The Effectiveness of Self-Guided Digital Interventions to Improve Physical Activity and Exercise Outcomes for People With Chronic Conditions: A Systematic Review and Meta-Analysis

Introduction

There is extensive evidence for the benefits of physical activity (PA) in managing chronic conditions given their impact on fitness, mobility and general health (1, 2). Interventions aiming to address physical inactivity do not appear to have been implemented in any meaningful way (3). This may in part be due to the limited availability of clinic-based, face-to-face interventions (4, 5) to address the unique needs of this population. Alternative methods of delivering PA and exercise (PA&E) interventions need to be explored.

Digital technologies and the internet offer a medium to deliver and support PA&E interventions. These can be defined as interventions that are delivered via a digital platform to support or encourage a person to perform PA or exercise, usually with the aims of improved health outcomes. They provide a mode of health care delivery for people who find standard care inaccessible due to physical, economic, or social barriers (6, 7). Advancements in technology and digital content have allowed the development of digital therapeutic interventions that encourage people to use them with minimal support. These interventions have minimal to no ongoing human involvement in their set up and can be delivered automatically. Applications, incorporating behavior change elements and persuasive features (8–11), can be incorporated into devices to offer interactive and personalized approaches (12).

Previous reviews have investigated the effectiveness of digital PA interventions in the general adult population and meta-analyses have demonstrated positive effects on PA (6, 13). However, people living with chronic health conditions face unique challenges accessing and undertaking PA and exercise. They express a desire for specialist knowledge; concern that exercising may exacerbate symptoms; and transport issues (7, 14–17). A previous review by Bossen and colleagues (18) investigated the use of web-based interventions with minimal human contact, designed to increase PA in people living with chronic health conditions. They reported mixed results with no clear conclusion.

Several factors mean that an updated review is warranted. First, all studies in the review by Bossen et al. (18) were published between 2005 and 2010. Innovations in technology and increasing acceptance of its use in therapeutic interactions have led to a change in the definition of minimal human contact. Second, the review did not include exercise-based interventions. Exercise is a subcategory of PA (19) and has proven benefits in many chronic health populations (20, 21). Therefore, in this review, our reporting of PA interventions will include exercise interventions. There has been no review to date investigating digital PA&E interventions that include behavioral intervention features, for people living with chronic health conditions, delivered with minimal human contact.

The aim of this systematic review was to determine the effectiveness of self-guided digital PA&E interventions to improve PA&E related outcomes for people living with chronic health conditions. A secondary aim was to determine key behavioral intervention features that were used in the selected studies.

Methods

The protocol was registered with PROSPERO; CRD 42019132464. The review has been conducted and is reported following the guidelines of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement Moher et al. (22).

Eligibility Criteria

Papers were included if they met the criteria as outlined in Table 1.

TABLE 1

Table 1. Eligibility criteria of papers.

The key elements in Table 1 were used to devise the search strategy. Specifically, key terms were derived for “intervention” (digital exercise) and “study design” (randomized controlled trial). The search strategy did not include search terms for “population” and “outcome” to keep the reach as broad as possible. The criteria for population and outcome were applied during the screening process. Search terms were combined using Boolean, wildcard, truncation, and proximity searching. The search strategy was tailored to specific databases. Table 2 shows the search conducted using an OVID database.

TABLE 2

Table 2. Search concepts and terms using OVID.

Databases

Literature searches were conducted in the following databases up until the end of December 2019. This was updated to include any new publications to the end of December 2021: Cumulated Index to Nursing and Allied Health Literature (CINAHL), MEDLINE, SPORTSDiscus through EBSCO Health Database, Allied and Complementary Medicine (AMED), Evidence Based Medicine (EBM) Reviews—Cochrane Methodology, EBM Reviews—Health Technology Assessment, PsycINFO, Ovid MEDLINE(R) through OVID, Scopus, clinicaltrials.gov, Cochrane Central Register of Controlled Trial (CENTRAL), and in Web of Science. The PEDro database was searched (using simplified broad key terms). Reference lists of relevant reviews and articles were also hand searched.

Data Extraction

All citations returned in the search were downloaded and saved into EndNote X8. Duplicates were removed and then titles screened by VS, according to the pre-defined inclusion criteria. A selection of titles (the first 100) was independently screened by a second assessor (NS). Any disagreements were reviewed and discussed to ensure consensus was reached. A third assessor (NK) acted as arbitrator. Thereafter, VS and NS would meet after every 200 titles. There was high agreement as to which titles warranted further review. This process refined selection criteria. The abstracts and then full texts of studies identified as probably meeting the inclusion criteria were reviewed by VS in consultation with NS to confirm the final set of included studies.

Key details from each of the included studies were recorded in a data extraction table in Excel. These included: author and country; participant numbers and characteristics; study design; treatment intervention parameters including duration, frequency, and follow up. The synthesis of findings from key papers in the area (25–29) were used to create a framework to guide data extraction and included: theoretical underpinning for the intervention; instruction on how to perform the PA or exercise; recording and tracking of PA or exercise; the use of goal setting; the use of action and coping planning; the type, use, and delivery of feedback and monitoring; the use and delivery of prompts; the use of any additional online PA or exercise resources; the use of PA or exercise testimonials; and the use of gamification. PA&E related outcome measures and results at the end of the intervention, and at follow up, if reported, were also recorded. If multiple impairment level outcomes were measured, only outcomes that have previously been shown to be correlated with the construct being measured (e.g., plantar flexion strength with walking) were extracted. For studies that were comparing more than two arms, data from arms comparing self-guided interventions to a control were included in the analysis.

Risk of Bias

Risk of bias for each of the included studies was categorized as low, having some concerns or high, drawing on the revised Cochrane risk-of-bias tool for randomized trials (RoB 2) (30). Initially, two studies were scored independently by two authors (VS, NK). Scores were compared and key points of disagreement were discussed, to improve interrater agreement on interpretation of the RoB 2 criteria. Following that, all included studies were independently assessed by both authors. Rates of agreement were calculated and are reported below. Disagreements were discussed to achieve consensus, with NS acting as arbitrator.

Data Synthesis

Data were narratively synthesized focusing on the characteristics of the studies and outcomes. Meta-analysis was conducted on all studies that used PA as an outcome. The PA outcome measures were categorized as self-report or instrumented, and the data from each were pooled separately for meta-analysis.

