The Health Impact of mHealth Interventions in India: Systematic Review and Meta-Analysis

Introduction

The use of mobile computing and communication technologies in health care and public health are seen as a rapidly expanding area within eHealth. The World Health Organization’s Global Observatory for eHealth defined mobile health (mHealth) as “medical and public health practice supported by mobile devices, like mobile phones, patient monitoring devices, personal digital assistants, and other wireless devices” [1]. Devices used in mHealth interventions include laptops, tablets, mobile phones, smartphones, palmtops, notebooks, and netbooks.

Features of mobile technology, including mobility, instantaneous access, and direct communication, permit faster transfer of health information, which aid in medical and public health practices. mHealth services range from simple apps to complex technologies including voice messaging, SMS text messaging, multimedia message service, Bluetooth technology, and others, which could transform the worldwide delivery of health services, especially in low- and middle-income countries [1].

Various functionalities such as SMS text messaging, voice messaging, mobile internet browsing, Voice over Internet Protocol services (eg, Skype), instant messaging services, photographic capabilities, and a wide variety of device-based applications available through mobile technology offer a range of opportunities for mHealth interventions, such as text message and interactive voice response campaigns and content to mobile phone–based imaging (which have potential diagnostic capabilities) [2,3]. This technology has a broad extent and accessibility, which can be efficiently leveraged for health care delivery in areas where access is a major constraint [4].

mHealth is increasingly being used for medical services and public health practice for patient communication, monitoring, and education [5,6]. The interventions have also shown to reduce the burden of diseases linked with poverty and an improvement in the accessibility of the health services in terms of clinical diagnosis, treatment adherence, and chronic disease management [1,7-9]. There is considerable interest in mHealth interventions with an enormous potential for beneficial effects on health and health service delivery processes, especially in resource-limited settings such as India [10].

This paper provides a review of evidence regarding the health impacts of mHealth interventions in India. The purpose of this review is to assess health impact in terms of measurable changes in mortality, morbidity, disability-adjusted life years (DALYs), and improved disease detection rates.

Methods

Study Design

This systematic review and meta-analysis was conducted in accordance with the PRISMA (Preferred Reporting Items for Systematic Review and Meta-Analyses) guidelines [11]. Randomized controlled trials (RCTs), non-RCTs (including cluster RCTs and quasi-experimental studies), and prospective parallel cohort studies conducted in India were included. Studies published between April 1, 2011, and March 31, 2021, were considered, and the search was initiated on September 10, 2020, until March 10, 2021. Studies reported in the English language and conducted in India, which addressed the impact of mobile technology, using SMS text messaging or cellular telephone interventions for any disease (eg, diabetes, hypertension, cardiovascular disease, chronic respiratory disease, and cancer) and maternal and child health, and measured outcomes including morbidity, mortality, hospitalization rates, behavioral or lifestyle changes, the process of care improvements, clinical outcomes, patient and provider satisfaction, compliance, and cost-effectiveness, were included in the review.

Literature Search

A literature search was conducted using a combination of text and Medical Subject Headings (MeSH) keywords in major databases, including PubMed, MEDLINE, Scopus, Cochrane Library, Web of Science, and Google scholar, to identify peer-reviewed publications. The MeSH keywords included the following: Text Messaging, Health Literacy, Mobile Applications, Smartphone, Cell phone, Health Impact Assessment, Developing Countries, Multimedia, Cell Phone, Telemedicine, Medication Adherence, India, Hypertension, Primary Health Care, Risk Reduction Behavior, healthcare cost, Health Information Management, and Information Systems. The search field was limited to the title or abstract (or both), and the type of publication was limited to original articles or full-length research articles. We excluded cross-sectional studies, letters, case reports, study protocols, reviews, opinions, gray literature, and non–peer-reviewed publications. The reference lists of articles were also examined to identify other potentially relevant articles. The protocol for this systematic review and meta-analysis has been registered in PROSPERO 2021 (CRD42021235315).

Study Selection and Characteristics

Two researchers (VJ and DO) independently screened the titles and abstracts to identify potentially eligible studies, and further assessment was performed by 2 authors (NKJ and YKJ). Only full-text articles published between 2011 and 2021, written in the English language, were included. The authors excluded duplicates and studies conducted outside India.

Initial searches identified 1393 titles. After removing duplicates, 1120 articles were included for initial screening. Of these, 920 articles were excluded after screening by title and abstract, leaving 200 articles, which were considered in more detail. A further 187 papers were subsequently excluded for not meeting the relevant criteria. Thirteen of the eligible studies were intervention studies, comprising 3 RCTs; 5 quasi–RCTs; 1 cluster RCT; 1 prospective, parallel-group cohort study; and 1 quantitative, single-arm, pretest, posttest interventional study (Figure 1).

‎

Data Extraction

The extracted data included the names of the authors, year of publication, study design, study location, sampling, and main results. All these details were captured and recorded in an Excel (Microsoft Corp) spreadsheet. The information reported in or calculated from the included studies was used for analysis. Corresponding authors of the articles were not contacted for unpublished or additional information. Disagreements related to the inclusion of an article were resolved through consensus among the authors.

