Strengths and limitations of the study
There are a number of methodological aspects to consider. Firstly, baseline participation was low and it is important, therefore, to assess whether selection bias, caused by selective participation and/or follow-up, may have influenced results. Of 1917 individuals invited to participate in the study, 58% agreed to participate, although only 88% of these subsequently returned a completed full-length baseline questionnaire. By definition, little data are available on non-participants. However, one can calculate that if all non-participants were pain-free at 3-month follow-up the prevalence of persistent disabling low back pain would still be around 20%.
Although participants and non-participants differed by age and sex, a weighted analysis, weighting the population back to the age/sex distribution of the target population, revealed similar results to the main findings. This would suggest that the demographic differences between participants and non-participants (insofar as we are able to examine them using available data) have not introduced a bias to the current study. Further, follow-up participation was high (95%) and it is unlikely, therefore, that bias was introduced through loss to follow-up.
Secondly, although we purport to have measured persistent low back pain, our exposures and outcome were measured in two consecutive cross-sectional surveys. Not having information about pain status between the two time points, it is possible that in some individuals we have identified recurrent low back pain, rather than persistent symptoms. Notwithstanding the fact that this may also be a valid outcome in general practice, this would serve to overestimate the prevalence of persistent low back pain. A corollary of this, however, is that the true associations between passive coping (and other risk factors) and outcome will actually be greater than we report here.
Thirdly, it seems somewhat counterintuitive that active and passive strategies do not elicit contrary responses: persons who reported high levels of passive coping behaviour experienced a significant increase in the risk of low back pain at follow-up, whereas those who adopted active coping strategies did not experience a decrease in risk. Active coping in the current study was defined as positive responses to statements such as, ‘I am able to ignore my pain’ and ‘I participate in all my usual leisure activities’; whereas a high passive coping score required positive responses to statements such as, ‘I wish my doctor would prescribe better pain medication’ and ‘I restrict or cancel my social activities’. However, rather than showing a strong inverse correlation between active and passive coping strategies, the current results show only a poor correlation (r = −0.13). This slight inverse correlation, confirming previous findings,19 goes some way to explaining the specificity of the results for passive coping. Furthermore, it could be argued that persons who consult their GP are already exhibiting characteristics of passive coping behaviour and, therefore, the range of coping strategies in our cohort may be restricted. However, this would make it more difficult to detect any association between passive coping strategy and low back pain. Despite this, individuals in the highest quartile of passive coping score were more than three times more likely to report a poor outcome at 3-month follow-up than those in the lowest quartile. This would suggest that across the full range of coping strategies the effect might be even greater.
Fourthly, a number of factors previously shown to be important in the epidemiology of low back pain are missing from our model. Previous studies have demonstrated that psychological distress and work-related psychosocial factors are important predictors of low back pain outcome,6 and these factors may influence coping strategies. However, our multivariable model performs well and among persons with all seven factors in the final model (that is, excluding age, sex and socioeconomic status), 87% report low back pain at follow-up. We suggest that the addition of other variables, such as psychological distress, would actually add little to the overall model.
Finally, some have suggested that a high passive coping score is only a proxy for severe baseline pain.22 Our results only partly support this theory — the risk ratio associated with a high passive coping score drops from 3.0 (2.3 to 4.0) to 1.5 (1.1 to 2.0) on adjustment for pain severity, disability and other baseline variables. However, passive coping remains independently and significantly predictive in the final model, suggesting that it is an independent construct and can precede chronic pain symptoms.
Comparison with existing literature
We have shown that, in a population of adults with a new consulting episode of low back pain, 39% report persistent disabling low back pain 3 months subsequently. This is in contrast to the common preconception that only a small proportion of patients experience chronic disabling symptoms. However, other authors have demonstrated similar findings: Croft et al showed that 50% of patients reported pain and disability 3 months after their initial consultation.4 Further, these authors also demonstrated that the same proportion were reporting symptoms 12 months post-consultation, although the majority will have stopped consulting with this pain.
Some authors have demonstrated the ability to identify patients at risk of poor work and functional outcome using a combination of pain intensity, fear avoidance, function and mood, but have been unable to predict persistent pain.23 Others have shown that psychological factors, in particular distress and somatisation, are strong predictors of chronic low back pain,24 and there is an emerging body of evidence to support the involvement of various aspects of coping behaviours in the epidemiology of low back pain, to which this study adds.
Our results are consistent with recent findings of Smith et al who reported that at 4-year follow-up, in addition to clinical factors, health-seeking behaviour was significantly associated with the onset, and persistence, of chronic back pain in the community.13 A smaller study, in general practice, demonstrated that initial pain intensity, a prior pain history, and low levels of active coping were predictive of low back pain outcome 3 months after consultation,14 and Burton et al report that aspects of coping strategy are important predictors of low back pain in acute patients (pain ≤3 weeks) but not those with subacute symptoms (>3 and <52 weeks).25 In addition, other authors have shown the importance of strategies for coping with pain in the prognosis of other musculoskeletal conditions: whiplash disorder,26,27 and rheumatoid- and osteo-arthritis.28,29
To focus on coping strategies, rather than other psychological and psychosocial indicators is important, as patients' coping strategies may be modified through cognitive-behavioural therapy.30 Several studies have demonstrated that intensive cognitive-behavioural treatments can be effective in improving pain, coping and disability in chronic low back pain patients,15–17 and others have shown that in patients with subacute low back pain a cognitive-behavioural therapy-based educational programme results in demonstrable improvements in pain and disability at 3- and 6-month follow-up.31 Other authors have shown that exercise programmes using cognitive-behavioural therapy principles can be effective at reducing pain and disability for up to a year,11 and these authors subsequently demonstrated that cognitive-behavioural therapy-based exercise programmes were most effective in patients with high fear avoidance behaviour.32 Although few GPs have quick access to individuals trained in cognitive-behavioural therapy, they may still provide education reassurance and advice, encouraging early mobilisation and recommending light activity.
Implications for clinical practice and future research
In summary, this is one of the first studies to examine, prospectively, the role of coping strategies in the prognosis of low back pain in primary care. While not dismissing the notion that chronic pain may lead to an increase in passive coping, our findings demonstrate that such coping strategies can also precede the occurrence of persistent disabling low back pain and that patients who report high passive coping strategies experience a significant increase in the risk of persistent symptoms. Further, although somewhat attenuated, this risk remains after adjusting for baseline levels of pain severity, disability and other measures of pain and pain history. We propose that the identification of such an ‘at risk’ group may help inform future treatment and management decisions — in particular, in targeting treatments that have been shown to be effective, but resource intensive, to the subgroup of patients who are at highest risk of a poor outcome.