Abstract
Background Correlations between clinical and quantitative measures of pain sensitivity are poor, making it difficult for clinicians to detect people with pain sensitivity. Clinical detection of pain sensitivity is important because these people have a different prognosis and may require different treatment.
Objective The purpose of this study was to investigate the relationship between clinical and quantitative measures of pain sensitivity across individuals with and without neck pain.
Methods This cross-sectional study included 40 participants with chronic neck pain and 40 age- and sex-matched controls. Participants underwent quantitative sensory testing of cold pain thresholds (CPTs) and pressure pain thresholds (PPTs). Clinical tests for pain sensitivity were the ice pain test and the pressure pain test. All tests were undertaken at standardized local (neck and upper trapezius muscles) and remote (wrist and tibialis anterior muscles) sites. Median and interquartile range (IQR) were calculated for neck pain and control groups, and parametric and nonparametric tests were used to compare groups. Correlation coefficients were calculated between quantitative and clinical measures.
Results There were significant differences for clinical and quantitative measures of cold and pressure sensitivity between the neck pain and control groups (eg, CPT neck pain group: median=22.31°C, IQR=18.58°C; control group: median=5.0°C, IQR=0.74°C). Moderate-to-good correlations were found between the clinical ice pain test and CPT at all sites (.46 to .68) except at the wrist (.29 to .40). Fair correlations were found for the clinical pressure pain test and PPT (−.26 to −.45). Psychological variables contributing to quantitative measures of pain sensitivity included catastrophization, sleep quality, and female sex.
Limitations Clinical pressure pain tests were not quantitatively standardized in this study.
Conclusions The ice pain test may be useful as a clinical correlate of CPT at all sites except the wrist, whereas the pressure pain test is less convincing as a clinical correlate of PPT.
Neck pain, whether idiopathic or caused by a whiplash injury, is a global health burden. It has been ranked as the fourth leading cause of years lived with disability for the last 2 decades.1 Data from systematic reviews tracking the course of recovery for both idiopathic neck pain2,3 and whiplash4,5 indicate that recovery is poor for many people, with at least 50% having long-term pain and disability. Factors that are associated with nonrecovery after neck pain are multifactorial and include psychological and symptom-related factors. The most consistent factor associated with nonrecovery is high initial self-reported levels of pain.2–5 Further work has shown that quantitative measures of pain sensitivity6–8 (in particular, increased cold pain sensitivity9–12) are associated with poor outcomes in neck pain. Therefore, detecting people with pain sensitivity in primary care is critical to identifying patients at greater risk of nonrecovery.
Quantitative sensory testing (QST) is a laboratory-based tool used to assess pain sensitivity.13 It has been used to characterize somatosensory phenotypes of various painful neuropathic conditions14 and has increasingly been used to test for the presence of pain sensitivity in cervical pain states.15–17 Laboratory equipment is able to quantitatively test both cold and pressure sensitivity components of the QST test battery18; however, this equipment is costly and unavailable to primary care clinicians. Lacking, therefore, are clinical tests that correlate with quantitative tests, which, in turn, would enable primary care clinicians to identify people with pain sensitivity.
To date, QST has been evaluated against clinical measures such as subjective reports of pain intensity and self-reported disability rather than more directly related physical measures. Correlations between subjective reports of pain intensity and disability have a weak association with quantitative measures of pressure pain thresholds (PPTs), as shown in primary cross-sectional studies19 and in our recently published meta-analysis.20 It could be argued, therefore, that a closer construct to PPT, a physical application of pressure, would be pressure applied manually. To date, however, correlations between manual application of pressure and PPT have not been investigated.
In contrast, manual application of ice has been suggested as a clinical method for assessment of cold pain sensitivity.21 To date, however, only one study has formally examined the association between this clinical ice test and the QST measure of cold pain threshold (CPT).22 In that study,22 the researchers applied ice to the skin, finding that a pain intensity rating of >5/10 indicated a 90% likelihood of cold hyperalgesia, assessed by CPT, being present. The main limitation of the study by Maxwell and Sterling is that only one body site was measured rather than multiple sites, which are often used to determine the presence of widespread pain sensitivity. This limitation is potentially important because there is some evidence that people with neck pain who have widespread pain sensitivity, including widespread cold pain sensitivity, have higher levels of symptoms and poorer prognosis than those with localized pain sensitivity.23
Also complicating the interpretation of QST results is the finding of pain sensitivity in pain-free individuals.24,25 Previous studies of a similar nature have often not included control participants.19,22 This lack of controls was considered important in the current study, as clinical correlates should be associated with the quantitative measure regardless of the pain status. In other words, clinicians need to be assured that the chosen clinical measure correlates with the quantitative measure in both individuals who are asymptomatic and those with presence of clinical pain.
