly either low or unclear [25]. Free et al [18,19] related that only 4 trials were at low risk of bias in all areas. All studies included in Car et al [13] had high risk of bias.
Baron et al used the McMaster University quality assessment tool and reported that the overall quality was poor [26]. Fanning et al used the Guide to Community Preventative Services data extraction form and reported that 7 studies were rated “fair” and 4 were “good” [16]. Three reviews [27] used the Cochrane Handbook for Systematic Reviews of Interventions [28] reporting varying methodological quality, with some providing insufficient information.
Two reviews reported adequate sequence generation for randomization [20]. De Jongh et al [17] reported adequate sequence generation in 3 of 4 and that none of the included studies were clear if the allocation was concealed. A potential for bias occurred from the apparent lack of blinding of outcome assessors. In Vodopivec-Jamsek et al [27], allocation concealment was considered adequate in 2 studies and unclear in 2 studies. Only one study reported on blinding of personnel collecting and analyzing samples. No mention is made of blinding of outcome assessors or researchers, which could have introduced bias.
Incomplete data analysis methods varied, with analysis and reporting based on intention-to-treat analysis and on only participants who completed the study, which could influence generalizability of the findings. Substantial heterogeneity was detected across analyses; however, a post hoc decision to conduct the main analysis using a random-effects model resulted in no difference in the interpretation of findings. Bacigalupo et al [29] also used Cochrane guidance [28], reporting that 4 of 7 studies presented low risk of bias. The only studies in which Grading of Recommendations, Assessment, Development, and Evaluation (GRADE) methodology [30] was used were those published in the Cochrane Database of Systematic Reviews [13,15,17,20,27].
Main Results
The reviews show a variety of results, as expected, for using different devices and different mHealth interventions on different populations. The most widely used and successful intervention was SMS addressed to chronic disease patients. Positive impact was reported on clinical outcomes, adherence to treatment and care, health behavior changes, disease management, attendance rates, and others. However, some reviews showed conflicting results, with no significant differences between intervention and control groups.
Clinical Outcomes
Asthma
Positive impact has been demonstrated with moderate-quality evidence that text messaging interventions showed greater improvements in the pooled symptom score (mean difference 0.36, 95% CI -0.56 to -0.17) compared with the control group [17]. Very low-quality evidence (GRADE Very Low) showed the following: increased office visits in the SMS group, whereas increased hospital admissions for the control group [17]; a reduction in hospitalizations and better symptom control using home spirometry transmission to physicians via SMS and telephone counseling [25]; increased unscheduled visits to the emergency department and hospitalizations using mobile phone-based interactive self-care software plus management feedback on pulmonary function [25]; and improvements in cough and nighttime symptoms and decreased daily doses of medication using peak flow and symptoms monitoring via SMS [31].
Cardiac Rehabilitation
Exercise capacity in cardiac rehabilitation improved a 6-minute walk test from 524-637 meters (P=.009) in monitored exercise training assisted by a mobile phone app. In an 8-week, nonrandomized clinical trial, there was 17.6% (SD 16.1) improvement in mobile (n=30) versus 11.5% (SD 35.9) in control group (n=32) (P>.05) [32].
Congestive Heart Failure
Mobile technology counseling led to fewer symptom complaints in congestive heart failure subjects [25]. There was relative risk reduction (20%) of death or hospitalization and better quality of life with nurse telephone intervention and cardiologist support [33].
Chronic Lung Diseases
An SMS program improved cough symptoms and sleep quality [34]. Mobile phones recorded respiratory symptoms during exercise training that increased walking distance [34].
Chemotherapy Symptoms
No statistically significant effects were demonstrated on chemotherapy-related toxicity symptoms when patients used a mobile phone app to report symptoms and receive self-care advice [18].
Diabetes
Educational group sessions for diabetic women via SMS showed positive effects on sleep, positive actions, and coping [35].
Surrogate Outcomes
Asthma
There was moderate-quality evidence that text messaging intervention led to greater improvements in peak flow variability (mean difference -11.12, 95% CI -19.56 to -2.68) compared with the control group. No significant differences existed between groups in impact on forced vital capacity or forced expiratory flow in 1 second (GRADE Moderate) [17].
Forced expiratory flow in 1 second improved after 4 months of home spirometry transmission to physicians via SMS and telephone counseling [25]. Using mobile phone-based, interactive self-care software plus management feedback on pulmonary function showed evidence of improvement in pulmonary function and health-related quality of life and decreased unscheduled visits to emergency departments and hospitalizations, plus an increase in the proportion of participants who received leukotriene inhibitors [25].
