Introduction

Physical activity, defined as any voluntary bodily movement that is produced by skeletal muscles and requires energy expenditure [1], is an essential component of successful disease self-management in people with arthritis. Extant research shows that it can improve pain, function, and quality of life [2, 3], and is safe for people with osteoarthritis (OA) and inflammatory arthritis (IA) [4]. Six million people currently live with arthritis in Canada, and this number is projected to reach 10.4 million by 2042 [5]. Among older adults, arthritis often coexists with frailty, which poses a greater risk of falls [6, 7], increased morbidity, physical dependence, and hospitalization [8, 9]. Physical activity can prevent the progression of frailty [10, 11], partly by mitigating the effects of arthritis [2, 3] and the progression of concomitant conditions such as cardiovascular disease [12] and type 2 diabetes [13].

Physical activity is beneficial during all phases of chronic disease management, from primary prevention through rehabilitation and ongoing care. However, only 49% of adults aged 18–79 in Canada engage in enough physical activity to meet the World Health Organization recommendations [14]. Among people with arthritis, physical activity participation is even lower than the general population [15,16,17]. Introducing physical activity to this population is challenging as their ability to engage in activities can be affected by their symptoms, their health status (including comorbidities), or other external factors.

Both clinicians and academics have identified the lack of precision in physical activity prescription and promotion as a limitation to supporting health behavior change. In a cohort of 172 people with OA and IA who were eligible for physical activity intervention studies [18, 19], Feehan et al. [20•] identified four distinct activity profiles based on the average time participants spent daily sleeping, sitting, and walking in different intensities. Their findings suggest that a range of individualized strategies may be required to help people achieve their optimal balance between activity and rest [21•].

Physical Activity Tailoring

If we want to improve physical activity participation among people with arthritis, we need to move away from a one-size-fits-all approach and toward a personalized approach. We define “physical activity tailoring” as a process whereby health professionals use patient assessments to shape a personalized strategy to support an active lifestyle [22•], taking into account the patient’s characteristics, preferences, needs, and context [23]. When a health professional is tailoring physical activity interventions, they assess individual factors (e.g., physical activity history, fitness, psychosocial factors) and health status (e.g., disease activity, symptoms) by inquiry, observation, physical examination, and/or monitoring. They then use this information to design the “what” (e.g., activity type, frequency, and intensity) and “how” to deliver the prescription (e.g., a combination of behavior change techniques (BCT), who delivers, using what mode of delivery) [24] to support individuals to engage in those activities. The purpose of this review is to first summarize the research on the use of physical activity tailoring in people with OA, IA, and related conditions (e.g., fibromyalgia). Secondly, we report the effectiveness of different tailoring strategies on physical activity behavior in this population.

Methods

Search Strategy and Literature Screening

The systematic review protocol has been registered at PROSPERO (https://www.crd.york.ac.uk/prospero/; Registration CRD42020215513). We followed the PRISMA guidelines and searched the PROSPERO database for similar reviews before commencing this review. This review was co-developed with people living with arthritis and health professionals (KT, AMH). We also consulted with the Arthritis Patient Advisory Board, a group of advocates who bring their personal experiences and expertise with arthritis to research decision-making. Their involvement included shaping the research question, refining the data extraction, reviewing the findings, and identifying priority topics for discussion.

The search strategy was developed with an experienced research librarian. The search was conducted in June 2020 and updated in May 2022. We searched five electronic bibliographic databases including PubMed, PsycINFO, CINAHL, Embase (Ovid), and MEDLINE (Ovid). Our search included terms for tailoring, physical activity, and arthritis. See Online Resource: Fig. 1 for a sample search strategy. In addition, we hand-searched reference lists of included articles and conducted forward citation searching.

Eligible articles (1) were original studies, (2) included people with arthritis and related conditions, (3) included a tailored intervention to promote physical activity participation, and (4) included a measure of physical activity behavior. Articles were excluded when either (1) the intervention did not include an assessment or a form of assessment for tailoring, and (2) the article was written in a language other than English.

