The Tampa Scale for Kinesiophobia is a measure commonly used to determine whether a person is afraid of moving because of beliefs about harm or damage, with a second scale assessing current avoidance behaviour. It has been a popular measure along with the pain-related fear and avoidance model and together with the model and measures of disability, catastrophising and pain-related anxiety, has become one of the mainstays within pain assessment.
There have been numerous questions raised about this measure in terms of reliability and validity, but the measure continues to be one that is widely used. The problems with reliability relate mainly to a long version (TSK-17) in which several items are reverse scored. Reverse scored items often state a negative version of one of the concepts being assessed by the measure, but pose problems to people completing the measure because it’s hard to respond to a double negative. In terms of validity, although the measure has been used a great deal and the original studies examining the psychometric properties of the instrument showed predictive validity, the TSK’s ability to predict response to treatment hasn’t been evaluated.
Chris Gregg and colleagues from The Back Institute and CBI Health Group studied a cohort of 313 people with low back pain attending one of the rehabilitation clinics in New Zealand. Participants completed the TSK at the beginning of treatment, and again at programme completion. Along with the TSK, participants also completed a numeric pain scale, a modified Low Back Outcome score, and indicated whether they were working or not. These latter measures were considered to be “Quality of Life” measures, although they’re not officially QoL scales.
Before I turn to the study design and statistics, I’ll take a look at the modified Low Back Outcome score. Now I don’t know if you’ve ever searched for something like this, but believe me when I say there are SO many versions of SO many different “modified” back pain questionnaires, it’s really hard to work out exactly which one is the one used in this study, nor how it was modified. I’m assuming that it’s the one mentioned in Holt, Shaw, Shetty and Greenough (2002) because it’s mentioned in the references, but I don’t know the modifications made to it. The LBOS is a fairly brief 12 item measure looking at pain intensity “on average” over the last week, work status, functional activities, rest, treatment seeking, analgesic use, and another five broad activities (sex life, sleeping, walking, traveling and dressing). It’s been described as having good internal consistency and test-retest reliability but validity isn’t mentioned in the 2002 paper.
Now, coming to this study, overall people improved at the completion of the programme. Pain reduced by 1.84 on the NPS, m-LBOS scores increased by 10.4 (a 28% improvement), and the TSK scores also improved by 5.5. Of course, we’d hope that at the end of a programme people would be doing better – though I’d prefer to see outcomes measured at least another three to 9 months after programme completion.
The authors looked at the relationship between the TSK and initial scores – there were small statistical relationships between these measures. They then examined the scores between pre-treatment TSK and QoL measures at the end of treatment to establish whether there was a relationship between kinesiophobia and eventual outcome. There wasn’t. At least, not much of a relationship. These authors conclude that the TSK is therefore not a good measure to employ to predict those at high risk of chronicity due to fear of movement. I was a bit disappointed to see that a subscale analysis of the TSK wasn’t carried out – so it’s not possible to know whether change was associated with reduced beliefs about fear of harm/reinjury or whether it was due to reduced avoidance, or both.
Now here’s where I get a bit tangled up. Wouldn’t you expect the underlying constructs of the TSK (fear of harm/reinjury, and avoidance) to be the targets of a back pain related treatment? Especially one that includes cognitive behavioural therapy, education and movement? If we’re using a measure I think we should USE it within our clinical reasoning, and deliberately target those factors thought to be associated with poor outcomes. If we’re successful, then we should be able to see a change in domains thought to be associated with those constructs. In this programme, given that people were given treatment based on sub-typing, including education and CBT, I would hope that pain-related fear and avoidance would be directly targeted so that people develop effective ways of dealing with unhelpful beliefs and behaviours. To establish whether that had happened I’d want to look at the association between post-treatment TSK and measures of function or disability.
And getting back to the timing of outcome assessment, given that we’re interested in people managing any residual back pain (and in this study people were left with pain scores on the NPS of 3.4 (+/- 2.4) they still had some pain), wouldn’t you be interested in how they were managing a bit further down the track? We can (almost) guarantee that people will make changes directly as an effect of attention and structured activities. Measuring what occurs immediately at the completion of a programme may not show us much about what happens once that person has carried on by him or herself for a few months. My experience with chronic pain programmes shows a typical pattern of improvement immediately at the end of a programme, then six weeks later, what can be called regression to the mean, or what we often described as “the dip” or “the slump” as reality hits the road. At a further six months down the track, results had improved a bit, and these were usually sustained (or thereabouts) at the following twelve month follow-up.
So, does this study provide us with evidence that the TSK isn’t useful as a predictive tool? I’m not so sure. I think it does show that there are improvements in TSK, pain, disability and work status immediately at the end of a programme. Unfortunately TSK scores at the end of the programme are not analysed into subscales, so we don’t know which aspects of pain-related fear and avoidance were affected – but we know that they were.