Cardiac Rehabilitation
Improvement in at least 1 risk factor (relative risk [RR] 1.4, 95% CI 1.1-1.7) in a randomized controlled trial (RCT) that assessed the impact of lifestyle counseling, mobile intervention, devices for home monitoring plus SMS messaging of recommendations, compared to lifestyle counseling alone. The mHealth group was more likely to achieve goals for blood pressure (BP) (62.1% vs 42.9%), HbA1c (86.4% vs 54.2%), and BMI (0.37 kg/m2decrease vs 0.38 increase). No significant differences in smoking cessation, cholesterol, or medication adherence [32].
Chronic Lung Diseases
Three RCTs showed nonsignificant results in lung function parameters [34].
Diabetes
Educational program via SMS for self-management improved HbA1c, low-density lipoprotein cholesterol, and microalbuminuria [33]. Cloud-based interactive voice response calls and automated email for clinicians reduced HbA1c [33]. SMS promoting medication adherence reduced fasting plasma glucose and 2-hour, postprandial glucose [33]. SMS with tailored instructions on diabetes mellitus care to adolescents and elderly patients improved HbA1c [34]. Diabetes self-management intervention reduced HbA1c [18]. Diabetes education and advice via mobile phone and SMS significantly reduced HbA1c [31]. A mobile phone-based, home glucose monitoring program study decreased HbA1c from 13.2% to 10.5% after 3-6 months [35]. Text messaging improved HbA1c, with positive results in 6 of 8 studies [36]. Daily glucose readings were transmitted via mobile phone to a physician who made adjustments plus clinic appointment [25].
The results were mixed on the impact of mobile telemonitoring supporting diabetes management and feedback on HbA1c but were more consistently positive for studies in type 2 diabetes. Ten of the 13 studies in type 2 diabetes and 4 of 7 studies on type 1 diabetes found mHealth led to HbA1c improvement [26]. Studies without health care professional feedback led to HbA1c improvement, suggesting professional feedback might not be necessary [26]
However, one study showed reduced benefit: educational group sessions for diabetic women via SMS showed higher diastolic blood pressure (+7 mmHg) and less spiritual hope at 6 months, and frequent texters had higher BMI and more sedentary time [35]
HIV Management
Text messages to maintain contact, monitor, and respond to medication issues in patients on antiretrovirals statistically significantly reduced human immunodeficiency virus (HIV) viral load by improving adherence [18].
Hypertension
Improvement in BP has been demonstrated with SMS or voice mail and immediate physician feedback [33]; SMS enabling interactive monitoring, where the provider sets reminders for patients, collects data, and schedules visits [34]; and electronic salt sensor and mobile sensor [34].
However, other reviews showed no benefit. Two trials showed no statistically significant reduction in BP [18]. Groups that did or did not receive alerts and reminders had nearly equal percentages of patients with controlled BP at follow-up [31].
Risk Factors for Coronary Artery Disease
Significant improvements were shown with automatic sphygmomanometer, blood glucose and lipid meter, and mobile phone [34].
Weight Loss
Moderate-quality evidence of short-term weight loss in overweight and obese adults with BMI of 25-39.9 using mHealth structured program was shown [29]. Mobile phone personalized advice and motivation observed a significant improvement in percent of body fat lost; however, BMI and systolic and diastolic BP were unchanged [31].
Behavioral or Lifestyle Changes
Physical Activity
Seven of 14 trials reported statistically significant benefits on self-reported physical activity outcomes, but no statistically significant change was demonstrated on trials using SMS to reduce calorie intake and increase physical activity or for trials targeting physical activity only, diet only, or diet and physical activity [18].
Smoking Cessation
Positive results have been demonstrated with moderate-quality evidence that mobile phone-based cessation interventions increased abstinence rates at 26 weeks (RR 1.67 95% CI 1.46-1.90, 12 RCTs in high income countries, 11,885 participants, GRADE Moderate). Six studies verified quitting biochemically at 6 months (RR 1.83 95% CI 1.54-2.19) [20]. SMS-based smoking cessation interventions doubled biochemically verified smoking cessation at 6 months [18].
Sexual Behavior
One study showed statistically significant benefits on behavior change [18].
Processes of Care
Antenatal Support
Pregnant women connecting to their health care provider through bidirectional mobile phone messaging were more likely to have skilled birth attendants [37].
Attendance Rates
A consistent improvement on attendance rates has been demonstrated. Text message reminders improved the rate of attendance at health care appointments compared with no reminders (RR 1.10; 95% CI 1.03-1.17) (7 studies, 5841 participants, GRADE Moderate) [15]. They had a similar impact to phone call reminders (RR 0.99, 95% CI 0.95-1.02) (3 studies, 2509 participants, GRADE Moderate). Text messages plus postal reminders improved attendance rates at appointments compared to postal reminders (RR 1.10, 95% CI 1.02-1.19) (1 study, 291 participants, GRADE Low) [13].
