Abstract

Introduction

In the last decade, mobile health (mHealth), the branch of eHealth broadly defined as “the use of mobile computing and communication technologies in health care and public health” [1], has been constantly expanding. Mobile applications for health can target heterogeneous audiences such as doctors, nurses, patients, or even healthy people [1]. Different features of mobile phones may address specific needs in different situations. Available literature suggests that the use of mobile phones serves a wide variety of purposes [2], such as smoking cessation, weight loss, diet and physical activity, treatment adherence, and disease management. The biggest advantages of using mobile devices, and in particular mobile phones, for health are that these devices are personal, intelligent, connected, and always with people [3,4]. Therefore, they can serve patients both in everyday life and during hospitalization or rehabilitation, as well as health care providers during emergency or routine visits. Current evidence suggests that the use of mobile technology can improve diagnosis and compliance with treatment guidelines, as well as patient information, and can increase administrative efficiency [5]. In particular, short message service (SMS) text messaging reminders have been shown to be a simple and efficient option for health services to use in order to improve service delivery, resulting in health benefits for the patients who receive them [6]. Mobile phone technologies have also been shown to be effective in smoking cessation, weight loss, physical activity, diabetes management, STD prevention and treatment, and hypertension [7].

The mobile phone market is constantly evolving. The first digital mobile phones appeared in the early 1990s, and since then, mobile technology has continued to be refined thanks to the development of new features and better networks. Current smartphones have been defined as “mobile telephones with computer features that may enable them to interact with computerized systems, send e-mails, and access the web” [8]. Over a third of US mobile phone users own a smartphone [3,9], and it is estimated that 67.6% of adults worldwide own a mobile phone [2,10], making it the most equitable communication technology [1]. It has been argued that mobile phones could be a solution to overcome the traditional digital divide derived from the introduction of the Internet because they provide new opportunities to reach underserved and previously unreachable parts of the population worldwide, especially in developing countries [2].

Mobile technology, with its diffusion and characteristics, holds a great potential for health care applications. However the use of mobile phones in health care delivery has not been fully explored, and the diverse outcomes of mHealth have barely been documented. Although some literature reviews cover one part or the other of the field [6,11,12], an overall picture is still missing, possibly due to the field’s constant evolution. A recent methodological review sought to map the domain of mobile phone health interventions [13], but it relied on describing the design of the interventions, with a clear focus on technology, rather than the outcomes. As the authors stated, their motivation lay in the fact that “effectiveness reviews can be best done at the level of a particular pathology”, while they wanted to draw a more comprehensive taxonomy of the field.

The main objective of this paper, as stated in the title, is to map the field, but without omitting the outcome measures. This means that our intention is to investigate how the impact of mobile phones on health has been assessed in peer-reviewed scientific literature. In particular, we are interested in understanding the evolution over the past decade, how the interventions have been developed, the main health care delivery areas where the impact of mobile technologies has been assessed, the methodology and features used, and finally, the type of outcome measures and general impact of the intervention.

The second objective of this review is to understand, after the 5 years since the introduction of the new generation of smartphones (eg, the iPhone in 2007), whether and how these devices have triggered research. The appeal of these new devices resides in the fact that they include several computer-like built-in features (eg, the GPS or the accelerometer) allowing the monitoring of a whole series of behaviors. Additionally, new mobile operating systems allow users to customize their devices according to their needs, by downloading apps available for free or for a low price from a central store. Klasnja and Pratt named this kind of feature “native application” [13], which is a typical complex and sophisticated application that can be implemented on major smartphone platforms (iOS, Android, Symbian, BlackBerry, webOS, and Windows Phone). In 2012, smartphone users spent US $8 billion for paid apps in the top 5 app platforms, and the European mobile app market size reached €1.68 billion [14]. Therefore, iPhones and similar devices are potentially very interesting for application in health care—they already integrate most of the features that researchers previously had to add to traditional mobile phones in order to use them for health-related purposes and monitoring [15,16].

Methods

The objective of this study was to provide a comprehensive picture of how the impact of mHealth was assessed in the scientific literature in its first decade of existence. For this purpose, a systematic literature review was conducted in which relevant studies were categorized in a two-step process. The first step included the review of the titles and abstracts of all publications that were identified as potentially relevant, with the goal of assessing whether they might meet the inclusion criteria for the systematic review. Selected abstracts were categorized at this stage using general typologies partly adapted from existing literature [1,2,12] and partly created inductively from a subsample of the publications. Categories referred to the type of methodology used, the impact area (ie, remote monitoring, data gathering, communication, self-management, training/education, improve adherence, health promotion), and the type of study. In a second step, all the publications not excluded during the abstract and title review stage underwent a full-text review. All publications that met all eligibility criteria (see below) made up the final sample.

