Development and Use of Mobile Messaging for Individuals With Musculoskeletal Pain Conditions: Scoping Review

Introduction

Background

Musculoskeletal conditions, those affecting the bones, muscles, and joints, are recognized as a global public health problem, although the prevalence and burden of healthy life lost are difficult to estimate with certainty because population studies are few [1]. Where representative studies have been conducted, they have consistently shown high prevalence of musculoskeletal conditions that increases with age and has a greater burden on female than male individuals [2-5]. In the Health Survey for England 2018, a total of 17% of adults reported having a long-term musculoskeletal condition (19.5% female vs 14.2% male), with prevalence increasing with age (4.7% at the ages of 16-24 years vs 39% at the age of ≥85 years). A total of 80% of people who reported having a long-term musculoskeletal condition also reported chronic pain (pain for >3 months), with 34.8% reporting pain that highly interfered with their life activities [4,5]. The Australian National Health Survey 2017 to 2018 reported that 29% of Australians were living with a chronic musculoskeletal condition (age standardized; adults aged ≥45 years: 51%; 55.3% female vs 47.3% male) [3]. musculoskeletal conditions were the second leading contributor to total burden of healthy life lost, equal to the burden of cardiovascular disease (13% of total burden in disability-adjusted life years), second only to cancer (18% of total burden) [2]. Prioritizing the delivery of effective treatments is necessary to address the substantial burden of musculoskeletal conditions.

With the ubiquity of consumer devices such as smartphones and tablets, technology may have a useful role to play in the management and self-management of musculoskeletal conditions; potentially improve accessibility of health care; and, in some circumstances, ease health system pressures. The use of technology for providing health-related activities is typically described as “digital health” and, more specifically, “mobile health” (mHealth) when referring to the use of mobile devices. While still a relatively new field, mHealth already has a considerable literature base, with examples of its use across most health disciplines and across the continuum of care from health promotion and prevention [6,7] to screening and diagnosis [8,9], therapy [10,11], and self-management [12,13] to cancer survivorship and palliative care [14,15]. While mHealth shows promise in improving aspects of health care, evidence to date is mixed, and caution is needed in interpreting the clinical value of mHealth for patients [16].

In this review, we focused on the development and use of mHealth for individuals with musculoskeletal pain conditions and specifically on health-related interactions that use text messaging as the delivery mechanism (SMS text messaging or messages provided via app-based push notifications), either alone or alongside another intervention. As one of the mobile technologies that have been established for longer, text messaging is familiar, easy to use, convenient, low cost, and available to anyone with a mobile device [17]. Messaging can be used as a vehicle to promote behavior change and guide self-management through prompts, reinforcement, reminders, activity recording, feedback, and adaptivity to the individual [17,18]. The effectiveness of messaging interventions has been assessed for a wide range of health problems, such as medication adherence and lifestyle change in diabetes; encouraging abstinence in smoking cessation; and, more recently, to encourage prevention behaviors during the COVID-19 pandemic [17,19-21].

In total, 2 previous reviews have explored the effectiveness of text messaging–based interventions for musculoskeletal conditions [22,23]. In a broad review of 19 randomized controlled trials (RCTs; 1086 participants) [23], 5 studies involved aspects of messaging [24-28], with 4 studies reporting improvements in pain [25-28] and functional disability [24-27] favoring digital interventions but not specifically favoring the messaging components [23]. A second review focused specifically on the effectiveness of text messaging–delivered interventions included 11 RCTs (1607 participants) [22]. Of the included studies, 5 assessed text messaging as an adjunct to usual care on treatment adherence and found improvements favoring text messaging [29-33]. In a further 5 RCTs, the effectiveness of text messaging as 1 component of a complex intervention was assessed [34-38], finding small but inconsistent effects on pain, functioning, adherence, and quality of life. In 1 RCT, text messaging was compared to telephone counseling, and similar effects on functioning were reported [39].

Objectives

These previous reviews focused on intervention effectiveness and synthesized data from RCTs only. The findings of observational studies have not been synthesized, and these studies may contain useful information to inform and, ultimately, improve the effectiveness and adoption of future musculoskeletal interventions delivered using text messaging. Furthermore, important characteristics of interventions, such as the configuration of digital content, method of presentation, dose, frequency, and preferences, have not been synthesized. Consequently, to inform the design, development, and evaluation of future messaging interventions for people with musculoskeletal pain, we need to explore the literature using a wider lens. Therefore, in this study focused on individuals with musculoskeletal pain conditions, we had three aims: (1) to map the literature related to the use of mobile messaging; (2) to identify information that could be useful in the design of future messaging interventions; and (3) to explore and summarize the findings on efficacy, effectiveness, and economics derived from previous experimental and observational messaging studies.