Finally, pain sensitivity is understood to be highly influenced by supraspinal processes.26 In neck pain, factors known to have an association with either CPT or PPT include general psychological distress, anxiety, catastrophizing, and fear avoidance.19,27 Recently, poor sleep also has been identified as a factor related to pain sensitivity.27 It is important, therefore, to consider these factors in any work investigating pain sensitivity.
This study, therefore, aimed to determine the association between clinical and quantitative tests of pain sensitivity in people with neck pain as well as pain-free controls. Specifically, we aimed to investigate correlations between: (1) pain ratings of manually administered pressure and laboratory measures of PPT and (2) pain and cold ratings with the application of ice with laboratory measures of CPT. An additional aim was to determine the psychological and clinical tests that are associated with the quantitative measures. In so doing, this information aims to facilitate better identification of pain sensitivity in clinical practice.
Method
Study Design and Participants
A cross-sectional study investigating people with chronic neck pain (n=40) and healthy age- and sex-matched controls (n=40) was conducted. Forty participants with chronic neck pain seen in private physical therapy practices in Sydney and Perth, Australia, participated in the study. Participants with chronic neck pain aged 18 to 65 years were included if they had: neck pain (with or without arm pain) of at least 3 months' duration, neck pain intensity of at least 3/10 on an 11-point numeric rating scale (NRS), and neck disability of at least 10/50 on the Neck Disability Index (NDI).28 People with neck pain were excluded if they had low disability (<10/50 on the NDI), cervical radiculopathy, known pathology such as fracture or tumor, or any implanted medical device such as a pacemaker.
Control participants were included if they had no history of neck pain or current neck pain. They were matched for age, sex, and local testing site with participants in the neck pain group to control for the known variation in QST measures according to both age and sex.29 Control participants were recruited in response to advertisements placed at the University of Sydney and Curtin University. Participants were recruited between October 2012 and December 2013. All participants provided written informed consent.
Procedure
All participants completed baseline demographic data and questionnaires covering symptoms, disability, and psychological status. Participants then attended on one occasion for clinical and quantitative (laboratory) testing. Clinical and quantitative testing were conducted independently by 2 different investigators, each blinded to the results of the previous examination.
Demographic Information and Questionnaire Assessment
Demographic data collected included age, sex, marital status, education level, and occupation. Symptoms assessed were average neck pain intensity and average referred pain intensity over the previous 24 hours and over the previous week as measured with the NRS. If several areas of referred pain were noted by the participant, the worst area of referred pain was recorded. Disability due to neck pain was assessed using the NDI,28 which has been extensively evaluated in neck pain populations.
Several psychological constructs were chosen to be assessed due to their reported association with pain sensitivity in musculoskeletal pain states, including neck pain. The short-form Depression Anxiety Stress Scale (DASS-21) was administered to assess stress, anxiety, and depression.30 The Pain Catastrophizing Scale31 (PCS) was administered to assess negative orientation toward pain and is a known construct associated with poor outcomes in neck pain states.8 Sleep quality was measured using the Pittsburgh Sleep Quality Index32 (PSQI), given the potential influence of sleep on pain sensitivity. The self-administered comorbidities questionnaire was used to document the presence of other health disorders.33
Clinical Testing
All participants attended one clinical assessment involving a standard clinical physical therapy neck assessment with the addition of the novel clinical tests to detect pain sensitivity. Clinical testing was conducted by specialist or titled musculoskeletal physical therapists (T.R., D.B., N.M., R.A.). The physical therapists initially conducted an upper limb neurological examination and the Spurling test34 to exclude cervical radiculopathy.
The first clinical test conducted to assess for abnormal sensory processing was the Brachial Plexus Provocation Test (BPPT), which assesses the sensitivity of the brachial plexus to movement.35 This test has been evaluated in participants with neck pain and neck-related arm pain, with findings that sensitivity of the neural tissue to movement during the BPPT is associated with nervous system sensitization.15,16,36 Participants were asked to indicate to the physical therapist when they first perceived the onset of pain anywhere along the tested arm. A test was considered positive if it reproduced the participant's symptoms (or concordant symptoms) and structural differentiation indicated that the sensitized structure was neutral. With the NRS, participants rated pain intensity experienced at onset of their arm symptoms or at the onset of increased pain if they had pain at rest.
