Study design
A randomised, controlled, single blind, between-group experimental design was used. Eligible participants were randomised to receive either moderate intensity aerobic exercise (control condition) or high intensity aerobic exercise (active condition), during two separate test sessions.
Randomisation
A randomisation sequence was computer-generated and held by the Physiotherapy Clinic supervisor at Curtin University. Randomisation was stratified for males and females. Prior to commencing each testing session, the research assistant contacted the holder of the allocation schedule to ascertain group allocation for each participant. This research assistant conducted the aerobic exercise sessions. The primary investigator (PI) (AM) who undertook all testing remained blind to group allocation.
Participants
A gender-stratified convenience sample of 68 participants with LE, aged between 18 and 60 years, was recruited from Perth, Western Australia. Recruitment took place from October 2017 until June 2018 through radio advertisements and a clinical trials recruitment agency. LE was diagnosed based on the criteria established by Haker and Lundeberg [18] (unilateral lateral elbow pain for a duration of at least 6 weeks reproduced on at least two of the following: palpation of the lateral epicondyle, isometric testing of the wrist extensors, middle finger extension test, passive stretch of wrist extensors, resisted hand gripping using a dynamometer, and upper limb neurodynamic test with radial nerve bias (ULNDT-RN)). Exclusion criteria included history of surgery/fracture (past 2 years), history of neurological/radicular dysfunctions, steroid injection (preceding 6 weeks), contraindication to cold application, current use of antidepressants for > 12 weeks, and the presence of widespread arthritis.
Prior to commencing the study, each participant underwent a thorough clinical examination by an experienced musculoskeletal physiotherapist, to confirm eligibility. Participants were also required to complete the Adult Pre-exercise Screening System (APSS) tool [26] to ensure they were eligible for the aerobic exercise intervention. Participants were asked to abstain from taking pain medications 24 hours prior to testing and to avoid any additional physical treatment 3 days before and on the testing day. Written informed consent was obtained from all participants.
Curtin University Human Research Ethics Committee approved the study (HRE2017-0198-02). The study was also prospectively registered on 9/2/2017 with the Australia New Zealand Clinical Trials Registry (ACTRN12617000219381). Further, the study adheres to CONSORT guidelines (http://www.consort-statement.org/) for reporting clinical trials.
Procedure
After confirming eligibility, the PI tested all participants for PPT at the affected elbow and at a site just proximal to the wrist [24]. These PPT values were used as the baseline value for the initial aerobic exercise response (Baseline 1). Following initial baseline PPT measurement, the PI left the laboratory and the aerobic exercise session was conducted under the supervision of a research assistant who had received training in the exercise protocol. Participants were allocated to receive either moderate (50% maximum heart rate (HRmax)) or high intensity (75% HRmax) aerobic exercise based on the randomisation schedule. Each participant completed two sessions, both at the same exercise intensity, three days apart. Following completion of the cycling exercise, the PI re-entered the laboratory and conducted either a CPM or a MIA assessment protocol (order randomised). In both cases, a second set of PPT measures (Baseline 2) were taken before the CPM or MIA intervention was applied. Additional PPT measures were then taken during and post the CPM and MIA protocols, as described below. All PPT assessments were performed by the PI, who remained blind to the experimental group assignment of each participant (Fig. 1).
Pain-related outcome measures
Pain-related measures of PPT, pain free grip (PFG) and ULNDT-RN were completed using previously reported methodology [24]. These were measured by an electronic digital algometer in kPa, an electronic digital dynamometer in N, and an M180 twin axis electrogoniometer in degrees, respectively. PFG and ULNDT-RN were used as secondary measures of the analgesic effect of the cervical lateral glide mobilisation. All measures were obtained in triplicate. Mean values were used in the analysis. PPT measures were assessed at wrist and elbow test sites.
CPM and MIA assessment protocols
CPM and MIA were assessed using the same protocols as reported in a previous publication [24]. Immersion in cold water (10 °C) was used as the conditioning stimulus for CPM and a cervical contralateral lateral glide mobilisation of the C5/6 motion segment was used to induce MIA. PPT was tested in the affected arm at baseline prior to, at 1 minute during, and at 1 minute post cold water immersion or cervical glide mobilisation, respectively. At each time point, PPT was measured three times. The mean value of the three measurements at each time point was used for analysis. The relative change in PPT at each test site from Baseline 2 to during and post cold water immersion and cervical mobilisation was considered as the CPM and MIA effects, respectively.
Tennis Elbow specific assessment
Before physical testing, all participants were asked to complete the Patient Rated Tennis Elbow Evaluation (PRTEE), which evaluates pain and function over the preceding week [27].
