Understanding prescribed dose in hand strengthening exercise for rheumatoid arthritis: A secondary analysis of the SARAH trial

Abstract Objective 1) To identify therapist or participant characteristics associated with prescribed dose of hand strengthening exercise in adults with rheumatoid arthritis and 2) To determine the impact of dose prescribed on outcome (hand function and grip strength). Methods Overall dose was calculated using area under the curve (AUC). Analysis 1 assessed the association between therapist professional background, therapist grade, baseline participant physical and psychological characteristics and prescribed dose. Analyses 2 and 3 estimated the relationship between prescribed dose and overall hand function and grip strength. Generalised estimating equation linear regression analysis was used. Results Analysis 1: Being treated by an occupational therapist (β = −297.0, 95% CI −398.6, −195.4), metacarpophalangeal joint deformity (β = −24.1, 95% CI −42.3, −5.9), a higher number of swollen wrist/hand joints (β = −11.4, 95% CI −21.6, −1.2) and the participant feeling downhearted and low all of the time (β = −293.6, 95% CI −436.1, −151.1) were associated with being prescribed a lower dose. Being treated by a grade 6 therapist (β = 159.1, 95% CI 65.7, 252.5), higher baseline grip strength (β = 0.15, 95% CI 0.02, 0.28) and greater participant confidence to exercise without fear of making symptoms worse (β = 18.9, 95% CI 1.5, 36.3) were associated with being prescribed a higher dose. Analyses 2 and 3: Higher dose was associated with greater overall hand function (β = 0.005, 95% CI 0.001, 0.010) and full‐hand grip strength (β = 0.014, 95% CI 0.000, 0.025) at 4‐month. Conclusion Higher dose was associated with better clinical outcomes. Prescription of hand strengthening exercise is associated with both therapist and participant characteristics.


| INTRODUCTION
Hand exercise is recommended by the National Institute for Health and Care Excellence (NICE) for individuals with pain and dysfunction of the hands and wrists caused by rheumatoid arthritis (NICE, 2018). This recommendation is underpinned by data from the strengthening and stretching for rheumatoid arthritis of the hand (SARAH) trial, which found that providing a tailored hand exercise programme in addition to usual care, is a clinically and cost-effective adjunct to the various drug regimens presently recommended .
Post-hoc analysis of the SARAH programme identified the techniques used to target grip strength (i.e. hand strengthening exercise) partially mediated improvements in overall hand function (Hall et al., 2017). Whilst NICE support using strengthening exercise for the rheumatoid hand, no information is provided about what dose may work best.
Developing a better understanding of the optimal dose for rehabilitation interventions has recently been identified as a research priority (CSP, 2018a). Further refinement of the dose of hand strengthening exercise may optimise clinical and costeffectiveness and inform future (more detailed) clinical guidelines.
Such refinement should be informed by data and evidence as far as possible. We used data from the SARAH trial to address the following objectives:

| OBJECTIVES
To identify the therapist and participant characteristics at baseline associated with the overall dose of hand strengthening exercise prescribed to participants in the SARAH trial (Analysis 1).
To identify the association between overall dose of hand strengthening exercise prescribed during the programme and hand function and grip strength (Analyses 2 and 3).

| Design
This study is a post-hoc exploratory analysis of the data from the SARAH trial. The SARAH trial was a pragmatic parallel-group trial conducted at 17 National Health Service sites across the UK. The complete methods of the SARAH trial are described in full elsewhere .

| SARAH hand exercise programme
The programme comprised six sessions of face-to-face contact (one assessment and five supervised exercise sessions) with an occupational therapist or physiotherapist. Seven mobility and four strengthening exercises were used. The four strengthening exercises (eccentric wrist extension, gross grip, pinch grip, finger adduction) used load (resistance) provided by bands, balls or therapeutic putty.
Therapists followed a predefined protocol for prescribing the dose (sets, repetitions and load) of each strengthening exercise. Intensity was set using the Borg rating of perceived exertion scale (Borg, 1982) and each exercise was progressed or regressed according to both participant capability and therapist judgement. The goal was for the participant to perform each exercise, where possible at a volume and load that was achievable while still providing a stimulus for physiological change (Heine et al., 2012).

| Data collection
We used data provided by participants at baseline and 4-month followup. Data describing the prescribed hand strengthening exercise was extracted from the exercise treatment logs that were completed by therapists at each exercise session , including dose parameters (sets, repetitions and load). Where exercise treatment logs contained insufficient/ambiguous information about dose, we utilised the personal exercise diaries completed by both the therapist and participant to assist with completion. for 3 months or more were recruited. 244 participants were randomly assigned to usual care and 246 to the tailored exercise programme.

| Participants
Usual care included information published by Arthritis Research UK, joint-protection education and, where indicated, functional splinting.

| Ethical approval
Ethical approval for this study was granted by Brunel University

| Prescribed dose of hand strengthening exercise
The overall dose of hand strengthening exercise that participants were prescribed at the five supervised exercise sessions was 900calculated using the area under the curve (AUC) method. This approach has previously been used for identifying response to an intervention and considers the change in the value of a parameter over time (Matthews et al., 1990;Pruessner et al., 2003). We focussed exclusively on the dose prescribed and completed at the five supervised exercise sessions due to well recognised problems with recording exercise adherence to home exercise (Nicolson et al., 2018). Table 1 provides a guide for a participant attending all five supervised exercise sessions. For example, if the participant was prescribed 1 � 10 repetitions using yellow band, ball or therapeutic putty for each of the four strengthening exercises used in the SARAH programme, the prescribed overall dose would be 640 AUC. A detailed description of the approach is provided (see Additional file 1).