Methods

We designed and conducted this review according to a preregistered and published protocol [40] developed using the Joanna Briggs Institute Manual for Evidence Synthesis [41] and the PRISMA-ScR (Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews) [42] guidelines. The methods are described in full in the published protocol and summarized in brief in the following sections.

Review Questions

Research question (RQ) 1 was as follows: In the context of musculoskeletal pain conditions, for which individuals, with which problems, and for what purpose, has messaging on mobile devices been used (eg, medication reminders, alerts, education, motivation, prevention, and data collection)?

RQ 2 was as follows: What information exists to guide the development of mobile messaging for musculoskeletal pain conditions (eg, frequency of messages, length of messages, duration of the intervention, and theoretical basis)?

RQ 3 was as follows: How have patients’ preferences been included in the design of a study, and how have their preferences been assessed?

RQ 4 was as follows: What methods have been used to evaluate the use of mobile messaging for musculoskeletal pain conditions (eg, how were outcomes assessed and what processes were involved)?

RQ 5 was as follows: Does the literature support the efficacy, effectiveness, and economics of messaging on mobile devices for individuals with musculoskeletal pain conditions?

Inclusion Criteria

Participants

We included studies on adult participants with acute or chronic musculoskeletal pain conditions.

Concept

The concepts of interest were the development or evaluation of patient-focused health-related messaging (eg, SMS text messaging and app push notifications) provided on mobile devices such as smartphones and tablets.

Context

We included articles that described messaging used in any setting either as a primary intervention or as an adjunct to other interventions. We excluded studies focused on spinal cord injury, traumatic brain injury, moderate to severe orthopedic injuries, surgical patients, and conditions related to mobile phone overuse. We also excluded studies focused on health conditions primarily unrelated to the bones, muscles, and connective tissue (eg, diabetes, asthma, cancer, and stroke).

Data Sources

We searched PubMed, CINAHL (via EBSCOhost), Embase, and PsycINFO (via APA PsycNET) using a strategy that combined controlled-vocabulary and free-text search terms related to messaging and musculoskeletal concepts. Because of resource limitations, we were unable to include gray literature in our searches.

Search Strategy

The search strategy is described in detail in the published protocol [40], and the search queries are provided again in this paper in Multimedia Appendix 1. The search strategy was developed through discussion among the team and an iterative process of pilot searches. The final searches were conducted by SSR. Because of resource limitations, we restricted our searches to articles published in English, and because the area of digital health is a rapidly changing field, we limited our searches to articles published in the previous 10 years.

Study Selection

We exported search results to EndNote (version X9; Clarivate Analytics) and Covidence (Veritas Health Innovation) [43] for duplicate removal and to manage the screening, selection, and record-keeping processes. We conducted study selection in 3 phases. First, using the predefined inclusion and exclusion criteria, 2 independent reviewers (from a pool of 7; SSR, JL, CEE, RE, CR, SR, and NA) screened the titles and abstracts to identify candidate articles for inclusion and to discard irrelevant articles. Second, 2 reviewers from the same pool reviewed the full text of each candidate article. Third, we searched the reference lists of the included papers to identify any further articles. At all stages, conflicts were resolved using a third reviewer from our pool.

Data Extraction

Data were extracted by one reviewer (JL) and independently confirmed by 2 others (NA and CEE). Data were extracted using predefined extraction forms, as described in the protocol [40].

Synthesis and Reporting

We described the results of the study selection process using a PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) flow diagram [44], with findings reported in accordance with the PRISMA-ScR checklist [42]. For each of our 5 review questions, we structured our findings using tables adapted from the Joanna Briggs Institute manual [41] refined as necessary at synthesis stage [40]. We then developed a narrative summary of the evidence for each of our review questions.

Protocol Deviations

There were 4 minor protocol deviations. First, we excluded studies that described the use of mobile messaging to collect data in cases in which those data were not subsequently used to inform care or self-management (eg, studies that simply tested the feasibility of using text messaging to collect data and studies that used text messaging as a data collection method to model recovery trajectories). Second, we included study protocols associated with evaluation studies if they provided useful information about messaging design and development. Third, we classified the level of development of the country in which the study was conducted using the Human Development Index (HDI) [45]. Finally, we reran our searches in 2022 and, therefore, included studies from a 12-year period rather than the originally specified 10 years.

Results

Overview

Literature searches were conducted in August 2020 and repeated in May 2022. In this section, we present the combined results of both searches. We identified a total of 8328 papers (published in 2010-2022) for screening, of which 50 (0.6%) were included in this review. A PRISMA flowchart of the article selection process is shown in Figure 1.