Participants then underwent clinical testing for pressure and cold sensitivity at both local and remote sites. The physical therapist performed a routine manual examination of the cervical spine to identify the most painful cervical level, which was used as the local test site. A second local site used was in the upper trapezius muscle, at the midpoint between the C7 spinous process and the acromion on the most symptomatic side. The remote sites tested were over the capitate bones of both wrists and over the tibialis anterior muscle. We considered the symptomatic wrist as the most symptomatic side for participants with neck pain, namely where pain may be referred from the cervical spine into the “symptomatic” limb. The tibialis anterior muscle site was located at the mid-substance muscle belly of the tibialis anterior muscle, located 5 cm distal to the tibial tubercle and 2 cm laterally for the right side only.
Clinical Pressure Pain Test
Clinical testing of pressure sensitivity was performed by the physical therapist applying mild, moderate, and firm manual pressure at the sites specified above. Physical therapists standardized their forces, using verbal feedback from pilot-tested patients. As a guide, physical therapists considered mild force to be equivalent to a grade I pressure applied during a “Maitland mobilization,” moderate force as a grade II pressure, and firm force as a grade III to IV pressure.37(p175) Pilot-tested patients gave verbal feedback on the pressure applied by all physical therapists doing the manual assessment until pressures were perceived to be similarly applied by all physical therapists.
Participants were asked to rate their pain intensity at each pressure using the 11-point NRS. The pain ratings from each of these pressures were amalgamated into one scale, which was designed to weight pain due to mild pressure more strongly (ratings for pain at mild pressure ×3, at moderate pressure ×2, and at firm pressure ×1). This approach gave a total amalgamated score/60 for the complete clinical pressure pain test.
Clinical testing of cold sensitivity was performed using ice application to the above sites for 5 seconds. Participants were asked to rate their pain intensity at each site on the 11-point NRS (clinical ice pain test) and to rate cold on a similar 11-point NRS scale (clinical ice cold test). We used 5 seconds based on clinical piloting of ice application to minimize any temporal summation or wind-up due to repeated applications.
Control participants were matched for age and sex with participants in the neck pain group and had the same sites tested. For example, if a participant with neck pain was most symptomatic to manual palpation over the left C3 facet joint, this same site was tested on the matched control participant.
Quantitative Testing
Pressure pain sensitivity was assessed by recording PPT values using a handheld pressure algometer with a probe size of 1 cm2 (Somedic AB, Hörby, Sweden) applied at a rate of 40 kPa/s.38 Pressure pain thresholds were tested at the same sites as previously described. Participants were asked to press the stop switch as soon as the stimulus started to feel painful. The mean of 3 tests was recorded.
Cold pain sensitivity was assessed by recording CPT values using a thermal sensory testing system (MSA Thermal Stimulator, Somedic). Cold pain thresholds were tested at the same sites as outlined in the clinical examination above. Participants received a continuously descending temperature stimulus starting at 30 degrees and reducing at a rate of 1°/s. They were instructed to terminate the stimulus using a cutoff switch as soon as the stimulus started feeling painful.
Data Analysis
We used IBM SPSS version 19 (IBM Corp, Armonk, New York) to analyze the data. Data were assessed using Kolmogorov-Smirnov tests for normal distribution. Demographic data and results from self-reported questionnaires were analyzed using descriptive statistics. Median and IRQ were calculated for clinical and quantitative variables. Group differences were investigated using independent t tests for normally distributed data. These statistical analyses are similar to those used in our previous articles.15,39
Bivariate correlation coefficients (Spearman rho [ρ]) were calculated between clinical measures and their related quantitative variables (ie, clinical ice cold rating and clinical ice pain ratings were calculated with CPT, and clinical pressure pain ratings were calculated with PPT). Relationships according to correlation coefficients were adopted from previous studies20,40 and considered as follows: <.25=little or no relationship, .25−.5=fair relationship, .5−.75=moderate-to-good relationship, and >.75=good-to-excellent relationship.
To investigate the psychological and clinical tests that are associated with quantitative measures of pain sensitivity, hierarchical linear regression models with PPT and CPT as outcome variables were built for each site on the basis of significant correlations (P<.05). Variables that demonstrated at least a fair correlation (ρ≥.3) with the outcome variable were selected as predictor variables.
In order to assess the unique contribution of the clinical tests, 2 models were built. In the first model, all potential predictors except the clinical tests were forced into the model. In the second model, the clinical tests were entered in a stepwise manner using a combination of forward selection and backward elimination, with P≤.5 for entry and P≥.1 for removal. Regression models were performed for all data and the neck pain group only to remove any dependency created by the age and sex matching.