We transferred all the retrieved articles from each bibliographic database to Covidence (www.covidence.org; Veritas Health Innovation, Melbourne, Australia) for screening, and removed all duplicate articles. Two reviewers (SR, AS) screened all articles by their titles and abstracts and then reviewed the full articles for those that appeared to be eligible. A third reviewer (JM) mediated any discrepancies. JM and DH completed the updated search following the same procedure.

Data Extraction and Analysis

We developed a data extraction form in consultation with Arthritis Patient Advisory Board members. Two reviewers (SR and AS) were trained by an experienced researcher (JM) and pilot-tested the data extraction with 10% of the papers for calibration. One reviewer (SR) extracted data from the remaining articles and a second reviewer (AS) checked them. The following information was extracted: study details, intervention details, change in physical activity behavior, and tailoring factors (timing, frequency, and content of assessments, who or what conducted the assessment and delivered intervention components, which assessments were explicitly paired with tailored intervention components).

We summarized the study and intervention characteristics in descriptive statistics. For each included study, we mapped the assessment and tailoring strategies used (Fig. 1). One experienced reviewer (JM) coded the tailored intervention strategies for BCTs using the 93 Behaviour Change Technique Taxonomy version 1 (BCTTv1) [24]. The BCTTv1 has previously demonstrated support for good inter-coder and test–retest reliability [25].

Fig. 1
figure 1

Tailoring checklist: summary of tailoring methods and considerations used across studies

Full size image

Appraisal of the Evidence

We assessed the quality of the studies with the Cochrane Risk of Bias (ROB) tool (for randomized controlled trials, RCTs) [26], and the Risk of Bias in Non-randomized Studies of Interventions (ROBINS-I; for non-RCTs) [27]. When interpreting overall risk of bias for randomized controlled trials, we did not consider blinding of participants given the challenge of blinding in a physical activity intervention. Random sequence generation, blinding of outcome assessment, incomplete outcome data, and other bias domains were prioritized in the overall Cochrane ROB appraisal.

Results

The systematic search retrieved 2121 records; of those, 257 passed the title and abstract screening. After the full-text screening, 78 articles (from 39 studies) were included (Online Resource File 1: List of Included Studies), with a total of 13,626 participants (individual study sample size varied from 3 to 8894 participants) (Online Resource Fig. 2: PRISMA Flow Diagram). The primary reasons for exclusion were (1) not an original article (n = 99), (2) did not measure physical activity behavior (n = 30), and (3) did not conduct a tailored intervention (n = 13). For the complete data extraction, see Open Science Framework Repository: https://osf.io/t7ybv/.

Study Characteristics

Of the 39 total studies identified (Table 1), 31 were randomized controlled trials [18, 19, 28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56]. Twelve studies were categorized as low risk of bias [18, 39,40,41, 44, 49, 51, 52, 54, 55, 57, 58], 13 as moderate risk of bias [28,29,30, 32, 35, 37, 38, 42, 50, 53, 56, 59, 60], and 14 as high risk of bias [19, 31, 33, 34, 36, 43, 45,46,47,48, 61,62,63,64] (Online Resource: Table 1 and 2). The mean age of participants varied from 48 to 72 years old. Twenty-three studies reported on physical activity tailoring in participants with OA [19, 28, 29, 31, 34, 35, 37, 42,43,44,45, 47, 48, 50,51,52,53, 56,57,58, 62,63,64], 13 in participants with IA [18, 30, 32, 33, 38, 40, 41, 46, 49, 54, 55, 59, 60], and 3 in participants with fibromyalgia [36, 39, 61] (Table 1). The disease duration of study participants ranged from 1.75 to 20 years.

Table 1 Study characteristics
Full size table

Intervention Details

The interventions employed in the studies lasted from 19 days to 2 years. Twenty-eight interventions were delivered in-person, while the rest (n = 11) were delivered remotely (e.g., via app, telephone, website) (Table 2). Most interventions (n = 31) were delivered by a health professional (physiotherapist, occupational therapist, nurse, physician, exercise physiologist). Providers who delivered interventions were variably trained in counseling, motivational interviewing, autonomy-supportive strategies, behavior change techniques, exercise prescription, stages of change, patient-practitioner collaboration, effective communication, and brief action planning.

