It is well documented that regular physical activity is associated with reduced morbidity and mortality attributable to such diseases as cardiovascular disease, diabetes, and some cancers [1-4]. Unfortunately, participation rates have remained dismally low in spite of significant scientific endeavors to enhance participation, reduce attrition, and increase maintenance of this important health behavior. In 2007, less than half of all adults in the United States achieved recommended levels of physical activity [5]. Furthermore, past research suggests that among those beginning a new exercise program, 50% will drop out within six months [6]. The advent of new and ubiquitous technologies offers a potential solution to increasing the effectiveness of and adherence to physical activity interventions. One such technology is the mobile telephone, which has shown promise as a delivery mechanism for health behavior interventions. These devices have become a more pervasive part of society with usage rates increasing exponentially. For example, in the United States some 83% of adults own mobile phones, while in younger populations (ie, 18-24 year olds) as many as 95% own mobile phones [7]. In the United States [8], Australia [9], and Western Europe [10], activated mobile phones outnumber citizens.

Declining cost and enhanced versatility in features are likely contributors to the rapid increase in mobile phone usage, and short message service (SMS, ie, text-messaging) has become an almost universal way to engage in brief conversations and convey short messages. In the United States, 73% of all adult cell phone users send daily text messages, at an average rate of 39.1 per day. Among younger Americans, a staggering 97% of mobile phone users send text messages at an average rate of 87.7 messages per day [7]. In total, nearly 2.1 trillion messages were sent in 2010 [8]. In the late 2000s, the growth of feature phones (ie, devices capable of basic voice and multimedia functions) was supplanted by surging growth in the smartphone market. These devices originally combined the computing power of handheld computers with mobile communication features serving primarily adult professionals. As of May 2011, 35% of all mobile consumers in the U.S. owned smartphones, and rates were higher in ethnic minorities [7]. Smartphones are equipped with advanced technological features that distinguish them from the feature phone. Typically they are capable of sending and receiving information via the Internet, connecting to local wireless networks and Bluetooth devices, utilizing global positioning system (GPS) data and allowing users to download countless mobile applications straight to their device from just about any location.

Behavioral scientists have begun to realize the potential of mobile devices to understand multiple health behaviors, and meta-analyses have supported the efficacy of mobile technology for influencing behaviors including diabetes management [11] and smoking cessation [12]. It is clear that some of the unique qualities of these devices would be attractive features for physical activity interventions allowing scientists to: collect objective and self-report measures of activity in real time; provide feedback and support at the point of decision; provide interactive, immersive, and individualized content that is automatically generated; and deliver materials on a device that is already carried by the individual [13]. A number of reviews exist examining the use of technologies that offer similar benefits to mobile devices and their effects on health behaviors. For example, Goode et al [14] reviewed telephone-based interventions for influencing physical activity and dietary behaviors. Multiple researchers have examined aspects of Internet-delivered interventions [15,16]. Others have examined the use of mobile technology to aid in disease prevention and management [17], as well as the influence of SMS technology on various health behaviors (eg, smoking cessation, diabetes self-management, asthma self-management; [18]). To date, however, no meta-analysis has been conducted that examines the efficacy of mobile devices for changing physical activity behavior. Given the increase in research applications of such devices, we believe that a comprehensive analysis of the influence of mobile devices on physical activity behavior would afford future researchers a foundation for guiding subsequent interventions. Herein, we present a meta-analysis of interventions that have utilized mobile devices (ie, PDA or mobile telephones) to influence physical activity behavior.

Previous reviews and meta-analyses have identified the mobile platform as an effective means to influence multiple health behaviors, including diabetes management [11] and smoking cessation [12]. The present meta-analysis provides some preliminary support for interventions using mobile technology to increase physical activity behavior. Although the use of mobile technology in physical activity research is still in its infancy, we believe that this finding serves as an important foundation for informing the development of appropriate and efficient intervention techniques using such technology to enhance this important health behavior.

However, this initial enthusiasm must be tempered by consideration of the limitations inherent in the individual studies reviewed. First, because of its broad accessibility during most of the included interventions, SMS is the primary technology utilized in this review, which hampers our ability to make evidence-based statements regarding the efficacy of interventions that utilize smartphones. Further, one primary weakness in the studies reviewed here is the inability to determine the unique contribution of the mobile device component to changing physical activity behavior. For the most part, mobile devices have been used primarily as data collection methods (eg, steps reported via SMS) or as supplemental materials (eg, provision of feedback via SMS) to a broader behavior change intervention relying on more traditional methods (eg, face-to-face counseling). When incorporated into larger interventions, SMS messages present a nimble and efficient means to relay feedback and information to participants, provide participants freedom in accessing the intervention message, and to an extent, SMS allows for real-time assessment of behavior.