A summary of intervention effects for each study was obtained by calculating Hedge's g standardized mean differences (SMD)s, 95% confidence intervals (CI)s, standard deviations (SD)s and effect sizes (ES). The Hedge's g values were calculated from the post-intervention time points while accounting for the pre-intervention differences. A positive ES indicates a result in favor of the intervention and a negative in favor of the control. When insufficient data were available for analysis, study authors were contacted. If the data were not received or could not be computated from published material, it was not included in any further analysis and was noted as not reported (NR). If standard errors or confidence intervals were presented instead of standard deviations (SD), SDs were calculated using recommended formulae (31). When required, means and SD were approximated from figures using WebPlotDigitizer (32, 33).

Given the clinical heterogeneity of the included studies, both fixed effects and random-effects models were considered for pooling PA data. The extent of heterogeneity was determined using a hypothesis test based on generalized Cochran's Q-statistic (34). High heterogeneity was assumed when the Q-test coincided with a significant value (p < 0.05) (35) in which case, a random effects model was used. I² statistic was presented if the random effects model was chosen (36). Meta-analysis results were reported as pooled Hedge's g and 95% CIs. Hedges'g ≤0.2, ≥0.5, and ≥0.8 were interpreted as small, medium, or large, respectively (37). A CI which did not overlap zero was considered statistically significant. An intervention was interpreted as effective at improving a construct when the estimated effect size was positive and had a 95% CI which did not cross zero. A category of interventions was considered effective at improving a construct when the meta-analysis effect size was positive and had a 95% CI which did not cross zero. Forest plots were also generated for pooled data. Unpooled data were presented in table format, allowing comparisons between each outcome. Analyses were performed in R (38) using the metafor package (35). UR and VS contributed to and confirmed the synthesis of the extracted data. Discrepancies in data synthesis were discussed amongst the authors until consensus were reached, with NK and NS serving as arbitrators.

With respect to the secondary aim, given the diversity of intervention features, context, and population, it was not possible to make a direct link between intervention features and outcome. Therefore, key behavioral intervention features of included studies were recorded and tabulated.

Results

Selection of Studies

A flow diagram of the identification, screening and selection of papers is presented in Figure 1. There was 98% agreement with the co-author (NS) during the screening phase, discrepancies were resolved by discussion. Following the full text review, 177 papers were excluded because they did not meet the a priori criteria (see Supplementary File 1 for a table of excluded papers). Three papers required discussion with arbitrator (NK). Sixteen papers met the criteria and were included in the review with ten included in the meta-analysis.

FIGURE 1

Figure 1. Flow chart summarizing the study selection process.

Risk of Bias

The risk of bias assessment for included studies is available in Supplementary File 2. Agreement between reviewers was high (92%). Consensus was reached where there was discrepancy. Overall, no studies were assessed as having a low risk of bias. The majority were assessed as having some risk of bias (39–48), with the remaining five judged to be at high risk of bias in at least one domain (49–53). As shown in the weighted summary plot (Figure 2), this could be attributed to having potential bias in the selection of reported results. Many studies did not have a published protocol, making it difficult to determine if analyses were carried out according to a pre-specified plan. Studies were also at risk of bias due to outcomes of interest being self-reported by participants who were unblinded to intervention status. Missing outcome data (for example, participants lost to follow up) also contributed to the bias.

FIGURE 2

Figure 2. Weighted summary plot of the overall type of bias encountered in the included studies.

Characteristics of Selected Studies

Included papers were published between 2013 and December 2021 (end of collection period). Publications in this research area have been increasing recently with half published in the last 2 years (40, 44–46, 51–54). Studies were conducted in Germany (41, 44), the Netherlands (39, 45), the United Kingdom (49, 54), Korea (42, 52), Australia (40), Brazil (48), Canada (53), China (47), Hong Kong (46), New Zealand (43), Norway (50), and the United States (51).

Interventions were compared to usual care or wait list groups in six studies (39, 41, 43, 46, 49, 54), to paper based interventions in four studies (42, 44, 47, 48), to another form of online intervention in three studies (40, 52, 53) and to a supervised intervention (50). One study used the intervention of interest (minimal contact self-guided intervention) as the comparison group when testing versions of online and blended interventions (51). Table 3 contains characteristics of the included studies.

TABLE 3

Table 3. Characteristics of the randomized controlled trials, participants, and interventions.

Characteristics of Participants

Participants were recruited from primary health care, community settings and online databases. The number of participants within studies ranged from 24 to 438, with a combined total of 2,439 participants across studies (median n = 93). The average age of participants across studies was 57.1 years with the mean age range from 43 to 67 years. Participants presented with the following chronic conditions: breast cancer (40, 42, 50), dizziness and vestibular syndrome (54, 55), heart disease (43, 46), chronic obstructive pulmonary disease (49, 52), fibromyalgia (48), hypertension (53), inflammatory rheumatic diseases (50), metabolic syndrome (47), progressive multiple sclerosis (44), osteoarthritis (39), and venous leg ulcers (51). Table 3 provides details for all included studies.

Outcome Measures

Outcomes of interest were those related to the PA or exercise interventions that measured an activity or body function (24). These were grouped into change in PA that was self-reported and change in PA that was measured by instrumentation. Changes in body functions and symptoms were also reported. Change in PA was meta-analyzed in ten of the included studies. This was measured using self-reported questionnaires (39–44, 46, 47) and instrumented devices (accelerometers, pedometers or wearable devices) (39, 44, 49, 53). The remaining relevant outcomes varied considerably among studies. Therefore, they were broadly grouped, as per the ICF framework (24), into change in body functions and symptoms and included measures such as walking endurance, measured by the 6-min walk test (44, 51, 52), strength (44, 51), vestibular symptoms (54, 55), peak oxygen uptake (43, 50), pain (48), foot and ankle mobility (51), dyspnea (49, 52), and range of motion (48, 49, 51–53). Self-reports of perceived effect and self-efficacy of exercise were also measured (39, 46, 47). Bossen et al. (39) and Kwon et al. (52) used the Knee Osteoarthritis Outcome Score and Hip Osteoarthritis Outcome Score. These measure symptoms, activities of daily living and quality of life; the scoring of each outcome precludes the ability to report these constructs separately.