For clinicians working in chronic pain programmes, where people are referred after having remained disabled and/or distressed despite having had prior treatment, the TSK may not be the most useful tool ever. The problems I’ve had with it are that scores in the fear of injury/reinjury subscale are lower when people have been given good pain “education” – but often present with a combined high score because of very high scores on the avoidance subscale.
A lovely study by Bunzli, Smith, Watkins, Schütze and O’Sullivan (2014) looked at what people actually believe about their pain and the associated TSK items. They found that many people DO believe their pain indicates harm, and they also found that people were worried about the effect pain would have on other things – and it’s this part that I find particularly interesting. It may not be the pain that matters as much as the anticipated losses and disruption to normal life that could occur.
The original authors of the “fear-avoidance” model, Vlaeyen and Linto (2012) reviewed the model after 12 years, and agree there is much to be done to refine assessment of pain-related fear. Self-report measures are only as good as the ability, insight and willingness of participants to complete them accurately.
So, is it time to throw the TSK out the window? I don’t think so – at least not yet. There’s more we need to do to understand pain-related fear and subsequent avoidance.
Chris D. Gregg, Greg McIntosh, Hamilton Hall, Heather Watson, David Williams, Chris Hoffman, The relationship between the tampa scale of kinesiophobia and low back pain rehabilitation outcomes, The Spine Journal (2015), http://dx.doi.org/doi:10.1016/j.spinee.2015.08.018.
Bunzli, S., Smith, A., Watkins, R., Schütze, R., & O’Sullivan, P. (2014). “What Do People who Score Highly on the Tampa Scale of Kinesiophobia Really Believe? A Mixed Methods Investigation in People with Chronic Non Specific Low Back Pain The Clinical Journal of Pain DOI: 10.1097/AJP.0000000000000143
Vlaeyen, J. W., & Linton, S. J. (2012). Fear-avoidance model of chronic musculoskeletal pain: 12 years on. Pain, 153(6), 1144-1147. doi: dx.doi.org/10.1016/j.pain.2011.12.009
HealthSkills Blog 
“Fear-avoidance” as a scientific model of chronic musculoskeletal pain has never made any sense to me. But at least it does seem to satisfy the academic psychologists and enable them to publish increasingly esoteric papers on the subject! I suspect that understanding threat detection and avoidance behaviours might provide us with a more realistic and common-sense approach to people in pain.
I’m not so sure John. I think the model is both practical and has opened up a great many opportunities to learn and treat/help people with high disability but relatively low levels of pain. I’ve certainly used it in practical terms over and again in my work with people living with pain. The TSK may not be an especially accurate measure of fear of pain or avoidance, there are others that I think work more effectively, but it does require some work to interpret well – which is what I think these researchers have not done. One of the problems with psychometric assessment is that it does require some understanding of how it was developed, what the subscales mean, and how it might change over time. I personally think it more useful to try to understand fear of pain, and then to look at avoidance – threat detection may be a component of that, but conditioned stimuli also produce fear even though they’re not directly relevant to threat. To me, it’s the avoidance that makes the major impact on disability, not the fear part.
Bronnie, how does one reach the conclusion that the pain of another person is at a relatively low level?
From their self-report.
Bronnie, then there may be a problem with the person’s threat detection apparatus such that it cannot turn itself off when no longer required.
And that’s OK – I think it’s one of the many underlying mechanisms of the experience of chronic pain. That’s in part why cognitive behavioural approaches can help, because cognitions and behaviours can be used to down-regulate the systems that are overdoing threat detection. Some of the work that can then be undertaken is to counter the avoidance component – if I experience pain I can choose to follow what my cognitions and emotions suggest – or act according to my values. I choose personally to work with my values because this leads to a more fulfilling life, despite experiencing ongoing pain.
Bronnie, I suppose this approach is the best we can offer under the circumstances. Apart from stress response systems being activated, so we have any idea of the other underlying biological mechanisms that might be in play? If so, what might they be?
To be honest, no I don’t. Immune responses seem to be implicated in some pain problems (and arguably in some depression as well). Attention, learning (through conditioned responses) – but I think the concept of threat detection implies all of these mechanisms as well as contextual cues. Fruitful areas for research!
Bronnie, one way out of this conceptual puzzle is to accept the proposition that “pain” can be but one component of a package that characterizes undampened activation of systems of stress response. Other components might be mood changes, sleep disturbance, fatigue, cognitive disturbances, dermatographia, etc. A close reading of our hypothesis (Lyon et al. 2011) will flesh out this hypothesis, which is firmly based upon evolutionary biological theory.