2.1. Search Strategy

An initial and updated literature search was conducted on August 20, 2018 and December 20, 2019 respectively, using the combination of key terms related to mHealth/eHealth, exercise/physical activity and older people in the databases of PubMed, Embase, Web of science and COCHRANE library. The search terms, MeSH and index terms, are attached in Appendix A.

3.1. Inclusion and Exclusion Criteria

All study designs were included in this review. Studies were included in the review, if they met all of these three conditions, (1) age of all participants was 55 years or above, (2) mHealth as an intervention for PA, (3) the outcome measure was related to usability, acceptability and/or effectiveness on PA.

Regarding the second criterion, studies with mHealth interventions in the form of mobile phone, smartphone or tablets (text messages, mobile phone or tablet enabled software or apps and smartphones with or without portable activity monitors) targeting PA solely or as part of another treatment (e.g. in diabetic management) were included in this review.

Studies on telephone interventions, video callings, Personal digital assistants (PDAs) and web apps were excluded if it was not clear whether mobile phones/tablets were included in the intervention or not.

Studies not fulfilling any of the above conditions, letters to the editor, conference abstracts and reviews were also excluded.

3.2. Selection of studies

As a first step, the titles and abstracts screening of the literature was carried out by two independent authors (first and third author). In the second step, full-text articles of the relevant studies were screened while in the third step, the reference lists of the suitable articles and the review articles on eHealth were searched for additional articles. At each step, any disagreement was discussed until consensus was reached. In the case of non-consensus, the study was discussed with the second author and disagreements were resolved.

3.3. Data Extraction

A protocol was developed for data extraction from the articles. The information related to article (year of publication and country of research), participants (male and female ratio and mean age), intervention (name and description of intervention, duration and frequency of reminders for PA), research design, PA measures and results (effectiveness, usability and acceptability of the intervention) was extracted.

3.4. Study Quality

The quality of studies was assessed by two authors (first and third author) with Cochrane Collaboration’s tool for assessing risk of bias (Higgins et al., 2011) which has six domains: (1) sequence generation for random allocation to conditions; (2) concealment of allocation to conditions; (3) addressing incomplete outcome data; (4) selective outcome reporting; (5) blinding of participants and researchers, and (6) other sources of bias. Domain five was not applicable in this review as participants and researchers were aware of allocation to conditions. Therefore, this domain was not assessed. Each study received a judgement in the form of low risk of bias (+), high risk of bias (−) or unclear risk of bias (?) on respective domains.

The overall risk of bias was rated as low, if the study scored low risk of bias on three or more domains, moderate when it scored low on two domains and high when it scored low on one or no domain. If a domain was not applicable (N.A.), it was considered having a high risk of bias.

When the design of a study was not clear, the corresponding author was contacted through email for the details of the study design.

3.5. Data synthesis

We conducted a systematic review because quantitative data synthesis (meta-analysis) was not possible due to the low number of studies and heterogeneity of the data.

4.1. Identified studies

The process of study selection for this review is summarized in figure 1. The literature search resulted in 662 articles after the removal of 167 duplicate articles, and 611 articles dropped out in the screening process. The major reasons for drop out of articles were the low age of participants, the lack of a mHealth component or the absence of usability/acceptability/effectiveness on PA. Only nine articles met the inclusion criteria and one article was found in the reference list of an eHealth review, so ten articles were included in this review.

Figure 1. PRISMA flow chart diagram for the selection of studies

4.2. Study Quality

The study design was not clear in two studies and the corresponding authors of the studies did not respond satisfactorily to query and reminders.

4.3. Study and Intervention Characteristics

Of the ten studies, six were related to PA or exercise in community living healthy elderly (van Het et al., 2014, Lyons et al., 2017, Knight et al., 2014, Kim and Glanz, 2013, Muller et al., 2016, Paul et al., 2017), one to increase PA in older cancer patients (Hong et al., 2015), one to improve PA in healthy residents (chosen by a physiotherapist) of a care home (Joosen et al., 2018), one to senior centres (Shake et al., 2018) and one to the participants visiting a geriatric clinic (Li et al., 2019).

All of the ten studies investigated the effectiveness of mHealth and three studies examined features of acceptability and usability as well (van Het et al., 2014, Lyons et al., 2017)(Paul et al., 2017, Hong et al., 2015, Joosen et al., 2018). Tablet or mobile-enabled app interventions were investigated in five of the studies (van Het et al., 2014, Lyons et al., 2017)(Paul et al., 2017, Hong et al., 2015);(Shake et al., 2018) . Simple motivational text message interventions were used in two of the studies (Kim and Glanz, 2013, Muller et al., 2016). A smartphone with pedometer intervention was used in one study (Knight et al., 2014). A smartwatch paired with a tablet was used in one study (Li et al., 2019). While a mHealth system intervention (consisting of a smartphone with wireless activity sensor mounted on the body for automatic continuous monitoring of PA) was used in one study conducted at a care home (Joosen et al., 2018). The duration of use of the intervention was 12 weeks in five of the studies (van Het et al., 2014, Lyons et al., 2017, Knight et al., 2014)(Muller et al., 2016, Hong et al., 2015), 10 weeks in two studies (Joosen et al., 2018, Shake et al., 2018), 6 weeks (Kim and Glanz, 2013, Paul et al., 2017) in two studies and 4 weeks (Li et al., 2019) in one study. Post-intervention measurements were taken immediately or within one week after the end of the intervention period and there were no follow-ups.