‎

We identified 3 previous systematic reviews, 2 (67%) of which we had already found while developing the protocol for this review [22,23] and 1 (33%) that was new [46]. One review focused specifically on the effects of text messaging for managing musculoskeletal pain conditions [22], while the remainder focused more broadly on digital health or mHealth for musculoskeletal conditions but covering some aspects of messaging [23,46]. The previous reviews were conducted in Australia, the United Kingdom, and the Netherlands, all countries classed as very highly developed according to their HDI. The characteristics of the reviews are shown in Table 1, and the findings are shown in Table 2. We did not identify any previous reviews related to design aspects of messaging for musculoskeletal pain conditions.

We included 47 papers describing 40 primary studies (22/40, 55% experimental; 16/40, 40% observational; and 2/40, 5% mixed methods). In total, 10% (4/40) of the experimental and observational studies had associated or embedded qualitative or mixed methods studies. The results of 5% (2/40) of the studies were multiply reported, and 8% (3/40) of the studies had either an associated design paper or a protocol paper containing design information. A total of 18 countries were represented, with the United States publishing the largest number of studies (9/40, 23%) followed by Australia (6/40, 15%) and Denmark (4/40, 10%). By HDI, most primary studies were conducted in very highly developed countries (36/40, 90%), 8% (3/40) were conducted in highly developed countries, and 3% (1/40) were conducted in a country of medium development. No studies were reported from countries of low development.

At the time of our search, 70% (35/50) of the previous reviews and primary studies had been published in the 3 years before our search. The characteristics of the primary studies are shown in Table 3 and Figure 2 [24,26,27,30-32,35-38,47-83].

‎

RQ 1: Individuals, Problems, and Purpose

Previous Reviews

In the previous reviews [22,23,46] (Tables 1 and 2), the most commonly reported messaging interventions were for people with rheumatoid arthritis (RA) and back pain. For RA, messaging was used to monitor medication and disease activity [31] and improve medication adherence [32,46] and for reminders to reduce daily sitting time [37,46]. For people with back pain, messaging was used mostly as a component of self-management, with approaches focused on education and behavior change strategies [24,25], supportive messages provided by a health coach during periods of low engagement with a digital self-management program [27], and motivating messages sent as part of a multidisciplinary pain program [28].

Other studies described uses of messaging for people with knee pain, systemic lupus erythematosus, frozen shoulder, chronic widespread pain, and limb injuries or conditions. For knee pain, one study reported a lifestyle intervention focused on behavior change [22,38], and another reported participation reminders and app-based messaging with a personal coach as part of an exercise, education, or cognitive behavioral therapy (CBT) or weight loss or psychosocial support program [23,26]. For frozen shoulder, reminder, encouragement, and education messages were used to promote exercise compliance and improve shoulder function [30]. For chronic widespread pain, a CBT intervention used SMS text message diary completion prompts, with those diary entries then informing the treatment used by a therapist [35]. For limb injuries and conditions, messaging was used to promote adherence to a home exercise program in one study [36].

Primary Studies

Rheumatic diseases accounted for the largest proportion of the included primary studies (19/40, 48%), followed by studies on multiple musculoskeletal conditions or pain sites (10/40, 25%), back pain (9/40, 23%), neck pain (1/40, 3%), and “other” (1/40, 3%; Table 3).

Rheumatic Diseases

Of the 19 rheumatic disease–related studies, 8 (42%) focused on osteoarthritis [26,38,52,53,55,57-59,78-80], 5 (26%) focused on RA [31,32,37,48,51,56,75], 2 (11%) focused on fibromyalgia [35,47,50], 2 (11%) focused on osteoporosis [49,76], and 1 (5%) each focused on ankylosing spondylitis [54] and chronic arthritis [77].

Of these 19 studies, 14 (74%) described the use of messaging to promote behavior change with the intention of improving levels of physical activity, assisting weight loss, improving sleep, or reducing stress [26,32,35,37,38, 47,48,50,51,53-59,75,78-80]. A total of 11% (2/19) of the studies described messaging for providing information [49,52], and 5% (1/19) described the use of messaging to collect data for disease monitoring and guide clinical care [31]. In total, 26% (5/19) of the studies described aspects of design and development of messaging systems for people with knee osteoarthritis [79,80], osteoporosis [76], and chronic arthritis [77]. The design and development aspects are described in later sections.

Osteoarthritis Studies

Of the 8 studies on osteoarthritis, 2 (25%) focused on behavior change based on personalized goals. In the first study, which proposed personalized goals based on machine learning, participants were sent daily push notifications to remind them of their goals together with an interesting fact or answer to a frequently asked question [59,80]. Similarly, the second study used messaging to provide reminders to complete individualized physician-assigned goals and tasks, for which participants also used messaging to provide confirmation, or otherwise, that they had completed their personalized goals [55].

