This is consistent with the current view that neck pain is an episodic condition that features intermittent periods of exacerbation and remission (Guzman et al 2008, Vos et al 2008). Because we used different definitions of recovery and recurrence as well as follow-up points that were different from previous studies, direct comparison of recurrence rates with previously described populations is not possible. Consistent with other

studies (Hendriks et al 2005, Hoving et al 2004), the disability measure at baseline was predictive of the disability score at 12 weeks. We did not however, find an association between baseline pain severity and time to recovery. An association between more SB203580 cost severe baseline pain and poor prognosis has been demonstrated in cohorts with predominantly chronic neck pain (Bot et al 2005, Hoving et al 2004). This suggests that unlike chronic neck pain, an acute episode EGFR signaling pathway (although initially a source of quite severe pain) is likely to resolve rapidly with a short course of treatment. This information might assist in alleviating anxiety and distress in those with severe baseline symptoms. Concomitant symptoms at baseline were prevalent among people seeking manual therapy care and some of these symptoms were predictive

of persisting pain and disability. Our results indicate that the absence of headache and upper back pain were features associated with faster recovery. Conversely, the presence of upper back pain or lower back pain was associated with persisting pain and activity limitation at 3 months. The divergent course of neck pain, depending on the presence or absence of concomitant symptoms, suggests that there is some validity in classifying neck pain syndromes according to symptom distribution. Just as these results demonstrate differing prognoses, it is plausible that subgroups based on distribution of concomitant symptoms might have different aetiologies. These subgroups might also differ with respect to the extent of pathophysiological and changes and thus might require

different treatment strategies. Consistent with previous studies, better prognosis was predicted by better self-rated general health and shorter duration of symptoms (Bot et al 2005, Hurwitz et al 2006). Also consistent with previous studies, factors that predicted persisting pain and activity limitation at 3 months included age (Hill et al 2004) and a past history of sick leave (Bot et al 2005, Hill et al 2004). Inexplicably, we found that being a smoker was strongly associated with a more rapid recovery. Given the known adverse health consequences of tobacco smoking (Vineis 2008), it is difficult to imagine the high rate of recovery in the 9% of smokers in this cohort being causally related to smoking.