Physical activity assessment
Participants also completed the Global Physical Activity Questionnaire (GPAQ) [28] to evaluate their physical activity levels. The total GPAQ score was calculated using the GPAQ guidelines [28] and expressed as Metabolic Equivalents (MET)-minute/week. The GPAQ is a reliable measure of physical activity with moderate to strong concurrent validity [29].
Experimental conditions
The method described here is based on a study by Naugle et al. [30]. Participants completed both stationary cycling sessions at either high or moderate intensity (randomly allocated) for 15 minutes. The PI was not present in the laboratory at any time during the exercise sessions. Before starting the first session, a target heart rate (THR) was calculated for each participant by the research assistant, based on age-predicted maximum heart rate (HRmax), where maximum HR = 220-age [31]. THR for those in the high intensity group was determined using the formula HRmax × 0.75 and for the moderate intensity group it was calculated as HRmax × 0.50.
All participants completed their exercise sessions using a cycle ergometer (828E Ergometer, Monark, Vansbro, Sweden). Heart rate was monitored during exercise using a chest heart rate monitor (Monark Heart Rate Monitor, Monark Exercise AB). The targeted exercise intensity level was achieved through adjusting the speed and the resistance of the cycle ergometer. Participants started the exercise session by cycling at low intensity (HR = 40% HRmax) for two minutes to familiarise themselves with their preferred cadence. The resistance was then gradually increased over the next three minutes (i.e. HR was elevated by an average of 10% and 3% per minute for high and moderate intensity exercise, respectively) to reach the desired target heart rate by the end of the first 5 minutes. Participants then continued cycling for the following 10 minutes, while maintaining the THR. Heart rate was continuously monitored to stay within a range of 10% above and 5% below the THR. Every five minutes during the cycling session participants were instructed to rate their perceived exertion (RPE) using the Borg-Scale (6–20) [32]. Heart rate (beats/minute) and workload (watts) were recorded every minute during the first five minutes and every 30 seconds and one minute, respectively, during the main exercise session. Mean RPE, HR and workload data collected during the 10 minutes of cycling at the THR were used for analysis.
Sample Size calculation
Sample size calculations were conducted using Stata/IC (version 15.0: StataCorp LLC, TX). Based on data from a large clinical trial comparing corticosteroid injections and physiotherapy management of tennis elbow [33] the minimal clinically important difference (MCID) in pressure pain threshold at the elbow was considered to be 88 kPa [34]. In determining our sample size, we used a more conservative difference value of 50 kPa (just above half of the MCID), as we expected that the influence of the aerobic exercise intervention assessed in this study would be more subtle than the influence of a corticosteroid injection, with a pooled standard deviation of 73.22 kPa resulting in an effect size difference of 0.68. An a priori power analysis (alpha = 0.05, beta = 0.80) indicated a required sample size of 68 (34 per group).
Statistical analysis
Data were analysed using Stata/IC (version 15.0: StataCorp LLC, TX). For all analyses, P < 0.05 was considered statistically significant. Descriptive statistics were based on frequency distributions for categorical data (gender and elbow tested) and means and standard deviations (SD) (age, PRTEE and RPE) or medians and interquartile ranges (IQR) for continuous data (duration of LE, GPAQ, HR, workload). Univariate group comparisons between intervention groups at baseline and during exercise sessions included χ2 and Fisher exact tests for categorical comparisons, and independent t-tests or Mann-Whitney U tests for continuous outcomes, as appropriate.
All outcome data were evaluated for normality using Shapiro-Wilk tests and graphical review. Non-normally distributed data (PPT, PFG, ULNDT-RN) were transformed using natural logarithms.
Linear mixed models with random subject effects were used to evaluate differences (relative to baseline measures) between time points (all participants) and between exercise groups over time for CPM and MIA measures for all outcome variables (PPT, PFG and ULNDT-RN). The respective marginal means, 95% confidence intervals (CI), and p-values of these differences were calculated. The analysis was controlled for PRTEE, GPAQ and sex.
Partial correlations and univariate regression models were used to determine the relationships between exercise induced analgesia and CPM and MIA, measured both during and post cold water immersion / cervical mobilisation at both test sites. The strength of the correlations were interpreted according to the guidelines defined by Cohen [35]: (small: 0.10 ≤ r ≤ 0.29; medium: 0.30 ≤ r ≤ 0.49; large: 0.50 ≤ r ≤ 1.0). Univariate regression models were used to calculate regression coefficients (B), and their 95% CI and p-values. The adjusted coefficients of determination (adj. R2) were also calculated in order to determine the proportion of variability in CPM /MIA PPT (dependent variable) that was explained by post cycling PPT. Due to the anticipated between-individual variability in PPT, baseline PPT (Baseline 1) was identified as a potential confounder for the association and therefore it was adjusted for in the partial correlations and regression analyses.