A total of 4 (50%) of studies focused on physical activity and exercise behavior change for people with knee osteoarthritis: of those, 1 (25%) used messages to decrease inactive behavior in people with knee osteoarthritis [53] and another (25%) used targeted personalized motivational reinforcement messages based on previous and current physical activity for people with osteoarthritis and sleep disturbance [57]. In the third study, which had an experimental design, the authors also explored patient attitudes and experiences of a self-directed digital health intervention incorporating automated messages to support strengthening exercises [78,79]. The fourth study, in which 77% of participants had knee osteoarthritis, described a digital care program that sent participants reminder messages if they did not engage with the program at the required intensity and also allowed participants to communicate with their health coach using messaging [26].

A single study focused on providing information for people with knee osteoarthritis, where messages were used to improve patients’ knowledge about their condition and treatment options before consultation with their specialist as part of shared decision-making [52].

A further study focused on knee osteoarthritis prevention, describing a self-management lifestyle intervention for young to middle-aged rural-dwelling women that incorporated messaging to provide key behavior reminders [38].

RA Studies

Of the 5 studies on RA, 2 (40%) used message reminders as part of a motivational counseling intervention to reduce sitting time [37,48,56], and 1 (20%) focused on physical activity behavior change with messaging used for coaching, prompts, reminders, and monitoring of physical activity program adherence [51,75]. A further study assessed the effects of text messages on medication adherence [32]. One study collected data using text or app-based messaging for symptom or disease monitoring and patient-reported outcome measures [31].

In a study that recruited women with chronic widespread pain (80% met the American College of Rheumatology criteria for fibromyalgia), text messaging was used to prompt diary completion and allow participants to exchange short messages with their therapist. The diary information was used by therapists to inform patient care [35,47]. A second guided imagery study also focused on people with fibromyalgia used text messaging to remind participants to practice their imaging exercises together with randomly selected reinforcement messages [50].

A study on patients with osteoporosis and nontraumatic fractures used text messaging to provide patients with treatment advice based on a validated fracture assessment tool and assessed whether the advice provided subsequently changed primary care physician management of their fracture [49].

Finally, one study described the use of social media messaging (WeChat) for people with ankylosing spondylitis, with messaging used for appointment reminders, for communication between physicians and patients, to record follow-up information, and for patients to provide feedback [54].

Multiple Musculoskeletal Conditions or Pain Sites

A total of 10 studies focused on multiple musculoskeletal conditions or pain sites (n=1, 10% each on the neck or back [64], neck, shoulder, or back [66], and chronic knee or low back pain [LBP] [67]). A total of 50% (5/10) of the studies recruited participants with a range of musculoskeletal problems typically seen in the general population [36,60,61,65], and 20% (2/10) of the studies recruited adults with chronic pain but not pain exclusively of musculoskeletal origin [62,63]. A further study focused on chronic musculoskeletal pain in veterans [81].

Of these 10 studies, 9 (90%) described behavior change interventions [36,60-63,65-67,81], and 1 (10%) was focused on providing information [66].

For neck and back pain, one study described the use of an artificial intelligence–enabled app that implemented evidence-based guidelines for self-management, with messaging provided within the app to remind participants to exercise and provide contact with the treating team [64]. A second study on workers with neck, shoulder, or back pain also described the use of artificial intelligence, wherein a chatbot provided messages with exercise instructions and suggestions for symptom improvement [66]. One study focused on chronic knee or LBP described a digital care program incorporating sensors and an app that allowed participants to communicate with a personal coach via SMS text messaging and app-based messaging [67].

Another 20% (2/10) of the studies included adults with chronic pain but not exclusively pain of musculoskeletal origin [62,63]. The first included patients being treated by a hospital-based pain management service for a range of conditions (LBP; cervical or upper-extremity, lower-extremity, abdominal or pelvic, and head or face pain; and multiple pain sites, with pain of ≥4 on a 0-10 scale). Participants used an app that incorporated reminders to complete daily assessments and also provided 2-way messaging [63]. The second study, with similar wide-ranging pain sites, used automated text messaging to prompt skill practice as part of an internet-delivered CBT program for chronic pain [62].

Regarding patients attending hospital physiotherapy services for a range of musculoskeletal problems, 10% (1/10) of the studies examined whether SMS text messaging could increase home exercise compliance [36]. In this study, compliance with exercises was encouraged via motivational SMS text messages sent by the physiotherapist. Similarly, the use of messaging to encourage home exercise compliance was described in a study on patients with musculoskeletal problems attending a chiropractic clinic [60].