In a limited study, Shetty et al [38] compared the effect of one SMS reminder sent to type 2 diabetes patients every third day for 1 year. Although there was no significant reduction in the mean HbA1c values in either group, the percentage of patients with HbA1c<8% decreased significantly in the SMS group. Free et al [18] included one study assessing SMS reminders on attendance for vaccination at different time points (as different studies). The relative risk was 1.19 (95% CI 1.15-1.23), but there was significant heterogeneity (I2=99.7%, P<.001) [33]. In 8 studies Free et al [19] observed that the pooled effect on appointment attendance using text message reminders versus no reminder was RR 1.06 (95% CI 1.05-1.07) [19]. There was no effect on the number of cancelled appointments (RR 1.08, 95% CI 0.89-1.30), and no difference in attendance using SMS reminders versus other reminders (RR 0.98, 95% CI 0.94-1.02) [19].
Hall et al [22] assessed 6 studies in low- and middle-income countries. The studies showed mHealth interventions to be beneficial, except a pilot study in rural Swaziland [39]. Mahmud et al [40] trained community health workers to use mobile phones for reporting on patient adherence, send appointments reminders, and answer physician queries. This evaluation was based on a retrospective observational study, with the possibility of recall bias.
Data Collection and Health Care Team Communication
Studies that included data collection as a primary mHealth function demonstrated that mobile phones effectively collect and report data, transfer patient-relevant information, and reduce the need for face-to-face communication [37]. There is a reduction in communication delays and improvement on data collection and reporting [23]. One trial reported a statistically significant improvement in nurse/surgeon communication using mobile phones [19].
Adherence to Treatment
mHealth strategies are beneficial to increase adherence to treatment in diabetes patients: SMS to increase adherence to prescriptions in type 2 diabetes [25], electronic blister packs with SMS communication [34], and insulin adherence among type 1 diabetes patients who received tailored text messages with goal-specific prompts [31].
Compliance with medication taking among memory-impaired, HIV-positive patients significantly increased compared to those without impairment. Hepatitis A and B dose vaccination schedules also increased among international travelers with reminders sent to mobile phones [31].
Hall et al reported improved adherence to tuberculosis treatment by using text messages and adherence to HIV therapy, with evidence of reduction of viral load [22], although the authors did not perform quality assessment. The authors commented that a risk-benefit analysis that assessed mHealth reminders to improve tuberculosis medication adherence showed increased mortality and disability-adjusted life years compared to directly observed treatments, and that an RCT in China observed no benefit with voice calls. They also reported limited evidence of contraceptive knowledge improvement with the use of an SMS education scheme; risk reduction of contracting dengue fever, but no significant improvement over alternative schemes; antenatal care improvement, with increases in using skilled birth attendants and women with 4 antenatal visits; and improved vaccination rates in rural Kenya [22].
Hamine et al observed in 2 studies that text messaging tailored to counteract negative beliefs about asthma and education was associated with improved adherence to medication [34].
Diagnosis
Hall et al mentioned studies showing improvement of diagnostic rates of dermatological conditions with mobile teledermatology [17]. However, two trials using mobile phones to transmit photos to offsite clinicians reported significant reductions in correct diagnoses compared to an onsite specialist [18]. One trial reported reduction in quality of electrocardiography (ECG) transmitted via mobile phone to an ECG transmitted by fax, but with no effects on ECG interpretation [18]. Krishna et al [31] showed that fewer days to diagnosis and treatment were reported among those who were notified of test results via text messages.
Cost
The following studies assessed costs, with evidence of reduction when compared to control groups. SMS reminders were more cost-effective than telephone and were equally efficacious [25]. SMS was found to be 35% and 45% less expensive, respectively, per attendance through reductions in research assistants’ work hours and in telecommunications costs. SMS reminders were less expensive than mobile phone reminders [33,40]. The relative cost of the text message per attendance was 55% and 65% of the cost of phone call reminders [13]. There was a reduction in patient burden to transportation time and costs in African countries [23]. The cost of text-based support per 1000 enrolled smokers was GB 278 per quitter, but when future health service costs were included, text-based support was considered a cost-saver [20].
Patient Satisfaction
The individual studies did not assess this outcome.
Potential Harms and Adverse Effects
Only 3 reviews assessed this outcome [15,17,27]. Vodopivec-Jamsek et al [27] reported one study where mobile phone messaging was used to support smoking cessation and that messaging did not have any impact on rates of pain in the thumb or finger joints (RR 1.08, 95% CI 0.74-1.59), or car crash rates (RR 0.88, 95% CI 0.58-1.35) at 26 weeks follow-up.