Table 2 Intervention characteristics
Full size table

Changes in Physical Activity Participation

Of the 39 studies, 27 demonstrated significant improvements in at least one measure of physical activity over time [18, 19, 28,29,30,31,32, 35, 36, 38, 40, 42,43,44,45,46,47,48,49,50,51,52, 55, 56, 58, 59, 63]. Sixteen interventions demonstrated significant improvements in physical activity in comparison to control groups [19, 28,29,30, 35, 40, 42, 44,45,46,47,48,49,50,51,52], though 4 of those interventions were appraised as having a high risk of bias.

Tailoring Methods

Figure 1 summarizes the tailoring methods and considerations used across the studies and provides a checklist to guide decision-making. The majority of studies conducted assessments throughout the intervention (n = 29), at baseline only (n = 2), or used both baseline and repeated assessments throughout the intervention (n = 8). Assessment was conducted by a human (n = 30), technology/device (n = 5), both (n = 3), or was not reported (n = 1). Both the interventionist and participant (n = 26), only the interventionist (n = 9), technology/device and participant (n = 3), or all of the above (n = 1) decided which tailored intervention components were delivered based on the assessment.

Interventionists and patients shared decision-making responsibility for the tailoring of the intervention in 10/12 interventions that demonstrated significant differences between groups in favor of the tailored intervention in at least one outcome of physical activity behavior (excluding high risk of bias studies). In all the effective interventions, assessments to inform tailored approaches were reassessed at multiple time points. The remainder of tailoring methods did not demonstrate consistent patterns across effective, low-moderate risk studies. See Online Resource: Table 3 for the full tailoring characteristic data extraction.

Assessment Factors and Intervention Strategies used in Tailoring

Figure 1 contains a summary of assessment and intervention factors used for tailoring across the included studies (labeled as “tailoring options”). Online Resource: Figs. 3 and 4 contain the assessment and intervention factors employed by each study. A total of 24 unique assessment factors and 23 intervention strategies for tailoring were used across the studies. The most commonly used assessment factors were past physical activity (n = 22), disease symptoms (n = 16), physical function (n = 14), fitness (n = 14), goals (n = 14), barriers (n = 13), and confidence (n = 11). The most commonly used tailored intervention strategies were goal setting (n = 27), physical activity prescription (n = 24), and problem solving (n = 17).

Among the low-moderate risk of bias interventions that demonstrated significant differences between groups’ physical activity behavior in favor of the tailored intervention, the assessment factors were barriers, confidence, demographics, disease symptoms, fitness, goals, mood, motivation, needs, past physical activity, physical function, and preferences (activity type, setting, general/not defined). The tailored intervention strategies included behavior change techniques (action planning, body changes, goal setting, environmental context and resources, information about health consequences, instructions on how to perform the behavior, problem solving, reinforcement), intervention dose, provider, and physical activity equipment, options, and prescription.

Discussion

This review provides a summary of tailoring methods used to date in physical activity interventions for people with arthritis. The evidence suggests that assessments to guide tailoring should be made repeatedly over time and that the health professional and the patient should make decisions together about tailoring. Health professionals may be able to select more appropriately tailored intervention strategies if they assess factors such as patient characteristics (e.g., demographics, disease symptoms, physical function), barriers (e.g., needs, confidence, motivation), and physical activity readiness and preferences (e.g., physical activity history, fitness, goals). Such strategies may include tailoring behavior change techniques (e.g., action planning, problem solving, provision of resources), the frequency of using these techniques, exercise prescriptions, and who should deliver the intervention.

Are Tailored Physical Activity Interventions Effective for People with Arthritis?