Results

The flowchart in Figure 1 summarizes the different steps of the literature search and review process. A first search identified 4039 articles. After checking for duplicates, 747 articles met the predefined inclusion criteria. Initially, articles were categorized by type of study: quantitative, qualitative, mixed methods, review, and system design. Since articles in the last typology described the development of a mobile technology but did not include any actual testing, they were excluded from further analysis together with reviews, reducing the final sample to 352 articles. More than half of the 352 studies (56%) included at least some testing of a mobile phone application via proper interventions or in small samples. Most of the studies analyzed (86%) applied a quantitative methodology and were designed to address simultaneously one or more impact areas. An upsurge was noted, starting from 2008, when the articles doubled in comparison with the previous year, and this upward trend reached its maximum in 2011 (36% of the total in a 10-year time period). The search of scientific databases without a time limit yielded an article distribution on the topic over 10 years, from 2002 to 2012.

Open in a separate window

Figure 1

Summary of literature search and review process.

The final sample for the full-text review included 117 [17-133] articles out of the 352 described above. After title and abstract review, an additional 157 articles were excluded because of no actual testing, while another 78 were excluded during full-text revision for different reasons (eg, no patients involved, mobile device other than phone, study duplicates). Looking at this past decade (Figure 2), we again observed an upsurge in the field: from 1 article in 2002 to 30 articles in 2011. The largest upsurge again came between 2007 and 2008, when the articles almost doubled, similar to what was already noted during the abstract review phase.

Open in a separate window

Figure 2

Distribution of articles over time.

In order to better reflect the objectives of our review and to mirror the development over time, all the results are presented by splitting our observation period in two halves (2002-2007 and 2008-2012). The first period includes 23 articles, while the second period includes 94. The 117 articles in the final sample were published in 77 different journals, which can be grouped in four disciplinary fields: technology, medicine, social sciences, and the intersection between technology and medicine. During the first half of the observed decade, most of the articles on mHealth were published in medical journals (52%) and in journals focused on medicine and technology (44%).The remaining 4% of articles were published in journals focused only on technology. In the second half, the share of articles published in medical journals grew from 52% to 60%, while coverage of the topic by technology journals did not change (4%). At the same time, a decrease in the share of articles published by journals dealing with medicine and technology was observed (from 44% to 35%). In the second period, we found one article from a new disciplinary field, the social sciences.

The geographical areas focusing on this type of research were mainly Europe (34%) and North America (33%). However, if we look at results split by time periods, Europe’s interest seems to decrease from the first period (52%) to the second (30%), and the same happens in Asia (from 17% to 10%). A different picture can be found on all the other continents, where the number of studies in the field increased. This is the case in North America (from 17% to 37%), Australia (from 13% to 15%), and especially Africa (from 0% to 6%) and South America (from 0% to 2%).

Specific health conditions addressed in the studies ranged from diabetes to mental health, from obesity to well-being and postoperative care. Figure 3 shows the number of articles for every health condition for which mHealth applications were studied. As shown in the graph, diabetes has received a great deal of attention. Moreover, after grouping the conditions into larger classes, it becomes clear that the focus of mHealth research is chronic conditions (74 studies), followed by prevention/well-being (22 studies), and acute conditions (21 studies).

Open in a separate window

Figure 3

Conditions addressed by mHealth applications in mHealth research.

In reviewing the background of the studies, we found that a description of the development of the intervention, and especially of how this was grounded, occurred more often in the second period (84%) than in the first (65%). During the last 5 years, only 1 study was uniquely theory-based (1%), while the majority was evidence-based (73%) or based both on theory and evidence (10%); 15 studies (16%) provided a more general description that was based on neither theory nor previous evidence.

From a methodological point of view, the majority of articles were clinical trials (50%), followed by pilot studies (44%). However, both of these study designs diminished over the last years of our observation period as new types of research designs were introduced, namely observational studies (2% of all articles in the second part of the observation period), case studies (2%), case series (2%), and cross-sectional studies (2%). The samples used to test mobile health applications were mostly small (less than 50 people) in both the first (61%) and the second half (49%) of our observation. Interestingly in the second half, the number of medium-sized samples increased (from 17% to 33%). Larger samples were used in 21 (19%) articles; however, they were more frequent in the first half of the observation (22%) than in the second half (18%).

Moving from research methodology to the actual target of investigation, ie, mobile phones, our classification highlighted a more rigorous and diversified description of the technology used in interventions. In recent years, new kinds of mobile phones have been used, such as smartphones (8%) and ad hoc phones (3%), which are devices developed specifically by the researchers to manage a specific condition. Unfortunately, the kind of mobile phone used was not even specified most of the time (71% of the overall sample).

