Optimising corticosteroid injection for lateral epicondylalgia with the addition of physiotherapy: A protocol for a randomised control trial with placebo comparison

Specific elbow manipulation techniques known as Mobilisation with Movement (MWM) will be applied, as described by Vicenzino (2003) [23], based on evidence of their immediate effects on pain and improved pain-free grip force; [24, 25]. A hydraulic grip dynamometer will be used to measure baseline pain-free grip force and the effect of the following glides – lateral elbow glide, postero-anterior radioulnar glide and de-loading of the common extensor origin [23]. With the participant's arm supported in elbow extension and pronation, the therapist will sustain a glide while the participant slowly performs a pain-free grip over approximately 6 seconds. If substantial improvement in pain-free grip force is noted with the glide application, the technique may be repeated 6–10 times during a single treatment session. If successful, self-treatment techniques (Figure 2) will be taught to the participant using dynamometer feedback in the clinic.

A comprehensive exercise program will be used to primarily address motor impairments, but also to address pain and stimulate tendon remodelling [22]. Three main groups of exercises will be pragmatically prescribed: (1) Sensorimotor retraining of gripping and forearm movements (Figure 3) and posture correction will be commenced early in the physiotherapy intervention. These exercises are based on identified motor control deficits [26‐28] and previous 'Occupational training' exercises for LE [29]. (2) Progressive resistance exercise for the wrist extensors will be prescribed based on identified strength deficits in LE [30, 31] and proposed effects of exercise on tendon remodelling [32, 33]. Combined concentric and eccentric exercise will be performed, given insufficient evidence for use of eccentric over concentric modes in LE [34, 35]. Thera-Band™ tubing of varied resistance levels with an attached handle will be used to provide progressive resistance (Figure 4). Restriction of painful segments of range will be recommended in early rehabilitation, with progression to maximal positions of loading involving elbow extension and forearm pronation as tolerated [36]. (3) Exercises geared towards general arm strengthening will be performed with free weights or weight bearing anti-gravity exercise to address known proximal and bilateral strength deficits found in LE sufferers [31]. Progression of these exercises to include work or sport-specific exercises may occur later in rehabilitation.

Physiotherapists will prescribe exercises based on the participant's capabilities at any given session to allow for optimal exercise volume and load setting without exacerbating pain [23]. The overriding rule for all exercise is that pain should not be provoked during or after exercise, including avoidance of delayed onset muscle soreness. Exercises will be performed in a slow manner with sufficient rest between sets to allow recovery, and correct form and posture emphasised throughout. Supervision of the home program at the commencement of every session and monitoring of exercise diaries by the treating practitioners will be used to facilitate program adherence and emphasise the importance of the exercise and self-treatment program.

The participant's self-perceived level of improvement will be measured with a 6-point Likert scale with categories: completely recovered, much improved, improved, no change, worse and much worse [10, 20]. The participant will be asked at each follow-up outcome assessment to rate the change in their elbow condition since their commencement in the study. A similar retrospective assessment scale has been shown to be more sensitive to change than serial visual analogue scale measures and more correlated with patient's satisfaction with change [37].

Success will be derived from the global perceived improvement scale as per previous work [10, 20]. The categories 'completely recovered' and 'much improved' will be dichotomised as 'success', while the other categories ('much worse' to 'improved') will be collapsed to represent 'no-success'.

Recurrence will primarily be defined as occurring when a participant rates a success at 4 or 8 weeks and a no-success beyond 8 weeks on the global perceived improvement scale [10]. This definition of recurrence which evaluates the pattern of recurrent events over time has been used in previous studies of LE [10, 20, 38]. In addition, participants will be questioned using a semi-structured interview at 12, 26 and 52 weeks about their experience of symptom aggravation and possible reasons for their recurrence. This data will enable better understanding of recurrences and best practice measurement of recurrence rates in future analyses and trials.

Two visual analogue scales (VAS) anchored by 'no pain' (0 mm) and 'worst imaginable pain' (100 mm) will be used measure the severity of participants' resting pain and worst pain experienced during the preceding week. The VAS is considered the most sensitive of all pain rating scales and has been specifically evaluated in the LE population with high test-retest reliability (r = .89) and moderate correlation with pain-free grip (r = .47) [39].