Characteristics of the Interventions

Modes of Delivery

There were a variety of modes of delivery and technologies used across interventions. Eleven of the interventions involved computers (39–43, 46, 49, 52–55). Mobile or smartphones were used in nine studies (41–44, 47, 48, 50–52). Almost half the studies were testing a website specific to the study (39, 41–43, 46, 54, 55) while three used existing platforms (40, 52, 53). Applications specifically designed for the study were used in seven studies and involved a smartphone or computer (44, 47–52). Communication with participants was carried out via automated email (39, 46, 49, 53–55), automated SMS text messaging (41–43) or the study app features (47, 48, 50–52). Biosensors were used in five studies (39, 43, 49, 51, 52).

Parameters of Intervention Delivery

Table 3 provides details on key intervention parameters. Intervention duration ranged from a one-off session to 24 weeks. The expected frequency of the exercise or PA&E performance varied. This ranged from a one-off instructional session (40) to asking participants to participate repeatedly throughout the day (51, 54, 55), once daily (44, 49, 52), several times a week (42, 43, 46, 50, 53), or as determined by the participant (39, 41, 52).

Theoretical Basis

Just over half of the included studies reported using theory to inform their interventions, with three referencing more than one theory. The transtheoretical model was the most frequently cited (40, 42, 53). Self-regulatory theory, cognitive behavioral theory, social cognitive theory, exposure-based behavior principles, implementation-intention-based principles, health belief model and operant behavior principles were also each used once (39–41, 43, 46, 47, 54, 55). No underpinning theory was reported for six of the self-guided interventions (44, 48–52).

Behavioral Intervention Strategies

The interventions used a combination of behavioral strategies and features to support PA&E behavior (see Table 4). A commonly used feature was instruction on how to perform the PA or exercise (n = 14). Variations included generalized automated PA or exercise information, tailored exercise provision, and tracking and recording of PA or exercise performed. The use of feedback and monitoring (n = 13) was also used in most interventions. Other strategies involved goal setting (n = 11) and the use of prompting features (n = 10). Fewer than half the interventions incorporated action and coping plans (n = 5), online resources of supplemental PA or exercise information (n = 4) or testimonials or case studies (n = 5). None of the included studies used gamification approaches. Most interventions employed several strategies concurrently.

TABLE 4

Table 4. Behavioral intervention features and strategies used.

Effectiveness of Interventions

Effectiveness was measured by change in PA&E and change in body function and symptoms. These were defined and measured in a variety of ways.

Physical Activity

Change in self-reported PA was the outcome of interest in eight of the included studies (39, 41–44, 46, 47). Figure 3 details the Hedge's g between group difference, with 95% CI, at the earlier assessment point, taken at the end of intervention (which ranged from 4 to 24 weeks). The Hedge's g between groups favored the intervention group with a small, estimated effect of 0.2 [95% CI (0.1, 0.3)].

FIGURE 3

Figure 3. Self-reported PA at end of intervention.

Four of these studies (39, 40, 46, 47) continued to measure self-reported PA at a follow up point (12–52 weeks from the end of intervention). Pooled results (39, 40, 46, 47), demonstrate the effect of the intervention. The Hedge's g between groups continued to favor the intervention with an increased estimated effect [Hedge's g 0.3, 95% CI (0.2, 0.5)] (see Figure 4).

FIGURE 4

Figure 4. Self-reported PA at follow up.

Change in PA, measured by instruments such as accelerometers or pedometers was described in four studies (39, 44, 49, 53). The Hedge's g between groups and 95% CI at the completion of the intervention are displayed in Figure 5. The small treatment effect on PA was in favor of the intervention group (Hedge's g 0.3) but the CI crossed the no effect line [95% CI (−0.2, 0.9)] and heterogeneity was 77.7%.

FIGURE 5

Figure 5. Instrumented measurement of PA at end of intervention.

Body Functions and Symptoms

Meta-analysis was not conducted on outcomes related to body functions and symptoms due to the high heterogeneity of measures employed. There was variation in the size of effect and statistical significance across the studies. Refer to Supplementary File 3 for details of each study, the extracted outcome measures, assessment points, mean group differences and estimated effects. Despite studies showing some risk of bias, there are some results that were of note. Effects (Hedge's g 0.3–0.8) were reported in two separate studies evaluating 6-week vestibular interventions at the 12- and 24-week post intervention time points (54, 55). Both interventions had a theoretical basis and incorporated several behavioral strategies. There was also evidence of a small positive effect of graded-activity on physical function at 12 weeks [Hedge's g 0.4, 95% CI (0.1, 0.7)] but not at the 52-week follow up in a 9-week PA intervention for people with knee and hip OA (39). Self-perceived effect and exercise self-efficacy were measured in two studies (39, 47). A moderate effect [Hedge's g 0.6, 95% CI (0.3, 0.9)] was noted for self-perceived effect after the 9-week PA intervention. Similarly, a moderate effect was seen in self-efficacy for exercise [Hedge's g 0.5, 95% CI (0.1, 1)] at the end of a 4-week intervention and maintained [Hedge's g 0.6, 95% CI (0.2, 1.2)] at the 12 week follow up. Theoretical underpinnings and behavioral features were also incorporated in these interventions. Studies that did not have a theoretical basis did not report any effects.

When looking across study exercise parameters, there was a large variation in dosage (amount x frequency x duration) of prescribed exercise. There did not appear to be a clear relationship between dosage and effect.

Discussion

Principle Findings

This is the first systematic review to conduct a meta-analysis of outcomes following self-guided digital PA&E interventions in people living with chronic conditions. This work advances the evidence in digital interventions for PA&E. The more stringent inclusion criteria for minimal contact used in this review meant that none of the studies in the review by Bossen and colleagues (18) met the threshold to be included.

The findings of our review indicate a positive effect of self-guided digital interventions on PA that were seen at the end of intervention and sustained at follow up, for people living with chronic health conditions. These findings differ from the findings of Bossen and colleagues of conflicting evidence on the effectiveness of web-based PA interventions in this population (18). Our findings reflect the developments in technology that have enabled persuasive and engaging elements to be embedded into digital interventions (56, 57). These may help overcome some of problems of low uptake and adherence (58).