We identified seven main categories of mobile features used in the studies, and an article could fall in one or more of these (ie, the categories were not mutually exclusive). Half of the studies (49%) applied text messaging, and 32% applied some features developed ad hoc for a specific condition. Add-ons (eg, a glucometer to measure blood sugar or a pedometer for physical activity) were used in 12% of the cases together with ad hoc features. Other features such as voice (10%), video (6%), and multimedia messaging service (MMS) (3%) were used less frequently. Native applications for smartphones were applied in 7 studies (6%) out of the 8 using smartphones. However, none of them applied already existing and publicly available apps.

The impact areas to which interventions were directed were coded into seven categories, again not mutually exclusive. The majority of articles addressed health promotion (38%) and self-management (33%), but also communication (22%), remote monitoring (21%), data gathering (21%), improvement of adherence (20%), and training/education (13%). The focus on most of these areas increased over time, eg, on self-management (from 30% to 33%) and communication (from 17% to 23%). Only health promotion (from 29% to 27%) and training/education (from 13% to 10%) had a slight decrease.

Regarding the aims of the interventions, both the evaluation of the technology itself (35%) and of its impact on health outcomes (43%) dominated in the first 5 years. In the second half, however, interest clearly moved toward evaluating the impact of mobile technology on health outcomes (73%). While the majority of the studies investigated only the impact of the mobile application on health outcomes (51%), some also assessed both the technology and its impact on health outcomes (22%).

Another point of interest was the outcome measure used to assess the impact of mobile phones. In the majority of cases, the outcome measures were a combination of both self-reported and objective data (44% of the overall sample). If we look at the evolution over time, self-reported measures increased (from 9% to 20%), whereas objective measures decreased slightly (from 39% to 36%), and this was also the case for the combination of self-reported and objective measures (from 52% to 43%).

Our examination of the type of data collected showed that clinical measures were often the only outcomes observed (30%), and this phenomenon increased over time (from 22% to 31%). 14% of the articles were focused only on user assessment of the technology, even if this decreased during the observation period (from 17% to 13%). Psychosocial measures were the outcome in 9% of the studies, and this increased slightly over time (from 9% to 10%). The remaining articles (53%) considered outcomes deriving from all possible combinations of these main three. The most frequent combination was clinical measures together with user assessment of technology (17%).

An overall positive impact of the intervention was described by a total of 69 studies (60%). In the first period, the impact of the interventions was mainly either mixed (43%) or positive (57%). In the second part of the observation, the number of interventions with a positive impact slightly increased (60%), while the number of those with mixed impact decreased (33%). In this second period, interventions with negative (6%) or no impact (1%) were reported as well.

Discussion

This systematic literature review has found encouraging trends pointing toward the development of mHealth as an autonomous field of study, which is of interest to different disciplines. In the case of social sciences for instance, this is shown by the publication of a special issue of the Journal of Health Communication dedicated entirely to mHealth in May 2012. We took this into account but did not include it in the current review because of the date of publication.

The increase in the number of articles over the past decade indicates an increasing interest in peer-reviewed scientific literature on the topic. In particular, the number of articles almost doubled from 2007 to 2008. All continents engaged in research on mHealth during the last 5 years. Many different conditions were addressed by the studies analyzed, with an evident focus on chronic conditions. More recently, the development of the interventions became more accurately described and grounded on past evidence or theory. In the last decade, the first in the history of mHealth, new methodologies were applied to the field, and while the majority of the studies were pilot studies or clinical trials, some new approaches were introduced. At the same time, the samples used in such studies grew, thus pointing toward a more reliable assessment. Newer and more advanced technologies were tested; however, the potential of smartphones does not seem to have been fully exploited yet. Indeed, half of the included studies applied very basic features of mobile phones, such as text messaging, which corroborates evidence already established in the field [7]. One third of the studies used features from add-ons, and only a few other examples used video, images, or MMS. Mobile technology interventions were directed to different impact areas, and most frequently to health promotion and self-management. Interestingly, research moved from an evaluation of the technology itself to an assessment of its impact on health outcomes, and an increase in the number of interventions with an overall positive impact was observed. These observations point toward a field that is becoming more structured, coherent, and solid. Among the outcome measures used, both self-reported and objective measures could be found, and in some cases were combined. In the last 5 years, the use of self-reported measures increased. Clinical measures only were observed as outcomes in almost one third of the studies, and an increasing trend was evident. Consistent with the observed trend towards an assessment of the impact of technology, we noticed an increase in psychosocial and clinical measures, sometimes in combination with users’ assessment of the technology, and a decrease of assessments of the technology alone.

Abbreviations

Footnotes

References