Quality Assessment and Assessment of Risk of Bias

Risk of bias of each study was assessed using the Risk of Bias 2 tool for RCTs and Risk Of Bias In Non-randomised Studies - of Interventions for non-RCTs [12,13]. Risk-of-bias grading for the different components of each study is shown in Table 1. Four of the intervention studies were graded as being at low risk of bias, 6 as moderate, and 1 as high.

Meta-Analysis

There was substantial heterogeneity among studies in their mHealth interventions and outcomes, except for studies on maternal, neonatal, and child health. Consequently, we performed a random-effects meta-analysis using the Mantel-Haenszel method in RevMan [27] for 3 comparable studies, which had all used cell phones rather than routine prenatal care as the intervention and had assessed increases in the number of antenatal check-ups, tetanus toxoids administered to pregnant women, institutional deliveries, and iron–folic acid to assess the effect of health care usage. However, as the relevant intervention for the purpose of this review, we exclusively compared the cell phone group to the usual care group in the meta-analysis. However, given the small number of studies, we did not undertake possible sensitivity analyses.

Results

Types of Outcomes Examined

Four studies examined the indicators of maternal, neonatal, and child health [14-17]—these reported the number of antenatal check-ups [14,15,17]; birth weight [15]; institutional delivery [14-17]; knowledge of the danger signs of pregnancy [14,15]; indicators of infant feeding and breastfeeding [14]; usage of antenatal, intrapartum, and postnatal care [14,15,17]; indicators of self-efficacy [15,17]; uptake of immunization [14,15]; and maternal health knowledge [17]. We found two studies evaluating the cost-effectiveness of mHealth programs [18,19]. Other outcomes included improvement in diabetes risk behaviors and increased awareness about the causes and complications of diabetes [20], improvement in medication adherence and the frequency of blood glucose testing [21], change in systolic blood pressure and hemoglobin A_1c levels [22], quality of care delivered by primary health workers [23-25], and oral health education [26]. The results are organized below in accordance with the types of outcomes examined in each study.

Effects on Maternal, Neonatal, and Child Health

A pre-post quasi-experimental study used an mHealth application in the Kaushambi district in Uttar Pradesh, India, to increase the quality of counseling by community health volunteers, resulting in improved uptake of maternal, neonatal, and child health services. A significant increase in coverage iron–folic acid supplementation and identification and self-reporting of illnesses or complications during pregnancy and after delivery were seen in the intervention area, but there was no change in the quality of antenatal care (ANC) care [14]. Similarly, an mHealth application was used in an open cluster RCT conducted in 22 primary health centers in 6 tribal blocks of Bharuch and Narmada districts in Gujrat, India, to assess the increase in the coverage of maternal, neonatal, and child health services and that of at least 2 home visits by accredited social health activists within the first week of birth. There were significant improvements in coverage home visits by accredited social health activists during the antenatal and postnatal period, early initiation of breastfeeding, and exclusive breastfeeding [15].

A pseudo-RCT conducted in Mumbai (Maharashtra, India) by Murthy et al [16], assessed the impact of age- and stage-based mobile phone voice messaging for pregnant women on reduction in low birth weight and child malnutrition and improvement in women’s infant care knowledge and practices. They observed that the intervention group performed well in infant care practice indicators: administering supplementary feeding to the infant at 6 months of age (odds ratio [OR] 1.4, 95% CI 1.08-1.82; P=.009) and fully immunizing the infant (OR 1.531, 95% CI 1.141-2.055; P=.005). Moreover, women in the intervention group had increased knowledge of giving infants solid food by 6 months of age and of the fact that the ideal birth weight is >2.5 kg [16]. A study from Jharkhand used a mobile app to support home visits by community health workers; Ilozumba et al [17] found that women receiving the mHealth intervention were more likely to attend 4 or more ANC visits and had significantly higher odds of delivering a baby at a health center than those receiving standard care and those receiving other interventions from a nongovernmental organization. Moreover, the usage of ANC services and delivery at a health center were associated with the education level of the spouse [17].

Cost-Effectiveness

Prinja et al [18] assessed the cost-effectiveness of the ReMiND (reducing maternal and newborn deaths) program in Uttar Pradesh, India; both the societal and health care perspectives were taken into account. Overall, the ReMiND program was considered a cost-saving intervention from the societal perspective. It resulted in a cost saving of US $90 per DALY averted US $2569 per death averted. From the health system perspective, the ReMiND program incurred an incremental cost of ₹12,993 (US $205) per DALY averted and ₹371,577 (US $5865) per death averted [18]. A study conducted in Gujrat, India, found the ImTeCHO (Innovative Mobile Technology for Community Health Operation) intervention to be cost-effective at an incremental cost of US $74 per life-years saved or US $5057 per death averted [19].