There was a small but significant effect seen with self-reported PA at the end of interventions. Direito and colleagues (9), investigating the effect of mHealth technologies on self-reported PA in healthy participants, found a comparable effect that was not significant. Kwan et al.'s (59) review of eHealth interventions promoting PA in older adults found that self-report PA was significantly increased compared with the control groups. However, since the interventions in both reviews included contact and personal consultation, a cautious approach to comparison is needed.

Objectively measured change in PA in the current review showed a small effect that was not significant. This has also been reported by others (9, 60) who have found insignificant effects favoring mobile and app-based interventions that were not strictly self-guided. There is an assumption that objective PA measures more accurately reflect actual changes in PA. However, Kayes and McPherson (61) argue measurement tools such as accelerometers and pedometers have not always been validated in people living with chronic health conditions. Despite significant advances in the objective measurement of PA in these populations, questions remain regarding the validity and reliability of these devices in some groups, such as people with slow walking speeds or those with higher levels of disability (62, 63).

Our findings of sustained PA improvement beyond the intervention, is in contrast with previous work. In-person PA interventions have been shown to increase PA levels in the short to medium term (7 weeks to 1 year) but not in the long term (at or over 1 year), in community dwelling adults with or without a long-term health condition (64, 65). Our findings are supported by Davies et al. (8) who conducted a review of web-based PA interventions and found an overall small but significant mean effect of sustained PA when looking at a follow up of at least 6 months after the intervention. Regardless of the length of follow up, the persistence of change seen, despite the removal of intervention, is encouraging.

The prolonged effect observed in some studies may be linked to the use of behavioral strategies in the interventions. The most common intervention strategies and features used were instructions on how to perform the PA or exercise, goal setting, and the use of feedback and monitoring. These align with the behavior change techniques (BCT) taxonomy clusters proposed by Michie and colleagues (27). The self-guided interventions with the larger effect sizes employed strategies from at least three of these clusters (39, 42, 46, 53–55). This finding is supported by research showing that the use of techniques from three BCT clusters are needed to produce effects on PA in face-to-face interventions (66, 67). For self-guided digital interventions, employing self-regulatory techniques may be the most effective way to support PA engagement (28, 29, 68). Michie et al. (68) found that interventions combining self-monitoring (using feedback) in combination with other features that encouraged self-regulation (goal setting, action and coping plans) were more likely to lead to intervention effectiveness (69). These strategies can influence behavior via mechanisms such as problem-solving, promoting self-efficacy or diminishing the impact of barriers to behavior change (70). They may also address the intention behavior gap (71). In this review, fewer than half the interventions reviewed used action and coping plans or other strategies that promote self-regulation. The lack of theoretical underpinning found in many of the studies may help explain their choice of intervention features and non-significant results. While we did not explicitly set out to determine the effectiveness of behavioral features, interventions that were effective did include behavioral features. Determining which specific features might best be implemented to support sustained use and engagement with self-guided interventions would be a future research direction.

The lack of an apparent relationship between exercise dose and effect suggests that the prescribed dose may be only one factor of an intervention that influences participant outcomes. The behavioral strategies embedded in the interventions influence exercise completion and, therefore, are an important factor in the reported interventions. The lack of reporting of exercise completion makes it difficult to appreciate how these factors contribute to outcome.

Strengths and Limitations

This review focused on investigating the effect of digital self-guided interventions on PA&E and fills a gap in the literature in this growing area. We included all forms of digital interventions that could be delivered remotely, without ongoing human interaction. The comprehensive search, using several databases, identified studies previously not included in similar reviews, and the broad range of outcomes make the findings generalizable to a wider range of populations, interventions, and environments. We identified behavioral strategies using well-established frameworks (27) that allows for transparency and a clear understanding of intervention features that may help to explain reported effects. This review has also reported metrics for discrete end of intervention and follow up outcomes and meta-analyses for PA outcomes for self-guided digital PA interventions for people living with chronic conditions.

People living with chronic conditions face additional challenges in undertaking PA&E compared to a healthy population. Acknowledgment of the inherent heterogeneity that comes with each health conditions should sit alongside any the generalization of these findings, particularly when a significant portion of the people included in this review were living with cardiorespiratory, neurological or oncology conditions and were under the age of 60. Despite this, there is much we can learn by looking across populations, particularly when those populations share characteristics beyond diagnosis such as chronicity and complex barriers to engagement with PA&E.

To be included in the review, papers needed to have been peer reviewed to address the question of effectiveness, which may have introduced publication bias. Other criteria were that interactions with the intervention had to be automatically generated. Screening of the papers for this criterion was not straightforward due to the lack of a standard definition of what constitutes minimal contact. We selected papers based on our operational definition and consequently many interesting and valuable digital PA&E interventions were excluded. For included papers, data were frequently incompletely or inconsistently reported, and some were analyzed as intention to treat while others were not. This made analysis of the results difficult and necessitated computation using reported data. The wide variation in health-related body function and symptoms outcomes prevented us from pooling this data. Therefore, we synthesized the results in a way that allowed the magnitude and range of effects to be appreciated.

The studies included in this review demonstrated some or high risk of bias and so findings should be interpreted with caution. The nature of the minimal contact digital intervention creates increased opportunities for bias with attrition, and the reporting of outcomes remotely.

Future Recommendations

Future work should continue to investigate which intervention features and exercise parameters lead to the best effect. Effort should continue to ensure complete reporting of the intervention, the behavioral interventions used, and treatment fidelity, including recording the participant's completion of the prescribed dose. Authors also need to ensure that adverse events are explicitly sought and consistently reported. This was not the case in many of the studies included in the current review. For interventions that are self-directed and have minimal contact with health professionals, understanding intervention safety is an important component of the trialing phase.

The proportion of participants lost to follow up in the studies in this review demonstrates that maintaining engagement with low contact digital interventions is challenging (72, 73). Given the added barriers experienced by these people (74), tailoring the intervention to the individual needs to be considered and addressed (75).

Researchers are encouraged to broaden the scope of populations involved in this type of research. For the current review, the population of interest focused on people living with chronic conditions that included a range of non-communicable diseases. Of note was the lack of studies involving people living with disabilities, a large group who also would benefit from self-guided and digital interventions. Finally, the participant's voice is notably absent from much of this research.

