This level of significance was chosen to decrease the likelihood of overlooking potential prognostic factors. Where there was a moderate or strong correlation (Pearson’s r > 0.4) between individual predictor variables, the variable with the best psychometric properties or ease of clinical application was selected.

The selected predictor variables were assessed using multivariate stepwise regression to identify the independent prognostic variables. One hundred and eighty-one participants were recruited between October Baf-A1 2006 and June 2008 from 11 primary care clinics in Sydney, Australia. Seven physiotherapists recruited 125 participants and five chiropractors recruited 56 participants. Of the 237 patients screened, 46 did not meet the eligibility criteria and 10 declined to participate. Three participants did not complete the course of four treatments. All participants completed baseline assessments with no missing data. Five participants withdrew from the study and were censored at the last date of data collection. Completeness of follow-up (Clark et al 2002)

was 96% of potential person-time for the time-to-recovery predictive model. Data were included from 176 (97%) participants for the predictive model for disability at 3 months. The baseline demographic and clinical characteristics of the participants are presented in Table 1. The mean age of participants was 38.8 (SD 10.7) years. Pain intensity at baseline was 6.1 (SD 2.0) with the average duration of neck pain 19.5 Everolimus ic50 (SD 20.1) days. The mean disability score was 15.7 (SD 7.4). Neck pain was frequently through accompanied by concomitant symptoms, most commonly upper limb pain (n = 144, 80%), headache (n = 117, 65%) and upper back pain (n = 115, 64%). One-hundred and fourteen participants (63%) had a past history of neck pain. Ninety percent of participants rated their general health as ‘good’ or better, and fewer than 10% were smokers. SF-12 Physical Component Score 43.5 (SD 8.2) and

Mental Component Scores 47.3 (SD 10.6) were less than one standard deviation from normal population values. Ninety-five participants (52%) experienced full recovery from neck pain during the 3-month follow-up period. The median time from commencement of treatment to recovery of pain was 45 days. Of those who recovered, 52 (55%) recovered within 3 weeks and 71 (75%) recovered within 4 weeks of commencing treatment (Figure 1A). The mean pain score for all participants decreased from 6.1 (SD 2.0) at baseline to 2.5 (SD 2.1) after 2 weeks of treatment, and to 1.5 (SD 1.8) at 3-month follow-up (Figure 2). Neck pain intensity in those participants who remained symptomatic (ie, excluding those who had recovered) showed rapid improvement with a mean pain score of 3.1 (SD 1.9) at 2 weeks (n = 143) and a mean pain score of 2.8 (SD 1.6) at 12 weeks (n = 77). The distribution of pain scores at the 3-month follow-up was skewed, with 153 (86%) participants rating residual pain as ≤3 out of 10 (Figure 3).