Overall, tailored interventions appear to be effective in improving physical activity participation among people with arthritis. However, we were unable to determine whether they are more effective than generic interventions given the lack of studies comparing the use of these approaches in this population. Meta-analyses have shown that tailored interventions demonstrate only small effects [65] or have demonstrated smaller effects than generic interventions for changing physical activity behavior in the general population [66]. However, select reviews and randomized controlled trials have supported the use of tailored interventions in improving physical activity behavior and demonstrated medium- to large-sized effects [67]. Well-defined moderators of physical activity behavior also signal the value of tailored interventions [68, 69]. Taken together, we argue that comparing the effectiveness of tailored vs. non-tailored interventions is premature without evidence for optimizing tailoring methods [22•]. Our summary of tailoring methods used across studies and recommendations for effective strategies is a step toward improving our understanding of optimal tailoring approaches. Further research is needed to define how tailoring approaches can be refined before quantifying or comparing its value in physical activity interventions. Therefore, while it is unclear whether tailored physical activity interventions are more effective than generic interventions among people with arthritis, our review notes the common use of certain tailored approaches in studies with effective interventions, which may indicate the importance of taking into account a patient’s characteristics, physical abilities, and psychosocial needs in client-centered care.

How do Health Professionals Learn to Tailor?

There is likely room for both systematic and humanistic approaches to tailoring. The Behaviour Change Wheel is one example of a systematic approach using evidence- and consensus-based links to tailor behavior change techniques to identified barriers [70]. The questions and checklist provided in this review may also help health professionals and researchers strategically plan tailoring in a systematic fashion. A key finding of this review is that a tailored physical activity intervention strategy has more chance of success if the health professional and patient collaborate on how to structure it. None of the included studies reported details of the decision-making process, possibly because the collaboration with patients was more intuitive than systematic (i.e., following a distinct procedure). Many health professionals are already tailoring strategies instinctively by employing aspects of motivational interviewing (compassion, acceptance, partnership, evocation) and principles of shared decision-making (e.g., situation diagnosis, option clarification, deliberation of patient preferences) [71, 72•]. Another point of consideration is that none of the included interventions mentioned self-tailoring, where patients alone decide on the intervention strategies and how they are tailored. Self-tailoring stems from the concept of self-management, whereby patients use self-management skills, decision-making skills, and problem-solving skills to apply knowledge to themselves as appropriate [73]. Collaborating with and equipping patients to self-tailor may be another tailoring strategy that benefits both health professionals’ workload and patient outcomes. As the literature for optimal physical activity tailoring strategies among patients with arthritis builds, it is important to consider not only “what” we tailor, but “how” and “with whom” we tailor.

Limitations and Future Directions

We acknowledge that this review has certain limitations. First, most studies did not provide enough detail for their tailoring methods to be reproducible. Therefore, the coding we conducted produced broad categories for tailoring, but important nuances like specific questions to ask or specific assessment tools to use are missing. Second, the checklist in Fig. 1 is a compilation of tailoring options that were included across the studies in this review and is not an exhaustive list. Furthermore, it is challenging to single out effective components of complex interventions as they likely intertwine and overlap. We encourage researchers and health professionals to build on our list and use principles of evidence-informed practice to decide with their patients which tailoring options are best suited for them [74]. Lastly, given the heterogeneity of the included tailoring approaches and limitations in study reporting, we were unable to define explicit links between assessments and strategies (i.e., if X finding is determined from the assessment, then Y strategy should be implemented). In the future, the specific links between assessment factors and tailored intervention strategies should be explored and more clearly defined.

Conclusion

Physical activity is a complex behavior that is influenced by demographic, physical, and psychosocial factors. Health professionals and researchers who want to move beyond generic interventions and increase physical activity participation among people with arthritis may benefit from using tailored approaches that involve patients in the decision-making process, can adapt to each patient’s changing needs over time, and are based on a broad range of relevant assessment factors. While much more work is needed to develop and refine methods for optimal tailoring, health professionals can start by familiarizing themselves with their options for tailoring and take a patient-centered approach to guide their decision-making and help more people with arthritis to be physically active.