Five of the studies were conducted in the USA, of which four included general older Americans (Lyons et al., 2017, Hong et al., 2015)(Shake et al., 2018, Li et al., 2019) and one included older immigrants of African descent (Kim and Glanz, 2013). The other five studies were conducted in Canada -North America (Knight et al., 2014), Belgium (Joosen et al., 2018), Switzerland (van Het et al., 2014), Malaysia (Muller et al., 2016) and UK(Paul et al., 2017).

Of the ten included studies, five studies were RCTs (Lyons et al., 2017, Knight et al., 2014, Kim and Glanz, 2013, Muller et al., 2016)(Shake et al., 2018), one was a pre-post randomized study (Joosen et al., 2018); two had a pre-post study design (Hong et al., 2015, Li et al., 2019), one was mixed-method study (Paul et al., 2017) and one had a preclinical exploratory design with pre-post measurements and a control group (van Het et al., 2014). There was no control group in three of the studies (Hong et al., 2015, Joosen et al., 2018)(Li et al., 2019), one study compared three groups (sedentary, exercise and comprehensive groups) based on different advice for activity (Knight et al., 2014), the participants were divided into four groups in one study for focal group discussion (Paul et al., 2017) while there was an exercise/training booklet control group in three studies (van Het et al., 2014, Kim and Glanz, 2013)(Muller et al., 2016), a waiting list control group in one study (Lyons et al., 2017) and a control group without specific exercise instruction in one study (Shake et al., 2018).

4.4. Usability and acceptance of interventions

Usability and acceptance of interventions were examined in three studies (van Het et al., 2014, Lyons et al., 2017)(Paul et al., 2017, Hong et al., 2015, Joosen et al., 2018). Lyons et al. (2017) found that the Jawbone Up app was usable in general older people and the acceptance rate was 80%, the app was considered user-friendly, convenient and comfortable in use by the participants. The participants also intended to continue use of the app in future.

Joosen et al. (2018) found that a smartphone-based solution is usable for monitoring PA in a care home. This system can monitor PA and present the level of PA in graphical form which can be translated into a fitness score. Furthermore, 58% of the participants consider the mHealth system beneficial to stay motivated for PA due to active monitoring. However, 74% of the participants asked for more encouragement to stay active in future.

Paul et al. (2017) reported on the acceptability and usability of the STARFISH app. In the pilot study, older people in focal group discussions gave positive comments on the acceptability and usability of the app for staying physically active. Many participants in the study found that the app was easy to understand and simple to use. The participants expressed that the app encouraged them to perform PA.

4.5. Effectiveness of interventions

Overall, eight out of ten studies found that mHealth interventions were effective on PA or exercise in older people irrespective of the kind of mHealth intervention (whether it was app-based, text message-based or mounted sensor-based for activity monitoring). In the studies of Knight et al. (2014) and Paul et al. (2017), the increase in step count for PA was not statistically significant. Further, the study of Joosen et al. (2018) reported the transient effect (increase in PA) of the mHealth system until the mid-period of the application of the intervention, when a motivation to use new technology was there. While the study of Muller et al. (2016) reported that PA behaviour diminished when the intervention was stopped.

Between group comparisons, next to ingroup comparisons, were made in six of the studies (van Het et al., 2014, Lyons et al., 2017, Knight et al., 2014, Kim and Glanz, 2013, Muller et al., 2016)(Shake et al., 2018). The between group comparisons have shown that the participants in the intervention groups improved more on PA in comparison to the control groups. Ingroup comparisons were made in all studies and participant improved on PA measures compared to baseline. Further details can be found in Table 2.

4.6. Motivational Reminders

One of the studies in our review reported a decrease in PA level after withdrawal of motivational reminders (Kampmeijer et al., 2016). It is difficult to say whether a single daily reminder is more effective than a reminder which is two or three times a day for optimal PA. In the studies, interventions were diverse which make a proper comparison difficult. The study on self-measured PA reported the efficacy of five different motivational text messages five times a week (Muller et al., 2016), similarly a study with three text messages (in the morning, afternoon and evening) three days a week also reported effectiveness on step count (Kim and Glanz, 2013). Furthermore, fortnightly email was found to be effective to improve PA in cancer patients (Hong et al., 2015). Similarly, interventions with daily reminders for exercises reported better results than no-reminder control groups (van Het et al., 2014).