Our research group is currently exploring many of these areas. For example, we are applying an interpretive descriptive study design to explore what makes a self-guided digital intervention more acceptable to users, with the aim of developing interventions which increase uptake and engagement (76). The findings from this current review and the interpretive descriptive study will inform our development a self-guided digital intervention to help treat shoulder pain for people living with spinal cord injury (76).

Conclusion

This review found a positive effect in favor of self-guided digital PA&E interventions on PA outcomes and a selected number of body functions and impairments at the end of intervention and at follow up, in people living with chronic conditions. Interventions that employ behavioral strategies, underpinned by a theoretical framework, have the potential to support self-regulation and sustained PA at least as well as interventions with supervision.

Data Availability Statement

The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author.

Author Contributions

VS, NK, and NS contributed to conception, design of the study, and analyzed intervention features. VS performed the initial search and organized the database. VS and NS reviewed study selection. VS and NK analyzed risk of bias of selected studies. UR provided statistical support. All authors read and approved the final manuscript.

Conflict of Interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Publisher's Note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Acknowledgments

The authors wish to thank Auckland University of Technology for their support of this study.

Supplementary Material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fresc.2022.925620/full#supplementary-material

References

1. Fransen M, McConnell S, Harmer AR, Van der Esch M, Simic M, Bennell KL. Exercise for osteoarthritis of the knee. Cochrane Database Syst Rev. (2015). doi: 10.1002/14651858.CD004376.pub3 (accessed June 09, 2022).

PubMed Abstract | CrossRef Full Text | Google Scholar

2. Furmaniak AC, Menig M, Markes MH. Exercise for women receiving adjuvant therapy for breast cancer. Cochrane Database Syst Rev. (2016) 2016:CD005001. doi: 10.1002/14651858.CD005001.pub3

PubMed Abstract | CrossRef Full Text | Google Scholar

3. Lee W-C, Ory MC. The engagement in physical activity for middle-aged and older adults with multiple chronic conditions: findings from a community health assessment. J Aging Res. (2013) 2013:8. doi: 10.1155/2013/152868

PubMed Abstract | CrossRef Full Text | Google Scholar

4. Reuben D, Cassel C. Physician stewardship of health care in an era of finite resources. J Am Med Assoc. (2011) 306:430–1. doi: 10.1001/jama.2011.999

PubMed Abstract | CrossRef Full Text | Google Scholar

5. Walsh N. Global initiatives in rehabilitation medicine. Arch Phys Med Rehabil. (2004) 85:1395–402. doi: 10.1016/j.apmr.2004.04.030

PubMed Abstract | CrossRef Full Text | Google Scholar

6. Davies C, Spence J, Vandelanotte C, Caperchione CM, Mummery WK. Meta-analysis of internet-delivered interventions to increase physical activity levels. Int J Behav Nutr Phys Act. (2012) 5:1–13. doi: 10.1186/1479-5868-9-52

PubMed Abstract | CrossRef Full Text | Google Scholar

7. Hoaas H, Andreassen H, Lien L, Hjalmarsen A, Zanaboni P. Adherence and factors affecting satisfaction in long-term telerehabilitation for patients with chronic obstructive pulmonary disease: a mixed methods study. BMC Med Inform Decis Mak. (2016) 16:1–14. doi: 10.1186/s12911-016-0264-9

PubMed Abstract | CrossRef Full Text | Google Scholar

8. Bennell K, Nelligan R, Schwartz S, Kasza J, Kimp A, Crofts S, et al. Behavior change text messages for home exercise adherence in knee osteoarthritis: randomised trial. J Med Internet Res. (2020) 22:p.e21749. doi: 10.2196/21749

PubMed Abstract | CrossRef Full Text | Google Scholar

9. Direito A, Carraca E, Rawstorn J, Whittaker R, Maddison R. mHealth tehnologies to influence physical activity and sedentary behaviors: behavior change techniques, systematic review and meta-analysis of randomized controlled trials. Ann Behav Med. (2017) 51:226–39. doi: 10.1007/s12160-016-9846-0

PubMed Abstract | CrossRef Full Text | Google Scholar

10. Matthews J, Win K, Oinas-Kukkonen H, Freeman M. Persuasive technology in mobile applications promoting physical activity: a systematic review. J Med Syst. (2016) 40:1–13. doi: 10.1007/s10916-015-0425-x

PubMed Abstract | CrossRef Full Text | Google Scholar

11. Win K, Roberts M, Oinas-Kukkonen H. Persuasive system features in computer-mediated lifestyle modification interventions for physical activity. Inform Health Soc Care. (2019) 44:376–404. doi: 10.1080/17538157.2018.1511565

PubMed Abstract | CrossRef Full Text | Google Scholar

12. van Gemert-Pijnen L, Kelders SM, Kip H, Sanderman R. EHealth research, theory and development: a multidiscliplinary approach: Taylor & Francis Group. (2018). doi: 10.4324/9781315385907

CrossRef Full Text | Google Scholar

13. Jahangiry L, Farhangi MA, Shab-Bidar S, Rezaei F, Pashaei T. Web-based physical activity interventions: a systematic review and meta-analysis of randomized controlled trials. Public Health. (2017) 152:36–46. doi: 10.1016/j.puhe.2017.06.005

PubMed Abstract | CrossRef Full Text | Google Scholar

14. Cowan RE, Nash MS, Anderson KD. Exercise participation barrier prevalence and association with exercise participant status in individuals with spinal cord injury. Spinal Cord. (2013) 51:27–32. doi: 10.1038/sc.2012.53

PubMed Abstract | CrossRef Full Text | Google Scholar

15. de Hollander EL, Proper KI. Physical activity levels of adults with various physical disabilities. Prev Med Rep. (2018) 10:370–6. doi: 10.1016/j.pmedr.2018.04.017

PubMed Abstract | CrossRef Full Text | Google Scholar

16. Jaarsma E, Dijkstra PU, Geertzen JHB, Dekker R. Barriers to and faciliators of sports participation for people with physical disabilities: a systematic review. Scand J Med Sci Sports. (2014) 24:871–81. doi: 10.1111/sms.12218

PubMed Abstract | CrossRef Full Text | Google Scholar

17. Jetha A, Faulkner G, Gorczynski P, Arbour-Nicitopoulos K, Ginis KAM. Physical activity and individuals with spinal cord injury: accuracy and quality of information on the internet. Disabil Health J. (2011) 4:112–20. doi: 10.1016/j.dhjo.2010.07.001

PubMed Abstract | CrossRef Full Text | Google Scholar

18. Bossen D, Veenhof C, Dekker J, de Bakker D. The effectiveness of self-guided web-based physical activity interventions among patients with a chronic disease: a systematic review. J Phys Act Health. (2014) 11:665–77. doi: 10.1123/jpah.2012-0152

PubMed Abstract | CrossRef Full Text | Google Scholar

19. Caspersen CJ, Powell KE, Christenson GM. Physical activity, exercise, and physical fitness: definitions and distinctions for health-related research. Public Health Rep. (1985) 100:126–31.