As the 2 pain intensity ratings (24 hours, last week) showed very high intercorrelations (ie, .9 and higher), a composite score was created. A sample size of 40 participants in each group was calculated a priori to have an 80% chance of detecting a moderate correlation at the 5% significance level.
Role of the Funding Source
Dr Rebbeck and Dr Beales are supported by a National Health and Medical Research Council Research Fellowship. Dr Hübscher was supported by a postdoctoral fellowship from the German Academic Exchange Service (DAAD).
Results
A total of 80 people participated in this study: 40 with neck pain and 40 age- and sex-matched controls. Demographic variables are shown in Table 1. A greater proportion of participants (n=50) were recruited from New South Wales compared with Western Australia (n=30). There were no differences between groups in terms of demographic variables; however, the neck pain group had significantly poorer scores for the DASS-21, PCS, sleep quality, and comorbidities. The most common symptomatic joint to palpation was C2 (50% of neck pain group).
Participant Characteristics and Group Differences for Pain, Disability, Psychological, and Health Status Domainsa
Group Differences
The differences between the neck pain and control groups for the clinical and quantitative data are provided in Table 2. There were significant differences between the 2 groups for quantitative measures of cold sensitivity (CPT) at all sites tested (P<.001). In addition, there were significant differences between groups for clinical measures of ice pain at all sites tested (P<.001). However, between-group differences for clinical measures of ice cold were significant only at the neck, wrist, and upper trapezius muscle sites (P<.05).
Group Differences for Quantitative and Clinical Measures of Pressure and Cold Pain Sensitivitya
There were significant differences between the neck pain and control groups for PPTs at all sites tested (P<.001). Clinical measures of pressure pain were significantly different at the neck, wrist, and upper trapezius muscle sites (P<.001) but were not significantly different at the tibialis anterior muscle site (Tab. 2).
Relationships Between Clinical and Quantitative Tests
Moderate-to-good correlations were found between CPT and clinical measures of ice pain at all sites tested (all data: ρ=.55 to .68; neck pain group only: ρ=.46 to .65; Tab. 3) except the wrist (all data: ρ=.40; neck pain group only: ρ=.29; Tab. 3). There were either no or fair correlations between CPT and clinical measures of ice cold for all data and neck pain group data only (Tab. 3).
Bivariate Correlation Analyses Between Quantitative and Clinical Measures for Cold Pain Sensitivitya
Fair correlations were found between quantitative measures of pressure sensitivity (PPT) and clinical pressure pain when considering all data (ρ=−.30 to −.44; Tab. 4) and the neck pain group only (ρ=−.26 to −.45; Tab. 4). In order to inform clinicians of the most appropriate force to apply, we also calculated the correlations between PPT and pressures applied at different forces (light, moderate, and firm). Correlations were generally higher if applied with light or moderate pressure (ranging from ρ=−.34 to −.55) when considering all data. When considering the neck pain group only, correlations also were generally higher and always statistically significant for moderate pressure (ranging from ρ=−.29 to −.37) compared with either light or firm pressure, where correlations were often not significant (eg, correlation between light clinical pressure and PPT at the neck was −.10; Tab. 4).
Bivariate Correlation Analyses Between Quantitative and Clinical Measures for Pressure Pain Sensitivitya
Summaries of the results of the regression models are provided in Tables 5 and 6. Variables that showed at least a fair-to-moderate correlation with CPT or PPT that were included in model 1 were sex, PCS score, DASS-21 score, sleep quality, pain intensity, and BPPT pain score. Age, which was initially considered associated with CPT and PPT, did not show a correlation. For CPT, these variables accounted for 29% to 45% of the variance, depending on the site tested (Tab. 5). When the clinical ice pain scores were added in model 2, a further 10% to 16% of the variance was explained when considering all data. When considering neck pain group data only, similarly a further 9% to 17% of the variance was explained with the addition of the ice pain test. Clinical ice pain remained the only factor that consistently predicted CPT across all body sites in the final regression models for all data and neck pain group data only (model 2; Tab. 5).