In the physiotherapy outpatient setting, the use of SMS text message reminders to reduce clinic nonattendance was described in 10% (1/10) of the studies [61].

A total of 20% (2/10) of the studies focused on specific populations. The first, a community-based study, aimed to improve the physical activity of older adults (aged ≥60 years, most of whom had musculoskeletal problems) and used social media messaging (WhatsApp) to inform participants of scheduled walks and promote social interaction between participants [65]. The second study focused on a chronic musculoskeletal pain program in veterans and used behavior change messaging for stress management and adoption of healthy sleep practices and to increase engagement and retention in the program [81].

Back Pain

A total of 20% (8/40) of the studies described behavior change interventions [24,27,68-73,82], and 5% (2/40) described the design and development (described in a later section) [82,83]. Of the 8 behavior change studies, of these 4 (50%) described the use of individual or personalized messaging for physical activity goal reminders and reinforcement [24], encouragement messages and physical activity suggestions [69], motivational notifications for self-management [73,82], and individual activity level–based feedback messages provided on a PDA to encourage behavior change [68]. A total of 13% (1/8) of the studies described a self-management app with notifications to encourage walk breaks and posture exercises [72].

A total of 38% (3/8) of the studies described the use of 1- or 2-way messaging with a health coach, physiotherapist, or sports scientist for support, encouragement, and participation reminders as part of self-management programs [27,70,71].

Neck Pain

Only 3% (1/40) of the studies focused specifically on neck pain. This study described a behavior change intervention for office workers with chronic neck pain incorporating weekly messages about caring for their pain with information about the importance of exercise and to provide encouragement to complete prescribed exercises [74].

Other Conditions

A total of 3% (1/40) of the studies, on patients with frozen shoulder recruited from an orthopedic outpatient clinic, used messaging to provide reminders, encouragement, and education to promote shoulder exercise compliance [30].

RQs 2 and 3: Design and Development and Patient Preferences

Overview

In this section, we report findings related to the design and development of messaging interventions. Because patient preferences, where accommodated, were generally addressed through participatory or co-design, we have reported the results of review questions 2 and 3 together. The findings are presented in three groups: (1) information found in papers specifically focused on the design and development of messaging interventions, (2) information found in mHealth design papers where some aspect of messaging was described alongside other mHealth functions, and (3) incidental design and development information found in papers that reported the results of messaging or mHealth interventions. The design-specific papers are shown in Table 4 and Figure 3 [76-83].

‎

Papers Focused Specifically on Messaging Design and Development

A total of 4% (2/47) of the papers comprehensively described the design and development of SMS text messaging interventions for knee osteoarthritis [79] and back pain [83].

An SMS Text Messaging Intervention to Support Home Exercise Adherence for People With Knee Osteoarthritis

In 2019, Nelligan et al [79] comprehensively described a formal two-phase process to (1) identify behavior change targets and (2) design a library of SMS text messages to support adherence to home exercises for people with knee osteoarthritis. The development was guided by the recommended steps for developing text messaging–based programs for health behavior change published by Abroms et al [100] in 2015.

The first phase of development, comprising 3 stages, focused on target behavior, barriers, facilitators, and behavior change techniques using the Behavior Change Wheel framework [97,98]. Stage 1 drew on the literature to define the problem in behavioral terms, explaining the behavioral target and context and the barriers and facilitators for people with knee osteoarthritis in terms of participating in exercise mapped to domains in the Theoretical Domains Framework [96]. Barriers and facilitators relevant to the target behaviors were organized using the Capability, Opportunity, and Motivation–Behavior model for behavior change [97]. Stage 2 mapped barriers and facilitators to select intervention functions and behavior change techniques appropriate for implementation using SMS text messaging [93]. In stage 3, behavior change techniques for each function were identified from the Behavior Change Technique Taxonomy (version 1) [99].

The second phase involved the development of SMS text messaging functionality, specifically, a message library of content and determination of message frequency and level of interaction. Messaging content was derived by taking each barrier- or facilitator-linked behavior change technique identified in the first phase and constructing a relevant SMS text message. Message content was derived with input from 12 participants (1 person with knee osteoarthritis, 7 researchers, and 4 physiotherapists). In total, 3 authors derived the final message bank. A fourth author reviewed the final SMS text message wording to ensure that it was consistent with the Behavior Change Wheel mapping process and the identified behavior change techniques. The final message bank was organized into a 24-week schedule, assessed using literacy tools for readability, and tested by the authors for functionality and errors.

Author-provided examples of the mapping process and resulting SMS text message content for example barriers and facilitators are shown in Table 5.

A Messaging Self-Management Intervention for LBP

In 2019, Fritsch et al [83] described the co-design process used to derive a bank of evidence-based lifestyle-focused messages for an LBP self-management text messaging intervention.