The Patient-rated Tennis Elbow Evaluation (PRTEE) questionnaire will provide a standardised quantitative assessment of pain and functional disability. It has been shown to have excellent test-retest reliability (r = .93) and good correlation with other functional scales including the Disability of Arm and Shoulder (DASH) questionnaire (r = .87) in the LE population [40]. Questions are scored on an 11-point Likert scale, with calculation of separate subscales for pain and function and a total score, ranging from 0 (no pain and no functional disability) to 100 (worst imaginable pain with a very significant functional disability) [40].

PFG force is well established as a highly reliable (ICC >.97) and convenient clinical assessment tool, which correlates more strongly with disability and perceived improvement than maximal grip strength in LE populations [41‐43]. PFG force will be measured using a digital grip dynamometer with variable handle position (MIE, Medical Research, UK). The participant will be positioned in supine with the tested elbow in relaxed extension and pronation [6]. The participant will be instructed to maximally squeeze the dynamometer on the unaffected side. On the affected side, the participant will be asked to grip the dynamometer at the same rate as the unaffected side but to stop the instant pain is experienced. The average of three repetitions with 20 second rest intervals will be used in further analyses.

Pressure pain threshold (PPT) will be used as quantitative measure of mechanical hyperalgesia over the lateral epicondyle as per previous studies [10, 20]. Pressure will be applied using a digital algometer with a probe size of 1 cm² (Somedic AB, Farsta, Sweden) at a rate of 40 kPa/s until the first sensation of pain is recorded. Triplicate recordings will be taken at each follow-up assessment and the mean values used for analysis. Measurements will be performed by the same blinded assessor (BC) as reliability has been found to be reduced between observers (r = .72–.98) [41].

The Hospital Anxiety and Depression Scale (HADS) will be used to identify and quantify the two most common forms of psychological disturbances – anxiety and depression [44]. It has been demonstrated to be an appropriate measure in musculoskeletal pain [45], with evidence of elevated levels of both anxiety and depression in the LE population [46]. HADS consists of 14 items which are independent of somatic symptoms and divided into two subscales for anxiety and depression. Each item is rated on a four point scale, with scores of >11 on either subscale considered to be a significant (probable or definite) case of psychological morbidity, whereas scores of 8–10 represent borderline (possible or doubtful) case and 0–7 normal (non-case) [45].

The Tampa Scale for Kinesiophobia (TSK) will be administered at baseline to assess the degree of kinesiophobia, also known as fear of movement or (re)injury [47]. While it has not been studied in the LE population, previous study of non-traumatic complaints of the neck and arm found positive associations between kinesophobia and disability and co-morbidity of musculoskeletal complaints [48]. The adjusted version of this scale (TSK-AV) will be used due to reported improved factor structure [48]. Each of the 13 items are scored on a 4-point Likert scale giving a total score ranging from 13 to 52, with higher scores indicating greater kinesiophobia.

The EuroQol EQ-5D instrument will be used to measure health-related quality of life, expressed as utility values ranging from 0 to 1, where 1 represents perfect health[49]. The utility weights captured by these preferences will enable the derivation of the Quality Adjusted Life Years (QALY) for each intervention and will be used in cost-utility analyses. One advantage of using this instrument is that results of this study can be compared with those of a previous trial which compared corticosteroid injection, physiotherapy and a wait and see policy for LE [17].

Cost data will be collected from a societal perspective, meaning that an attempt to capture all of the costs, regardless of which party or parties incur them, will be made. It will include a) direct health care costs, such as visits to doctors, therapists, investigations and prescribed medication; b) direct non-health care costs, such as costs of over-the-counter medication, hours of paid and unpaid household help, transportation and other out-of-pocket expenses; c) indirect costs, such as absence from work, housekeeping and other daily activities [50]. Participants will be asked to recall their out-of-pocket expenses incurred over the preceding month during a structured telephone interview by a research assistant not involved in outcome measurement. The initial interview date will be randomised by a computer-generated number sequence to 4, 8 or 12 weeks following commencement of the study and then subsequent interviews performed at three monthly intervals. This method of sampling was chosen to provide sufficient information while minimising the time commitment required of participants [51]. In addition, participants will be asked to provide consent for their Medicare data on in-hospital, out-of-hospital and pharmaceutical services to be provided by Medicare Australia. These data will capture expenditures incurred under Australia's universal, compulsory health care financing scheme, Medicare. This includes all expenditures incurred for general practitioner and specialist services, as well as pharmaceuticals that are listed on the Pharmaceutical Benefits Scheme.