4.7. Effectiveness in relation to form of mHealth

Comparison of the effectiveness of mHealth in relation to simple and complex form of mHealth on objective measurement of step count was inconclusive. A simple motivational text message intervention in the study of Kim & Glanz (2013) resulted in an increase of approximately 680 step counts in 6 weeks. Though, a complex mHealth intervention in the study of Lyons et. al. (2017) resulted in an increase of approximately 1100 step counts in 12 weeks. Due to the difference between studies in the intervention period and population characteristics, it is difficult to say whether the simplicity or complexity of mHealth interventions have an effect on PA or not. It is noticeable that none of the studies in our review compared a text message intervention with an app intervention to improve PA in older people.

5.1. Motivational support

Motivational support is considered as an important aspect of mHealth interventions to improve PA (Kampmeijer et al., 2016). Motivational reminders may be important in interventions, but the optimal frequency of these reminders is not clear yet. If WHO guidelines are followed, the recommended PA is at least 150 minutes a week (accumulative) that can be like 30 minutes a day and five times a week in healthy older people while older adults with poor mobility should exercise for at least three days a week (World Health Organization. Global Strategy on Diet, Physical Activity and Health. https://www.who.int/dietphysicalactivity/leaflet-physical-activity-recommendations.pdfua=1. Published, 2011). A reminder on three or five days a week seems an appropriate choice, but the frequency of reminders also depends on the personality type (sedentary or active), personal need and preference of the person using it. Future research could investigate the optimal frequency of reminders for an optimal PA in different older people.

5.2. Simple and complex forms of mHealth

It would be interesting to investigate differences in effectiveness between simple and complex forms of mHealth in the future. In the future, apps could be more prevalent in advanced regions like Europe, the USA and China where more people own a smartphone while text messages are a good solution in developing countries in Asia and Africa where mobile phones are more common than smartphones and text messages are not expensive. It may be investigated in future research whether a motivational text message may have the same effect as an app or a complex system of mHealth (with mounted sensor) in older people.

5.3. Optimal Step Count

Step count is the objective method of measurement of PA and the average step count in healthy older people may range from 2000 to 9000 and pedometer-based interventions could cause an increase of 775 step count (Tudor-Locke et al., 2011). The recommended (increase in) daily step count for good physical health in older people is not specifically operationalized yet. In fact, the increase in step count is operationalized for some physiological conditions. For example, an increase of 600 or more daily step count is related to a low risk of hospital admission in older patients of Chronic Obstructive Pulmonary Disease (COPD)(Demeyer et al., 2016). More research is needed to determine the recommended (increase in) step count for older people.

5.4. WHO on PA in bouts of 10 minutes

The WHO recommends aerobic PA in bouts of at least 10 minutes (World Health Organization. Global Strategy on Diet, Physical Activity and Health. https://www.who.int/dietphysicalactivity/leaflet-physical-activity-recommendations.pdfua=1. Published, 2011). However, the included studies in this review did not explicitly use the bouts of 10 minutes. This might be related to the fact that the activity performed in bouts of less than 10 minutes could be equally effective, as the recommendation of PA in bouts of 10 or more minutes is supported by limited data (Murphy et al., 2009, Glazer et al., 2013). Further, intensive exercises of short duration do not give so much health benefits, when there are long periods of sedentary behaviour (Chastin et al., 2011). In the future, the WHO may further investigate/elaborate its recommendation on PA for older people regarding the bouts of 10 minutes.

5.5. mHealth and Population groups

An effective PA intervention for the general population might need some adaptations to stay effective in the subgroups of that population(Boekhout et al., 2017). In older people, the effectiveness of the same mHealth interventions may vary in different subgroups based on age, educational status, digital literacy, ethnicity and cultural perspectives. So, older people may not be treated as a single group. In the future, it might be interesting to compare an intervention in different groups of older people.

5.6. Future of mHealth

The future of mHealth interventions for PA in older people looks promising. mHealth interventions have several benefits in comparison to traditional interventions. These benefits are related to low costs, the flexibility of time and place to work with the interventions, and the sustainability of mHealth interventions. Traditional interventions with coaching professionals are usually expensive and require a dedicated time and place. Further, a large number of older people that need PA interventions makes mHealth really promising.

6.1. 2.8. Recommendations for future research

We recommend future research on the comparison of simple mHealth solutions with complex mHealth interventions. We recommend evaluation of the effectiveness of text message’ reminders delivered on a mobile phone compared to an app’ reminder. We suggest an investigation of the frequency of optimal reminders for PA. Only few studies were conducted on usability and acceptability of mHealth and more research is needed on it. Further research is recommended for evaluation of usability, acceptability and effectiveness of mHealth in older people in relation to digital literacy, physical health, socioeconomic status and follow-up of long duration.