The authors used an iterative 2-phase co-design approach based on a framework used to design prevention messages for patients with cardiovascular disease previously published by Redfern et al [106] in 2014.

Phase 1 consisted of two 2-hour workshops intended to develop the concept, initial content, and messages. Workshop participants were researchers, clinicians with specific knowledge related to LBP, and consumer representatives from the support group Musculoskeletal Australia. At the first workshop, participants identified important domains relevant to LBP (exercise, education, mood, use of care, sleep, and medication) through reference to an evidence-based consumer resource (Managing your pain: An A-Z guide; Musculoskeletal Australia). The second workshop was focused on identifying sources of content for messages and duration, frequency, and timing of messages. Identified sources of content were relevant peer-reviewed literature, Australian and international clinical practice guidelines for LBP, and consumer group patient educational resources. Message frequency (4 messages per week) and timing (9 AM, 12:30 PM, 4 PM, and 6 PM) were based on previous work in coronary heart disease [107]. The development team considered that an intervention program duration of 12 weeks would be appropriate, with exercise domain messages being sent twice per week (emphasizing the importance of remaining active) and 1 message sent per week for each of the other domains.

This phase of the development process was also informed by previous work on factors related to engagement, perceived usefulness, behavior change, and delivery preferences for patients with coronary heart disease [105].

Following the workshops, a team comprising 2 researchers and 2 consumer representatives drafted evidence-based behavior change messages following the same theoretical approach by Redfern et al [106]. Messages were focused on education, motivation, or behavior change in the domains of providing information or encouragement; prompting about consequences, intention formation, monitoring self-behavior, and barrier identification; advice about setting graded tasks; and strategies aimed at relapse prevention and the use of prompting and cues. The team drafted an initial set of 82 positively phrased messages (by domain: 40 exercise messages, 10 education messages, 10 mood messages, 8 use of care messages, 7 sleep messages, and 7 medication messages) to take forward to the second phase of development.

In the second phase, the authors used a web-based survey of leaders in the field of LBP management to assess the appropriateness of the message content, gather opinions on the likelihood that the messages would be clinically effective, and make recommendations for message content improvement. The mean score for the messages from the expert review was 8.30/10. Messages with a score of <8/10 (34%) were modified in response to accompanying feedback. Subsequently, consumers scored each text message on utility of content, understanding, and language acceptability. Text messages with a consumer review score of <12/15 (31%) were revised according to feedback (mean score 12.5/15 points). Most frequently, consumer feedback focused on making the content more specific and less technical and including more examples.

Papers Describing the Design and Development of Messaging Within mHealth Apps

A total of 9% (4/47) of the papers described the design and development of more general mHealth interventions, where those interventions contained some use of messaging (alongside other mHealth features) for people with knee or hip osteoarthritis [80] and back pain [82], pain self-management for veterans [81], and women newly diagnosed with osteoporosis [76]. In total, 2% (1/47) of the papers focused on feature preferences for an app to support the self-management of chronic arthritis [77].

In each case, the design of the overall intervention was typically well described; however, the design of the content, timing, and frequency of the messaging components was not described in detail (Table 4). Because these papers provided little useful messaging-specific design and development information, we do not discuss them any further.

Incidental Design and Development Information Contained in Papers Reporting the Results of mHealth Interventions

We found little useful design-related information contained within the papers describing results of interventions. Typically, the papers described the purpose and behavior of the messaging component within their intervention, but the design processes used to determine message content, timing, and frequency were described incidentally or not at all (studies shown in Tables 6 and 7) [30,35,36,38,47,50,60,62,63,65,66,72,74]. For example, one paper provided examples of messages intended to provide encouragement, education, or motivation but provided no explanation of how these were derived [30]. Similarly, some papers (4/47, 9%) made a passing reference to co-design processes involving patients and clinicians but provided limited detail [37,48,55,56].

Some papers (17/47, 36%) described the use of messaging adaptivity (ie, dynamic system-initiated changes to the delivery of messaging to personalize content, frequency, or timing of messages based on automated or manual triggers) or individualization. Triggers for adaptivity included self-reported exercise adherence [79], automated physical activity data derived from wearables [57,68,73], self-reported data [54,66,71], personalized goals [37,48,55,56], and manual adaptivity triggers initiated by study personnel [31,63] and health coaches [26,27,67,70]. However, in these papers, no substantial detail was provided on the design considerations or processes related to the development of the intervention’s adaptive behavior.

RQs 4 and 5: Evaluation Methods, Efficacy, Effectiveness, and Economics

To avoid repetition, the findings of review questions 4 and 5 are reported together. A total of 28% (11/40) of the studies directly compared the use of messaging with an alternative; a further 60% (24/40) of the studies evaluated mHealth interventions with embedded use of messaging.