| Candidate predictors for prescribed dose
Based on theoretical knowledge and the clinical experience of physiotherapists within the research team, we selected candidate predictors potentially associated with the overall dose prescribed (Table 2).

| Hand function and grip strength
Hand function and grip strength at 4 months follow-up (closest to the supervised exercise sessions ending) were the outcomes used in our model to evaluate association with exercise dose. MHQ overall hand function was the primary outcome in the SARAH trial, and hand grip strength is known to partially mediate overall hand function (Hall et al., 2017).

| STATISTICAL ANALYSIS
Data was analysed using IBM SPSS Statistics for Windows,

| Analyses 2 & 3
4.2.1 | Steps 1. Each predictor variable identified in analysis 1 with a p < 0.10 was included in the univariate analysis.
2. GEE Univariate analysis: Along with prescribed overall dose (independent variable), we evaluated whether each predictor variable was associated with the dependent variable of outcome (4month overall hand function or full-hand grip strength). Adjustment was made for baseline overall hand function and full-hand grip strength respectively. Only complete cases were used.
3. Variables associated with both prescribed overall dose and outcome (p-value < 0.05) were included in the multivariate analysis.

| Prescribed overall dose
Overall dose of hand strengthening exercise was calculated for 222/ 246 (90.2%) participants (Table 3). Of the four exercises, gross grip had the highest overall dose prescribed compared to eccentric wrist extension, finger adduction and finger pinch.

| Analysis 1 (factors associated with prescribed overall dose)
Following univariate analysis (

| Analysis 2 (overall hand function)
Of the 246 participants, we excluded 29 (11.7%): 24 where the outcome (overall hand function) was missing at baseline and/or 4month follow up and 5 because their exercise treatment logs were recorded as missing (e.g., unable to calculate dose). Potential con- Limited evidence exists to ascertain the most effective dose of hand exercise in RA. Higher intensities of exercise is tentatively recommended over lower intensities (Bergstra et al., 2014;Hammond & Prior, 2016). This study supports clinicians aiming to prescribe higher overall dose with their patients in order to achieve better outcomes.
What is less well understood from our results, is whether volume (i.e. sets and repetitions) is more or less important than load (i.e. resistance) used. Exercise-based clinical trials may better inform the development of future guidelines if more detailed dose-response information is offered as part of the dissemination process.
Both therapist professional background and job grade (our surrogate for clinical experience) were associated with dose. Clinician professional background, years of experience, knowledge, beliefs, attitudes, and behaviour towards exercise have previously been associated with how exercise is prescribed in the rehabilitation setting (Bennell et al., 2014;Eulenburg et al., 2015;Hansen et al., 2018). Differences in professional training may be one possible reason. Physiotherapy as a profession appears to place a stronger emphasis on movement and exercise (CSP, 2018b). Consequently this profession maybe more confident to progress participants or pre- Swelling has been proposed to influence the range of joint movement and grip strength (Fraser et al., 1999;Scott & Houssien, 1996).
Participants with greater grip strength measured at baseline had a higher dose. Previous research suggests grip strength has been identified as an important marker for hand function (Higgins et al., 2018). Two participant psychological factors (mood and confidence) were found to be associated with dose. Exercise and its positive effects on mood are well known (Cooney et al., 2013). Less understood are the effects of mood on dose prescribed in exercise-based clinical trials. Depression is considerably higher amongst individuals with RA (Katon & Schulberg, 1992). Participant reporting of feeling downhearted or low all of the time was associated with lower dose being prescribed. Higher participant confidence to exercise without fear of making their symptoms worse was associated with higher prescribed dose. In a qualitative study that interviewed SARAH trial participants, confidence was identified as a facilitator for performing and adhering to the exercises (Nichols et al., 2017). Those participants with lower confidence levels may need more support to engage and progress the exercises. Evaluating these factors may help therapists to work with participants to achieve greater doses of hand strengthening exercise.

| STUDY STRENGTHS AND LIMITATIONS
This study utilised a relatively large trial data set and our analyses have controlled for a range of variables relating to baseline function, condition severity and participant characteristics. We recognise our study has some important limitations. These findings are based on observational data from within an RCT and our analyses were not pre-planned as part of that trial. Whilst we believe the overall dose calculated for the five supervised sessions acts as a reasonable proxy for dose completed over the 12 week programme, it may not fully reflect changes in the participants ability to perform the strengthening exercises (for example during symptom flare-up, injury or illness). We assigned a numerical rating to each level of resistance to help with calculating dose. This was because we were unable to obtain information on resistance level for exercise balls and putty (i.e. colour equating to kg). We accept this approach may have influenced the overall dose calculated. There may also have been other factors influencing prescribed dose (e.g. therapists knowledge/training, beliefs or access to equipment such as exercise band/putty). Considering some of the difficulties in calculating overall dose, future trials using strength-based exercise should explicitly report load/resistance in metric terms (e.g. kilograms).

| CONCLUSION
Clinicians using the SARAH hand programme with their patients should aim to prescribe higher overall dose of strengthening exercise if they wish to help patients achieve greater overall hand function and grip strength. Further research into understanding about how therapists select, weight and combine information gathered during the healthcare consultation when prescribing dose may be useful for informing future clinical practice.