For example, Fjeldsoe et al [34] developed an SMS-based intervention to increase physical activity in post-natal mothers who were sent bi-weekly, social cognitive-based motivational messages tailored to the study participants. In addition to these messages, a goal-setting refrigerator magnet and face-to-face and telephone counseling were provided at baseline, and telephone counseling alone was provided at week six. The researchers created a useful and innovative model of population-specific text-messaging; however, the inclusion of the counseling and magnet components highlights the inability of current SMS-based interventions to be all-encompassing. This is in contrast to the current movement by many in the area of Internet-based health research who recommend full automation (ie, minimal researcher contact, few supplementary materials, and automatic generation of individualized user content). Hurling et al [35] used an automated design in creating an Internet-based physical activity intervention that utilized SMS to deliver motivational and reminder messages to participants. The intervention was effective; however, our understanding of the efficacy of the mobile component is again clouded by its role as an SMS supplement to the broader, Internet-based trial.

Unfortunately, the simple adoption of mobile technologies does not ensure effective intervention methods. Theoretical frameworks should guide interventions to help researchers understand which components were most effective in successful (or unsuccessful) trials. For example, Prestwich et al [38] instructed participants to develop implementation intentions and a goal relative to brisk walking. These intentions are based in part on Gollwitzer’s [40] position that anticipating and planning goal-directed responses removes some burden of responsibility from the individual and delegates it to the situation. That is, implementation intentions are formed in such a way that one self-regulates in a manner following “when I encounter X, I will respond by Y” [40], thus ensuring that when a particular situation is encountered, the desired response occurs automatically. After forming goals and implementation intentions, subjects were assigned to one of three groups: one that received SMS reminders of their implementation intentions, one that received SMS reminders of their brisk walking goal, and a control group. Both SMS groups significantly increased the amount of walking relative to control. Further, the SMS plus goal group better recalled their goals, while the SMS plus implementation intentions group better recalled their implementation intentions, indicating that this form of mobile intervention successfully supported the underlying theoretical principles of the study.

It is worth considering, however, that much can be added to current theoretical models of behavior change so that they are better suited to design mobile interventions and interpret results. Ritterband et al [41] have developed a behavior change model for Internet-based interventions that can be used to guide mobile interventions, given the similarities of the two platforms. The authors note that at the time of writing, no theoretical model existed to guide development and testing of Internet-based interventions. More recently, Riley and colleagues [13] questioned whether or not current behavioral theories are adequate for designing and implementing mobile interventions. They acknowledge that current models account for the state of the participant at baseline and challenge that they do not adequately account for the interplay between user experiences and the dynamic and adaptable nature of mobile interventions. For Internet and mobile interventions alike, theories that account for inter- and intra-individual change over time may be a better fit.

One of the unique and advantageous features of mobile devices, and smartphones in particular, is their use of the popular and widespread native applications or “apps”. Rather than relying on Internet connectivity to deliver content that resides on a remote server (ie, the method utilized by “web apps”), native apps are developed for the mobile operating system, reside on the user’s mobile device, and may store data locally or exchange it via the Internet. Importantly, native apps allow a greater degree of flexibility and complexity to software and intervention designers. Since 2008, application development has continued to grow across multiple platforms at an exponential rate and has mirrored the increase in smartphone users worldwide [42].

Though limited in number at the time of writing, research using native applications gives cause for optimism. For example, King et al [30] developed software that cued individuals to participate in a short survey twice daily. This survey assessed types of physical activity, context of physical activity, and behavioral/motivational factors. The physical activity program was grounded in social cognitive principles and included an assessment of barriers and enablers; self-regulation of step counts; and a goal-setting component, from which personalized PDA-delivered graphical/textual feedback was developed. The authors reported a mean increase in moderate to vigorous physical activity of 177.7 minutes per week, as compared with a mean decrease of 80 minutes in the control group over eight weeks. This early native application−based intervention demonstrated the versatility and potential efficacy of this mode of delivery.

Finally, inclusion of advanced sensors (eg, integrated accelerometer and GPS devices) holds promise for more accurate assessment of physical activity behavior in real time. Ecological Momentary Assessment (EMA) is a measurement strategy that aims to collect data reflecting behavior and the social/physical context that influences such behavior. Often this is accomplished by providing a prompt to the participant, cuing them to submit reports relative to, for example, their current location, the type of activity they are currently participating in, and their social context [43-45]. In combination with smartphone technology, integrated motion sensing can bolster the accuracy of activity measurement, while GPS data may provide geographic location information, allowing for a more detailed examination of the environmental context in which activity does or does not occur [44]. Bergman et al [46] have questioned the validity of mobile phone−based accelerometry, although their study examined only a single mobile application meant to estimate number of steps taken. Recently, Wu et al [47] used accelerometer and gyroscope data obtained by smartphone devices to classify activity types using machine learning algorithms. They found that they could accurately classify walking and jogging activities with greater than 90% accuracy. However, there still remain issues to resolve relative to the best location on the body to obtain accurate movement data on larger and more diverse populations, and the most accurate algorithms for quantifying different types and intensities of activity.