Studies Comparing the Use of Messaging With an Alternative

Of the 11 studies that directly compared messaging to an alternative, 9 (82%) had an experimental design and 2 (18%) were observational. In most cases, the comparator or control condition was no messaging or treatment as usual, with outcome measures varying by the intent of the intervention. Of these 11 studies, 3 (27%) [30-32] were described in the previous review on the effectiveness of text messaging interventions on the management of musculoskeletal pain [22], and the remainder were not, likely because they did not meet the inclusion criteria or were published later [49,52,53,60-63,66].

Overall, the outcomes either favored the messaging condition or were equivocal.

Examples of outcomes favoring messaging interventions included improved knowledge of the illness and the available treatment options and physical activity for knee osteoarthritis [52,53], improved medication adherence and physical functioning for RA [31,32], improved attendance to outpatient physiotherapy [61] and engagement with general practitioner [49], and improved exercise compliance for frozen shoulder [30] and mixed musculoskeletal conditions [60]. However, despite participants sometimes reporting messaging as acceptable [62] or appealing [63], and while improved pain intensity was found in participants with neck and shoulder pain and LBP [66], some studies (4/40, 10%) reported equivocal findings for important patient outcomes such as time spent physically active [53], pain [53,63], and quality of life [31,53].

In no studies did the primary outcome favor the control condition. In only one study, a secondary outcome (clinician responsiveness) favored the control condition. In this study, patients with chronic pain recorded their progress using an app, with intervention recipients also having access to messaging with their clinician (controls could report progress but had no messaging). Control participants perceived their clinicians to be more responsive to their progress reports [63].

No studies reported economic outcomes.

The studies are summarized in Table 6.

mHealth Studies With Embedded Messaging Components

A total of 24 studies evaluated mHealth interventions containing some form of embedded messaging component (n=11, 46% experimental; n=11, 46% observational; n=1, 4% observational with a qualitative component; and n=1, 4% qualitative embedded within an experimental study).

The results of efficacy and effectiveness were mixed, but because messaging was embedded within a larger mHealth intervention, it was not possible to isolate the messaging-specific effects from the overall intervention effects.

A total of 8% (2/24) of the studies reported economic outcomes—avoided surgery costs associated with a digital education program for chronic knee pain, in which messaging was used for coaching or peer support and program engagement reminders [26], and reduced travel time associated with a self-management mHealth tool for ankylosing spondylitis, in which social media messaging was used for communication between physicians and patients [54].

While it was not possible to isolate messaging-specific effects, these studies are included for completeness and summarized in Table 7.

Discussion

Principal Findings

To our knowledge, this is the first study to comprehensively map how mobile messaging has been used in the treatment and self-management of musculoskeletal conditions. We mapped the conditions and purposes for which messaging has been used and the approaches used to design and develop messaging interventions and summarized the evidence of efficacy, effectiveness, and economics from both experimental and observational studies. Our intent was to draw together all the available relevant information to help inform the future design of messaging interventions for people with musculoskeletal conditions and identify research gaps.

While previous reviews in this area are few, this work builds on 3 existing syntheses of the effectiveness of messaging interventions for people with musculoskeletal conditions. One review focused specifically on text messaging interventions for musculoskeletal pain [22]. The review included studies across a range of musculoskeletal problems and included both studies in which messaging was added to and compared with usual care (findings of positive effects only on exercise and medication adherence) and studies in which messaging was a component of a larger intervention (reporting some small effects on pain intensity, function, care-seeking behavior, exercise and medication adherence, and quality of life). Overall, the quality of the evidence was low. The 2 other reviews focused more generally on digital health for managing musculoskeletal conditions [23] and mHealth interventions for people with RA [46].

In this review, all the included studies that assessed intervention efficacy or effectiveness (on pain [30,53,63,66], function [30,31,63], disability [53], adherence to the intervention [30,60,63], physical activity levels [53], appointment attendance [49,61], health care contact [31], mood [63], quality of life [31,53], remission [31], and disease activity [31,32]) reported either equivocal findings or findings favoring messaging.

The notable absence of studies reporting negative outcomes may suggest publication bias. The lack of economic studies is also concerning; no messaging-specific studies reported economic outcomes. While, of the 40 studies, 2 (5%) digital or mHealth studies with messaging components did report economic outcomes, including avoided surgery costs and reduced travel time [26,54], the embedded nature of messaging means that it is not possible to attribute the observed savings specifically to the messaging components. While a previous review has shown messaging to be cost-effective in some circumstances, there is no information on the economic effects on musculoskeletal pain conditions; in cases in which messaging interventions are shown to be effective, further studies should be conducted to assess their economic effects [22,108].