PubMed Abstract | Google Scholar

20. Lee I, Shiroma E, Lobelo F, Puska P, Blair SN, Katzmarzyk P, et al. Impact of physical inactivity on the world's major non-communicable diseases. Lancet. (2012) 380:219–29. doi: 10.1016/S0140-6736(12)61031-9

PubMed Abstract | CrossRef Full Text | Google Scholar

21. Warburton DER, Bredin SSD. Health benefits of physical activity: a systematic review of current systematic reviews. Curr Opin Cardiol. (2017) 32:541–56. doi: 10.1097/HCO.0000000000000437

PubMed Abstract | CrossRef Full Text | Google Scholar

22. Moher D, Liberati A, Tetzlaff J, Altman DG, The The PRISMA Group. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. PLoS Med. (2009) 6:264–269. doi: 10.1371/journal.pmed.1000097

PubMed Abstract | CrossRef Full Text | Google Scholar

23. Bernell S, Howard SW. Use your words carefully: what is a chronic disease? Front Public Health. (2016) 4:159. doi: 10.3389/fpubh.2016.00159

PubMed Abstract | CrossRef Full Text | Google Scholar

24. World Health Organisation. Towards a Common Language for Functioning, Disability and Health. Geneva: WHO (2002).

Google Scholar

25. Eysenbach G. CONSORT-EHEALTH: improving and standardizing evaluation reports of web-based and mobile health interventions. J Med Internet Res. (2011) 13:25–34. doi: 10.2196/jmir.1923

PubMed Abstract | CrossRef Full Text | Google Scholar

26. Kelders SM, Kok RN, Ossebaard HC, Van Gemert-Pinjnen JE. Persuasive system design does matter: a systematic review of adherence to web-based interventions. J Med Internet Res. (2012) 14:e2104. doi: 10.2196/jmir.2104

PubMed Abstract | CrossRef Full Text | Google Scholar

27. Michie S, Richardson M, Johnston M, Abraham C, Francis JJ, Hardeman W, et al. The behavior change technique taxonomy (v1) of 93 hierarchically clustered techniques: buidling an international consensus for the reporting of behavior change interventions. Ann Behav Med. (2013) 46:81–95. doi: 10.1007/s12160-013-9486-6

PubMed Abstract | CrossRef Full Text | Google Scholar

28. Morrison LG, Yardley L, Powell J, Michie S. What design features are used in effective e-health interventions? A review using techniques from critical interpretive synthesis. Telemed J E Health. (2012) 18:137–44. doi: 10.1089/tmj.2011.0062

PubMed Abstract | CrossRef Full Text | Google Scholar

29. Webb TL, Joseph J, Yardley L, Michie S. Using the internet to promote health behavior change: a systematic review and meta-analysis of the impact of theoretical basis, use of behavior change techniques, and mode of delivery on efficacy. J Med Internet Res. (2010) 12:18. doi: 10.2196/jmir.1376

PubMed Abstract | CrossRef Full Text | Google Scholar

30. Sterne JAC, Savovic J, Page M, Elbers R, Blencowe N, Boutron I, et al. RoB 2: a revised tool for assessing risk of bias in randomised trials. BMJ. (2019) 366:1–8. doi: 10.1136/bmj.l4898

PubMed Abstract | CrossRef Full Text | Google Scholar

31. Borenstein M. Effect sizes for continuous data. In: Cooper H, Hedges LV, Valentine JC, editors. The Handbook of Research Synthesis and Meta-Analysis. 2nd ed. New York, NY: Russell Sage Foundation (2009).

Google Scholar

32. Drevon D, Fursa S, Malcolm A. Intercoder reliability and validity of WebPlotDigitizer in extracting graphed data. Behav Modif. (2017) 41:323–39. doi: 10.1177/0145445516673998

PubMed Abstract | CrossRef Full Text | Google Scholar

33. Rohatgi A. WebPlotDigitizer. California, USA2020. Available online at: https://automeris.io/WebPlotDigitizer (accessed December 15, 2020).

Google Scholar

34. Bowden J, Tierney J, Copas A, Burdett S. Quantifying, displaying and accounting for heterogeneity in the meta-analysis of RCTs using standard and generalised Q statistics. BMC Med Res Methodol. (2011) 11:1–12. doi: 10.1186/1471-2288-11-41

PubMed Abstract | CrossRef Full Text | Google Scholar

35. Viechtbauer W. Conducting meta-analysis in R with the metafor package. J Stat Softw. (2010) 36:1–48. doi: 10.18637/jss.v036.i03

CrossRef Full Text | Google Scholar

36. Huedo-Medina TB, Sanchez-Meca J, Marin-Martinez F, Botella J. Assessing heterogeneity in meta-analysis: Q statistic or I² index? Psychol Methods. (2006) 11:193. doi: 10.1037/1082-989X.11.2.193

PubMed Abstract | CrossRef Full Text | Google Scholar

37. Cohen J. Statistical power analysis for the behavioral sciences: Routledge. (2013). doi: 10.4324/9780203771587

CrossRef Full Text | Google Scholar

38. Bunn A, Korpela M, R Core Team. R: A Language and Environment for Statistical Computing. Vienna: R Foundation for Statistical Computing (2018).