Effect on Chronic Conditions

Study conducted by Pfammatter et al [20] to examine the effect of mDiabetes—a text messaging program to improve diabetes risk behaviors—on fruit, vegetable, and fat intake and exercise among Nokia phone users in India. A greater improvement in the health behavior composite score over 6 months was observed among participants who received the text messages than among those who did not receive text messages [20]. An RCT conducted by Kleinman et al [21] at 3 sites in India assessed the impact of an mHealth diabetes platform on clinical outcomes, patient-reported outcomes, patient and provider satisfaction, and app usage. There was decrease of 1.5% in mean hemoglobin A_1c levels in the intervention group and 0.8% in the usual care group, an improvement in self-reported medication adherence from baseline, and an increase in blood glucose testing in the intervention group from baseline compared to that in the control group (39.0% vs 10.3%, respectively; P=.01) [21]. Prabhakaran et al [22] conducted a cluster-RCT using the mWellcare system for integrated management of 5 chronic conditions (hypertension, diabetes mellitus, current tobacco and alcohol use, and depression). No evidence of difference in systolic blood pressure and hemoglobin A_1c levels was observed between the intervention and control groups [22].

Other Effects

Garner et al [23] determined the effectiveness of an mHealth application to improve hypertension health literacy among vulnerable populations in India. A significant improvement in the understanding of hypertension through the innovative animated application was observed [23]. In the RCT conducted in rural areas of Tamil Nadu, India, Gautham et al [24] observed that mobile app–based procedural guidance for rural frontline health care providers had significant potential for attaining consistently standardized quality of care with patients’ acceptance. Praveen et al [25] showed that implementation of a mobile clinical decision support system for cardiovascular disease management by public nonphysician health care workers and physicians in a rural Indian setting increased the number of referrals to the physician and had potential to help improve cardiovascular disease outcomes, but system-level barriers have an impact on limiting the access to medical care. Jadhav et al [26] assessed the effectiveness of the reinforcement of oral health education SMS text messages and reported that mean Oral Hygiene Index and Gingival Index scores in the intervention group were significantly lower than those in the control group (P<.01).

Effect on Health Care Usage

Among pregnant women, those using mHealth interventions were more likely to take a complete dose of iron–folic acid supplements (OR 14.30, 95% CI 6.65-30.75; Figure 2), both doses of the tetanus toxoid (OR 2.47, 95% CI 0.22-27.37; Figure 3), and to attended 4 or more antenatal care check-ups (OR 1.82, 95% CI 0.65-5.09; Figure 4) than those who received routine prenatal care. No strong evidence of differences regarding institutional deliveries (OR 1.14, 95% CI 0.26-4.95) were found.

‎

‎

‎

Discussion

Principal Findings

mHealth is an implicit, promising tool for addressing several health care system limitations in transitional countries, such as a limited health care workforce, scarce resources, high burden of disease, rapid population growth, and challenges of extending health care to underserved populations. We identified 13 studies showing the impact of mobile technology–based interventions designed to improve health care service delivery processes in the Indian setting. Most studies were at moderate and low risk of bias. Heterogeneity among studies did not allow the calculation of a pooled estimate for all the parameters. However, a meta-analyses of 3 studies arbitrated to be sufficiently homogenous showed that mHealth interventions used for maternal and child health improved the usage of prenatal services including the intake of a complete dose of iron–folic acid supplements, taking both doses of the tetanus toxoid, and attending 4 or more antenatal care check-ups. No strong evidence of differences regarding institutional deliveries were found. A similar review conducted by Lee et al [28] for low- to middle-income countries showed that mHealth technologies are rapidly being used to promote health care use, improve the quality of pre- and postnatal care, and collect data on pregnancy and child health.

In our systematic review, we could not use economic evaluation–tailored reporting standards (such as the CHEERS [Consolidated Health Economic Evaluation Reporting Standards] checklist [29]) for full economic evaluation due to the lack of sufficient economic evaluation studies, as indicated by Iribarren et al [30], who described the evidence related to economic evaluations of mHealth interventions in low- to middle-income countries and in the evaluation of 2 mHealth interventions in India: ReMiND [18] and ImTeCHO [19]. These studies included a comparison of the effectiveness of a health-related outcome and reported economic data. Both the studies showed a positive economic impact considering the societal perspective.

All the studies included in this review provide evidence that the interventions conducted for the chronic diseases had an impact on clinical outcomes, patient and provider satisfaction, app usage, and improvement in health behavior (except for the study conducted by Prabhakaran et al [22]). Similar findings were described in the review conducted by Beratarrechea et al [31] for chronic diseases in transitional countries, which addressed more than 1 outcome and reported a positive impact on chronic disease outcomes.

Limitations and Conclusion

This paper reviews the comprehensive use of mHealth technologies in all sectors of health care in India. We used a thorough, extensive, and highly sensitive literature search technique in this systematic review, which analyses both health and health care usage indicators, encompassing the entire scope of relevant mHealth technologies including those focusing on maternal and child health and chronic diseases. All comparative reviews have been conducted for low- to middle-income countries and mainly focused on the either chronic disease or maternal and child health [28,30-38].

However, due to a small number of studies for a single set of interventions, a meta-analysis for all the impact indicators was not conducted. Additional work is needed to improve and test this with a larger set of interventions, and to determine how to best integrate it with different conceptual frameworks that have been published.