We identified studies describing the use of messaging across a range of musculoskeletal conditions, with rheumatic diseases representing almost half (19/40, 48%) of the included studies, of which two-thirds (13/19, 68%) focused on osteoarthritis and RA. Somewhat surprisingly given its high population prevalence, back pain was represented by less than a quarter of primary studies (9/40, 23%). A further quarter of the studies (10/40, 25%) addressed multiple musculoskeletal conditions, but most (30/40, 75%) targeted single musculoskeletal conditions and pain sites despite evidence that musculoskeletal conditions often do not occur in isolation (eg, in Australia, 64% of people with back pain and 74% of people with arthritis have at least one other chronic condition) [3].

More than 80% of the included primary studies (34/40, 85%) focused on the use of prompts and reminder messages to foster positive behavior change at the individual level, most commonly in combination to encourage movement (eg, to increase physical activity, reduce sitting time, and improve compliance with prescribed exercise); compliance with prescribed medication; the practice of coping skills; and the meeting of personal goals. While a small number of studies (10/40, 25%) described the use of unidirectional or 2-way messaging with a health coach or exercise or sports scientist for support and encouragement, more studies (11/40, 28%) described the use of automated and unidirectional messaging, which, while economical on resources, may limit effectiveness in fostering behavior change.

While most studies (34/40, 85%) focused on influencing individual behavior change, there appears to be limited research into the use of messaging to improve treatment or self-management at the broader system level (eg, to improve health care processes, handover communication, and continuity of care between providers). One study on RA used SMS text messaging–based monitoring of medication adherence and disease progression to inform follow-up nurse contact but found no difference in the primary outcome of remission [31]. A second study on a digital health platform for ankylosing spondylitis management consisting of a patient and physician portal and 2-way chat via social media reported improvements in the proportion of patients with inactive disease and an avoidance of 29.1% of in-person clinic visits [54]. Future studies could focus on addressing the gaps in knowledge on more process- or system-focused interventions.

Overall, we found limited information about messaging design. One study (1/40, 3%) used development processes previously described by Redfern et al [106], which, while originally focused on cardiovascular event prevention messaging, have since been more widely adopted and adapted in the co-design of text messaging interventions, including for diabetes prevention [109], endometriosis support [110], and support after breast cancer treatment [111]. Similarly, one study (1/40, 3%) used guidance for the development and testing of messaging in health behavior change developed by Abroms et al [100].

Only 4% (2/47) of the papers provided comprehensive information about message design and development [79,83] and highlighted the importance of taking a formal approach and having a theoretical underpinning and meaningful consumer involvement. Beyond these 2 papers, design was typically poorly described or not described at all, and many projects appeared to leapfrog from concept to implementation with a limited or absent design phase, perhaps not recognizing the importance of formal design for subsequent adoption.

While some papers (8/47, 17%) did describe elements of participatory design or co-design, some papers (4/47, 9%) provided limited detail or had limited consumer involvement, or in some cases, consumers were involved after the design had already been conceived by researchers and clinicians. In most cases, we found that papers described the theoretical basis underlying their intended behavior change well, but consistent with a previous review [22], we found few detailed descriptions of the process through which the content, timing, and frequency of the messages were derived. This remains an important weakness in the musculoskeletal literature specifically and has been identified more generally by others [100,106]. Further work should be conducted to elicit preferences regarding these processes from people across the spectrum of musculoskeletal conditions.

The strengths of this review include our comprehensive search strategy and the inclusion of a wide range of studies and designs, providing a rich map of the literature expanding the insights provided by previous effectiveness-focused reviews. A limitation is that we focused specifically on the use of messaging in patient care and self-management, and it is possible that there are other messaging applications relevant to people with musculoskeletal conditions. However, we made this review as broad as possible within available resources. Because a comprehensive synthesis was time-consuming and we last conducted the searches in 2022, there may be important, more recent studies that we missed in this review. Similarly, resources did not allow us to review the gray literature.

Conclusions

In conclusion, messaging has been used for the care and self-management of a range of musculoskeletal conditions with generally favorable outcomes reported. Nonetheless, there are areas that should be addressed by future research to improve the quality of intervention design, which will hopefully translate into uptake and sustainability. First, preferences related to messaging content, timing, and frequency should be further explored specifically among people with musculoskeletal conditions, eliminating the reliance on information from other disciplines. Second, teams should incorporate digital intervention design expertise, follow formal design processes, and clearly describe design considerations and processes used. Finally, in cases in which messaging interventions are shown to be effective, further studies should be conducted to assess their economic effects and practical considerations related to implementation and sustainability.