Google Scholar

39. Bossen D, Veenhof C, Van Beek KE, Spreeuwenberg PM, Dekker J, De Bakker DH. Effectiveness of a web-based physical activity intervention in patients with knee and/or hip osteoarthritis: randomized controlled trial. J Med Internet Res. (2013) 15:e257. doi: 10.2196/jmir.2662

PubMed Abstract | CrossRef Full Text | Google Scholar

40. Chapman J, Fletcher C, Flight I, Wilson C. Pilot randomized trial of a volitional help sheet-based tool to increase leisure time physical activity in breast cancer survivors. Br J Health Psychol. (2018) 23:723–40. doi: 10.1111/bjhp.12313

PubMed Abstract | CrossRef Full Text | Google Scholar

41. Holtdirk F, Mehnert A, Weiss M, Mayer J, Meyer B, Brode P, et al. Results of the optimune trial: a randomized controlled trial evaluating a novel internet intervention for breast cancer survivors. PLoS ONE. (2021) 16:e0251276. doi: 10.1371/journal.pone.0251276

PubMed Abstract | CrossRef Full Text | Google Scholar

42. Lee MK, Yun YH, Park HA, Lee ES, Jung KH, Noh DY. A Web-based self-management exercise and diet intervention for breast cancer survivors: pilot randomized controlled trial. Int J Nurs Stud. (2014) 51:1557–67. doi: 10.1016/j.ijnurstu.2014.04.012

PubMed Abstract | CrossRef Full Text | Google Scholar

43. Maddison R, Pfaeffli L, Whittaker R, Stewart R, Kerr A, Jiang Y, et al. A mobile phone intervention increases physical activity in people with cardiovascular disease: results from the HEART randomized controlled trial. Eur J Prev Cardiol. (2015) 22:701–9. doi: 10.1177/2047487314535076

PubMed Abstract | CrossRef Full Text | Google Scholar

44. Nasseri NN, Ghezelbash E, Zhai Y, Patra S, Riemann-Lorenz K, Heesen C, et al. Feasibility of a smartphone app to enhance physical activity in progressive MS: a pilot randomized controlled pilot trial over three months. PeerJ. (2020) 2020:e9303. doi: 10.7717/peerj.9303

PubMed Abstract | CrossRef Full Text | Google Scholar

45. Van Vugt V, Van der Wouden JC, Essery R, Yardley L, Twisk J, van der Horst H, et al. Internet based vestibular rehabilitation with and without physiotherapy support for adults aged 50 and older with a chronic vestibular syndrome in general practice: three armed randomised controlled trial. BMJ. (2019) 367:1–11. doi: 10.1136/bmj.l5922

PubMed Abstract | CrossRef Full Text | Google Scholar

46. Wong EML, Leung DYP, Chair SY, Sit JWH. Effects of a web-based educational support intervention on total exercise and cardiovascular risk markers in adults with coronary heart disease. Worldviews Evid Based Nurs. (2020) 17:283–92. doi: 10.1111/wvn.12456

PubMed Abstract | CrossRef Full Text | Google Scholar

47. Wong EML, Leung DYP, Tam HL, Wang Q, Yeung KW, Leung AYM. The effect of a lifestyle intervention program using a mobile application for adults with metabolic syndrome, versus the effect of a program using a booklet: a pilot randomized controlled trial. Clin Interv Aging. (2021) 16:633–44. doi: 10.2147/CIA.S303920

PubMed Abstract | CrossRef Full Text | Google Scholar

48. Yuan SLK, Couto LA, Marques AP. Effects of a six-week mobile app versus paper book intervention on quality of life, symptoms, and self-care in patients with fibromyalgia: a randomized parallel trial. Braz J Phys Ther. (2021) 25:428–36. doi: 10.1016/j.bjpt.2020.10.003

PubMed Abstract | CrossRef Full Text | Google Scholar

49. Crooks MG, Elkes J, Storrar W, Roy K, North M, Blythin A, et al. Evidence generation for the clinical impact of myCOPD in patients with mild, moderate and newly diagnosed COPD: a randomised controlled trial. ERJ Open Res. (2020) 6:1–10. doi: 10.1183/23120541.00460-2020

PubMed Abstract | CrossRef Full Text | Google Scholar

50. Haglo H, Wang E, Berg OK, Hoff J, Helgerud J. Smartphone-assisted high-intensity interval training in inflammatory rheumatic disease patients: randomized controlled trial. JMIR Mhealth Uhealth. (2021) 9:e28124. doi: 10.2196/28124

PubMed Abstract | CrossRef Full Text | Google Scholar

51. Kelechi TJ, Madisetti M, Prentice M, Mueller M. FOOTFIT physical activity mhealth intervention for minimally ambulatory individuals with venous leg ulcers: a randomized controlled trial. J Wound Ostomy Cont Nurs. (2020) 47:173–81. doi: 10.1097/WON.0000000000000631

PubMed Abstract | CrossRef Full Text | Google Scholar

52. Kwon H, Lee S, Jung EJ, Kim S, Lee JK, Kim DK, et al. An mHealth management platform for patients with chronic obstructive pulmonary disease (efil breath): randomized controlled trial. JMIR mhealth and uhealth. (2018) 6:e10502. doi: 10.2196/10502

PubMed Abstract | CrossRef Full Text | Google Scholar

53. Liu S, Brooks D, Thomas SG, Eysenbach G, Nolan RP. Effectiveness of user- and expert-driven web-based hypertension programs: an RCT. Am J Prev Med. (2018) 54:576–83. doi: 10.1016/j.amepre.2018.01.009

PubMed Abstract | CrossRef Full Text | Google Scholar

54. Geraghty AWA, Essery R, Kirby S, Stuart B, Turner D, Little P, et al. Internet-based vestibular rehabilitation for older adults with chronic dizziness: a randomized controlled trial in primary care. Ann Fam Med. (2017) 15:209–16. doi: 10.1370/afm.2070

PubMed Abstract | CrossRef Full Text | Google Scholar

55. van Vugt VA, van der Wouden JC, Bosmans JE, Smalbrugge M, van Diest W, Essery R, et al. Guided and unguided internet-based vestibular rehabilitation versus usual care for dizzy adults of 50 years and older: a protocol for a three-armed randomised trial. BMJ Open. (2017) 7:e015479. doi: 10.1136/bmjopen-2016-015479

PubMed Abstract | CrossRef Full Text | Google Scholar

56. van Gemert-Pijnen JEWC, Nijland N, van Limburg M, Ossebaard HC, Kelders SM, Eysenbach G, et al. A holistic framework to improve uptake and impact of ehealth technologies. J Med Internet Res. (2011) 13:e111. doi: 10.2196/jmir.1672