A contingent valuation study will be conducted to elicit an individual's willingness-to-pay (WTP) for relief of the symptoms of LE. The contingent valuation approach results in monetary valuation of the benefits produced by a health program or health state [52]. An advantage of this approach is it allows valuation of transitory health states, which may be important given previous trials suggest considerable temporal variance in the effectiveness of interventions for LE [10]. Participants will be questioned at baseline assessment regarding the greatest amount they are willing to pay, based on the following hypothetical scenario: "Suppose that your current elbow pain will persist over the next 12 months if untreated. Imagine that there is a new treatment that is quick and non-invasive. If it works, it will provide an immediate and complete cure of your tennis elbow. However, this treatment only works on 1 in 2 people. The treatment is not covered by Medicare or private insurance." Bias will be minimised by using both a bidding game approach and a binary response approach. The bidding game approach involves exposing respondents to repeated bids and identifying the point of indifference between those bids. In the binary response approach, the respondent is asked to respond to only one bid. In both approaches, four bids ($50, $650, $1250 and $1850 in Australian Dollars) will be randomly allocated to respondents. The randomisation of bids is a further measure to reduce the bias that may be generated by the choice of a starting bid [53]. Specifically, it is known that high (low) starting bids may lead to higher (lower) WTP valuations. The starting bids will be based on the mean ± 2 standard deviations derived from a similarly designed pilot study of WTP in LE participants.

All adverse events, defined as any negative or unwanted reactions to intervention, will be recorded. These will include pain (lasting longer than 7 days), skin pigmentation changes, subcutaneous atrophy or any other reported physical discomfort. As well as the adverse effects committee (chaired by BV) that will manage adverse reactions at the time of the adverse reaction, we will administer a questionnaire at 8 and 52 weeks to record all such patient identified events. This will capture additional ill effects not reported at the time of their occurrence. These are likely to be minor aggravations in symptoms. The questionnaire will use open-ended questions and will be placed in a sealed envelope to be opened by a research assistant not involved in outcome measurement. The intervention-specific incidence of each adverse event will be computed.

All analyses will be conducted on an intention to treat basis by an investigator who is blind to group allocation. The outcomes measured at 4, 8, 12, 26 and 52 weeks will be used to generate efficacy and recurrence indices that will be analysed using linear mixed and logistic regression models [10]. Baseline scores for each dependent variable will be entered as a covariate, relevant participant characteristics entered as a random effect and treatment conditions and time as fixed factors. Variables such as age, gender, duration of condition, nature of employment will be included as covariates in the analysis, if found to significantly influence outcomes over time. Regression diagnostics will be used to check for normality of the measures and homogeneity of variance where appropriate. Alpha will be set at 0.01 to compensate for the possible increase in type I error rates that may result from multiple testing. The dichotomous measures of success and recurrence will be analysed using relative risk, and numbers needed to treat in order to provide a meaningful indicator of treatment efficacy to practitioners.

Participants' utility weights will be derived from EQ-5D responses. These QALY computations will be used along with cost data to enable cost-effectiveness analyses to be undertaken. As Australian weights for the EQ-5D are currently unavailable, we propose to use the available scoring algorithms for the UK and NZ to compute QALY estimates, with the variation between results forming the basis for a cost-per-QALY sensitivity analysis [49]. Capturing participants' WTP for treatment will provide an alternative measure of the marginal benefits associated with intervention and will be used in cost-benefit analyses. Care will be taken to avoid double counting the benefits estimated via the contingent valuation approach. The distribution of the WTP data will be analysed to examine not only the distribution of the responses, but also the effect of the starting bid and of respondent income on WTP. Modern statistical procedures, such as bootstrapping, will be used to assess the uncertainty surrounding costs and effects.