Regression Analyses for Variables Associated With Cold Pain Threshold at All Body Sitesa
Regression Analyses for Variables Associated With Pressure Pain Thresholds at All Body Sitesa
With regard to PPT variables, sex, PCS score, DASS-21 score, sleep quality, pain intensity, and BPPT pain score accounted for 35% to 43% of the variance in model 1 (Tab. 6). When the clinical pressure pain score was added (model 2), a further 1% to 5% of the variance was explained for all data and 0% to 13% for the neck pain group data only, depending on the site tested (Tab. 6). Significant variables associated with PPT were consistently female sex and clinical pressure pain scores for all body sites, except the upper trapezius muscle (only sex was significant). These associations remained significant when considering the neck pain group only.
Discussion
This study demonstrated that the clinical ice pain test has a moderate-to-good correlation with the laboratory measure of cold pain threshold in people with and without neck pain. However, only fair correlations between the clinical pressure pain test and laboratory measures of pressure pain threshold were found. Both clinical tests remained significant in multiple regression models predicting their quantitative equivalent after accounting for psychological, pain intensity, and sleep variables. These tests, therefore, could be used with some confidence in the clinic to assist with the identification of people with pain sensitivity.
Moderate-to-good correlations between the ice pain test and cold pain thresholds were found at all body sites tested except the wrist, where the association was only fair. In retrospect, the wrist as a test site may need to be reconsidered given the lower correlations and the insignificant differences between neck pain and control groups with the ice pain test (Tab. 2). It was chosen as a site based on the remote site used in low back pain studies41; however, it is noted that, in studies involving upper limb populations, the forearm is often a preferred test site.15 Anecdotally, we found the wrist relatively less sensitive to tests compared with the other upper limb sites. Hence, in future studies, an alternate site in the upper limb should be considered.
Otherwise, the reported moderate-to-good correlations corroborate the findings of the only other study22 investigating the ice pain test, where high pain scores with ice application predicted cold hyperalgesia at the neck. Correlations with the ice pain test in our data also were higher than those with the ice cold test, thereby suggesting the ice pain test is more closely related to CPT testing.
Our data also clearly demonstrate that contributions to cold pain sensitivity are not unidimensional. In this study, up to 43% of the variance in CPT was explained by psychosocial variables such as anxiety, depression, PCS score, and sleep quality. Of these contributions, the PCS score was the one variable that reached significance before the clinical ice pain test was added. This finding supports previous work where pain catastrophization27 was highly correlated with CPT in people with whiplash-associated disorder. However, given that the ice pain test remained the only significant factor in the regression models across all body sites when PCS and other variables were accounted for, we argue that the ice pain test should be considered associated with CPT.
The pressure pain test, however, showed only fair correlations with PPT. In addition, although significantly associated with PPT in the regression models, the clinical pressure pain test accounted for less of the variance associated with PPT than the ice pain test did with CPT. This result may have occurred because other factors such as female sex consistently contributed to PPT across all body sites. Pressure pain thresholds are known to be significantly less in women than men,24 a finding further supported by these data. Interestingly, sleep quality also was predictive of PPT for the neck and upper trapezius muscle sites. Sleep quality has more recently been found to be associated with pain levels but seems to be more consistently associated with mechanical hyperalgesia.42,43 We argue, therefore, that although the clinical pressure pain test was still correlated with PPT, its clinical utility as a substitute for PPT is less convincing compared with the ice pain test.
The limitations to this study may explain some of the low correlations between PPT and the clinical pressure pain test. First, we did not standardize the quantity of the force applied by the physical therapists as mild, moderate, and firm; hence, almost certainly the clinical pressure forces between physical therapists would have differed. However, when designing this study, we were attempting to replicate clinical practice as closely as possible, where standardization of manual forces is not possible. Second, both clinical and quantitative testers were not blinded to group allocation; hence, they may have subconsciously altered the force of their testing if they knew they had a “pain patient.” It is noted, however, that the correlations did not differ greatly when the pain participant groups were considered separately.