PubMed Abstract | CrossRef Full Text | Google Scholar

57. Yardley L, Ainsworth B, Arden-Close E, Muller I. The person-based approach to enhancing the acceptability and feasibility of interventions. Pilot Feasibility Stud. (2015) 137:1–7. doi: 10.1186/s40814-015-0033-z

PubMed Abstract | CrossRef Full Text | Google Scholar

58. Kohl L, Crutzen R, de Vries N. Online prevention aimed at lifestyle behaviors: a systematic review of reviews. J Med Internet Res. (2013) 15:e146. doi: 10.2196/jmir.2665

PubMed Abstract | CrossRef Full Text | Google Scholar

59. Kwan R, Salihu D, Lee P, Tse M, Cheung D, Roopsawang I, et al. The effect of e-health interventions promoting physical activity in older people: a systematic review and meta-analysis. Eur Rev Aging Phys Act. (2020) 17:1–17. doi: 10.1186/s11556-020-00239-5

PubMed Abstract | CrossRef Full Text | Google Scholar

60. Romeo A, Edney S, Plotnikoff RC, Curtis R, Ryan J, Sanders I, et al. Can smartphone apps increase physical activity? Systematic review and meta-analysis. J Med Internet Res. (2019) 21:e12053. doi: 10.2196/12053

PubMed Abstract | CrossRef Full Text | Google Scholar

61. Kayes NM, McPherson KM. Measuring what matters: does 'objectivity' mean good science? Disabil Rehabil. (2010) 32:1011–9. doi: 10.3109/09638281003775501

PubMed Abstract | CrossRef Full Text | Google Scholar

62. Sasaki JE, Sandroff B, Bamman M, Motl RW. Motion sensors in multiple sclerosis: narrative review and update of applications. Expert Rev Med Devices. (2017) 14:891–900. doi: 10.1080/17434440.2017.1386550

PubMed Abstract | CrossRef Full Text | Google Scholar

63. Sliepen M, Brandes M, Rosenbaum D. Current physical activity monitors in hip and knee osteoarthritis: a review. Arthr Care Res. (2017) 69:1460–46. doi: 10.1002/acr.23170

PubMed Abstract | CrossRef Full Text | Google Scholar

64. Kunstler B, Cook J, Freene N, Finch C, Kemp J, O'Halloran P, et al. Physiotherapist-led physical activity interventions are efficacious at increasing physical activity levels: a systematic review and meta-analysis. Clin J Sport Med. (2018) 28:304–15. doi: 10.1097/JSM.0000000000000447

PubMed Abstract | CrossRef Full Text | Google Scholar

65. Richards J, Hillsdon M, Thorogood M, Foster C. Face-to-face interventions for promoting physical activity. Cochrane Database Syst Rev. (2013) 2013:CD010392. doi: 10.1002/14651858.CD010392

PubMed Abstract | CrossRef Full Text | Google Scholar

66. McEwan D, Beauchamp MR, Kouvousis C, Ray C, Wyrough A, Rhodes RE. Examining the active ingredients of physical activity interventions underpinned by theory versus no stated theory: a meta-analysis. Health Psychol Rev. (2019) 13:1–17. doi: 10.1080/17437199.2018.1547120

PubMed Abstract | CrossRef Full Text | Google Scholar

67. Stockwell S, Schofield P, Fisher A, Firth J, Jackson SL, Stubbs B, et al. Digital behavior change interventions to promote physical activity and/or reduce sedentary behavior in older adults: a systematic review and meta-analysis. Exp Gerontol. (2019) 120:68–87. doi: 10.1016/j.exger.2019.02.020

PubMed Abstract | CrossRef Full Text | Google Scholar

68. Michie S, Abraham C, Whittington C, McAteer J. Effective techniques in health eating and physical activity interventions: a meta-regression. Health Psychol. (2009) 28:690–701. doi: 10.1037/a0016136

PubMed Abstract | CrossRef Full Text | Google Scholar

69. Conner M, Sandberg T, Norman P. Using action planning to promote exercise behavior. Ann Behav Med. (2010) 40:65–76. doi: 10.1007/s12160-010-9190-8

PubMed Abstract | CrossRef Full Text | Google Scholar

70. Lippke S, Wiedemann A, Ziegelmann J, Reuter T, Schwarzer R. Self-efficacy moderates the mediation of intentions into behavior via plans. Am J Health Behav. (2009) 33:521–9. doi: 10.5993/AJHB.33.5.5

PubMed Abstract | CrossRef Full Text | Google Scholar

71. Gollwitzer P. Implementation intentions: strong effects of simple plans. Am Psychol. (1999) 54:493–503. doi: 10.1037/0003-066X.54.7.493

CrossRef Full Text | Google Scholar

72. Eysenbach G. The law of attrition. J Med Internet Res. (2005) 7:e402. doi: 10.2196/jmir.7.1.e11

PubMed Abstract | CrossRef Full Text | Google Scholar

73. Leslie E, Marshall AL, Owen N, Bauman A. Engagement and retention of participants in a physical activity website. Prev Med. (2005) 40:54–9. doi: 10.1016/j.ypmed.2004.05.002

PubMed Abstract | CrossRef Full Text | Google Scholar

74. Bullard T, Ji M, An R, Trinh L, Mackenzie M, Mullen S. A systematic review and meta-analysis of adherence to physical activity interventions among three chronic conditions: cancer, cardiovascular disease, and diabetes. BMC Public Health. (2019) 19. doi: 10.1186/s12889-019-6877-z

PubMed Abstract | CrossRef Full Text | Google Scholar

75. Allen KD, Arbeeva L, Callahan LF, Golightly YM, Goode AP, Heiderscheit BC, et al. Physical therapy vs internet-based exercise training for patients with knee osteoarthritis: results of a randomized controlled trial. Osteoarthr Cartil. (2018) 26:383–96. doi: 10.1016/j.joca.2017.12.008

PubMed Abstract | CrossRef Full Text | Google Scholar

76. Stavric V, Saywell N, Kayes NM. Development of a self-guided web-based exercise intervention (SPIN) to treat shoulder pain in people living with spinal cord injury: protocol of a mixed methods study. BMJ Open. (2019) 9:e031012. doi: 10.1136/bmjopen-2019-031012

PubMed Abstract | CrossRef Full Text | Google Scholar