An important aspect of this study was that correlations were tested in healthy controls as well as participants with neck pain. It is known that pain sensitivity can occur in people who are asymptomatic24 and that not all of those with symptomatic neck pain will have pain sensitivity. Our median and IQR data demonstrate an overlap between quantitative measures of CPT (eFigure) and PPT between the neck pain and control groups. The finding of a spread of pain sensitivity profiles at each site among both healthy controls and people with neck pain leads to the question regarding the genetic contribution to pain sensitivity. Although research in this area is only recent, data are accumulating that between 30% and 60% of the variation in chronic pain syndromes may be due to heritable factors.44–48 Furthermore, data from more than 90 pairs of twins indicate that up to 60% of the variance in cold pressure pain sensitivity and 26% of the variance in heat pain sensitivity is genetically mediated.49 Specific genotyping studies have demonstrated that higher pain sensitivity occurs in people harboring specific genes or specific single nucleotide polymorphisms of genes (eg, the A118G single nucleotide polymorphism of the μ-opioid receptor gene),50 which also may be associated with increased risk of developing chronic pain conditions in these individuals.51
Future Research
Clinical detection of pain sensitivity is one step toward understanding prognosis and treatment for such individuals. Future longitudinal research is needed to investigate the relationship between clinical measures of pain sensitivity and long-term outcome. Identification of subgroups of pain sensitivity profiles (eg, local versus widespread pain sensitivity) utilizing clinical measures is needed. Determining cutoff scores for the clinical tests in longitudinal studies also may improve our understanding of their diagnostic accuracy. Other predictors of both CPT and PPT clearly still need to be found, given that our models accounted for between 30% and 55% of the variance. We have hypothesized that heritability49,51 may be among these factors.
Our specific recommendation for improving the correlations between clinical and quantitative measures is to test a different site than the wrist, given our low correlations between the ice pain test and CPT. Our data suggested that moderate pressure is most likely to be associated with PPT (given the overall higher correlations compared with light or firm pressure), but ideally quantitative standardization of the moderate clinical pressure applied may yield better correlations.
Clinical Implications and Recommendations
Our data suggest that the ice pain test may be a useful clinical correlate for CPT at the neck, upper trapezius, and tibialis anterior muscles, given correlations are moderate to good at all locations. The wrist as a test site should be reconsidered at this stage, given the fair correlations only. We would recommend the ice pain test above the manual pressure pain test, given the higher correlations and the higher prediction levels with quantitative measures. The clinical ice pain test could potentially enable the clinician to determine if pain sensitivity is local or widespread, as treatment recommendations may arguably differ depending on the pain sensitivity profile.
The application time for the ice pain test varied from 5 to 10 seconds between our study and that of Maxwell and Sterling.22 Given the results of both studies, it is not clear that there is a significant difference in utilizing either protocol clinically. Anecdotally, for people with high pain sensitivity, immediate ice application is associated with significant pain; therefore, sustaining the ice application may not be desirable for such people. Other observed responses during our testing included persistence and referral of pain after the removal of the ice in some patients, perhaps indicative of temporal summation or wind-up. Given this finding, we favor the 5-second application to minimize patient discomfort.
Finally, clinicians need to understand that sensitivity to pain is multidimensional and multifactorial. The simple clinical tests evaluated here test one stimulus at a time. How this information is processed and perceived depends on a multitude of factors mediated both peripherally and centrally. Our models suggest that clinicians also should assess factors such as pain catastrophization, depression, and sleep quality, as these factors potentially contribute to pain sensitivity.
Clinicians can be confident that the ice pain test is associated with laboratory measures of CPT when used at the neck, upper trapezius, and tibialis anterior muscles. Further development of the manual pressure pain test is recommended before clinicians use this test as a substitute for PPT. An alternate upper limb site to the wrist should be considered in future work investigating CPT and ice pain test correlations. Other contributions to quantitative measures of pain sensitivity in this study were pain catastrophization, sleep quality, and sex.
Footnotes
Dr Rebbeck, Dr Hübscher, Mr Waller, Dr Moloney, and Dr Beales provided concept/idea/research design. All authors provided writing. Dr Rebbeck, Dr Moloney, Ms Azoory, Mr Waller, Ms Gibbons, and Dr Beales provided data collection and study participants. Dr Moloney and Dr Hübscher provided data analysis. Dr Rebbeck and Dr Beales provided project management and institutional liaisons. Dr Rebbeck provided fund procurement. Dr Rebbeck, Dr Moloney, Mr Waller, and Dr Beales provided facilities/equipment. Dr Rebbeck, Dr Moloney, Ms Azoory, Mr Waller, and Miss Gibbons provided consultation (including review of manuscript before submission). The authors acknowledge the assistance of their honors students across the span of this work: Camilla Montague, Basil Chen, Alexandra Griffin, and Natalie Chui-Cheng-Lam.
Human ethics approval was obtained from the University of Sydney Human Ethics Committee (Protocol No. 14417) and the Curtin University Human Research Ethics Committee (Protocol No. PT0205).
Dr Rebbeck and Dr Beales are supported by a National Health and Medical Research Council Research Fellowship. Dr Hübscher was supported by a postdoctoral fellowship from the German Academic Exchange Service (DAAD).
- Received August 21, 2014.
- Accepted May 21, 2015.
- © 2015 American Physical Therapy Association