Previous results

Follow up was measured at 3, 6 and 12 months and outcomes were summarised as the change from baseline score. At 12 months, the mean change from baseline in the RMQ score was 1.1 (95% CI 0.39 to 1.72) in the control (AM) group and 2.4 (95% CI 1.89 to 2.84) in the AM+CBA group. The dierence of 1.3 (95% CI 0.56 to 2.06) is statistically signicant (p= 0.01), and can be delivered at low cost to the

provider. 1,465 people of the 9,771 initially identied appeared to be eligible, and after excluding those who no longer had back pain, or who had only infrequent pain, or pain that was not troublesome, or were pregnant, 705 were randomised of whom four did not provide baseline data, leaving 701 for the trial analysis. Of these, 70% experienced pain every day, 55% had moderately troublesome pain and the remain- der had very or extremely troublesome pain.

The average age of the trial participants was 54 years, 60% were women, most had left full-time education before age 15, and half were currently working, mostly full-time; most of those not working had retired. Stratied block randomisation was carried out by an independently administered telephone randomisation service at the MRC clinical trials unit in London.

There was no statistically signicant dierence in the follow-up questionnaire return rates between the two arms of the trial at any of the observation points nor any evidence of a systematic dierence in the baseline characteristics of participants who provided follow-up data and those who did not. Attendance at all six of the cognitive behavioural sessions was achieved by 25% of the participants allocated to the AM+CBA arm, 63% attended at least three sessions and were considered to have received the basic elements (adhered), and just over 10% did not attend any. There were no dierences between the patients who were adherent and those who were not in any of RMQ score, troublesomeness of pain, fear avoidance beliefs, sex or Modied Von Kor disability at baseline. Adherent patients had slightly lower MVK pain scores at baseline (mean dierence 4.5, 95% CI 0.95 to 8.19) and were older (average 4.5 years older, 95% CI 1.8 to 7.5). There were 62 groups run with mean size of 8 (SD 1.62, range 4-12) and a range of times were oered to meet the needs of participants. There was no evidence of either group or therapist eects.

Improvements in the AM only arm were, on average, 1.1 RMQ points, with change occurring between baseline and three months and no further improvement thereafter. The change in the CBA arm was almost double by three months, and the treatment dierence continued to widen at six and twelve month follow-up points. The mean treatment dierence was 1.1 Roland Morris Questionnaire points at three months, 1.4 at six months and 1.3 at twelve months, and all were statistically sig- nicant.

The MVK disability scale showed improvement over both arms of the trial, with the improvement in the active management arm occurring between baseline and three months and declining thereafter, and the cognitive behavioural approach arm showing a greater improvement. The mean dierence between the two arms was 4.3% at three months, 8.1% at six months, and 8.4% at twelve months. The MVK pain scale also showed improvement over both arms, with the active manage- ment arm improving gradually over the twelve months, and greater improvements

in the cognitive behavioural approach, particularly at three months. The dierence between the two arms was 6.8% at three months, 8% at six months and 7% at twelve months, and all were statistically signicant.

The SF12 physical subscale showed improvements in both arms of the trial in the rst six months, but greater improvement in the cognitive behavioural approach arm, and in the active management arm there was no dierence from baseline by twelve months. The mean dierence between the two arms was 2.2 at three months, 1.8 at six months, and 4.1 at twelve months, all statistically signicant. The SF12 mental subscale showed no signicant improvement or dierences between the two arms at twelve months. There were statistically signicant improvements in the cognitive behavioural arm at six months, but by twelve months the improvement disappeared. Fear avoidance beliefs did not change in the active management arm but substantial improvements occurred in the CBA arm between 0 and three months and were maintained to twelve months. The mean dierence at three months was 2.6, 3.1 at six months and 3.0 at twelve months, and the dierences between treat- ments were signicant at all time points.

Pain self-ecacy improved in the cognitive behavioural arm, with peak im- provement at six months maintained at twelve months. In the active management arm, there were no discernible changes in pain self-ecacy, the mean dierence at three months was 3.2, 4.1 at six months and 3.8 at twelve months.

Distributions of patient scores is given in Table 3.1. Subgroup analysis re- vealed that fear avoidance beliefs at baseline was not associated with treatment eects measured by the RMQ, but there was a statistically signicant interaction between baseline fear avoidance and outcomes measured by MVK for disability, with those not fear avoidant at baseline having a larger treatment eect. However these observations were not consistent across all primary outcomes. The treatment eect was larger for people with moderately troublesome back pain as opposed to severe pain. Active management had little or no eect for these patients which accounts for the dierence, and the interaction was not signicant. There was no dierence in the results between the observed case analysis and an analysis based on multiple imputation, and the ndings appeared insensitive to the method of dealing with missing data.

Variable Male Female Total

Number 161 (39.6%) 246 (60.4%) 407 (100%)

Mean age in years (sd) 54.2 (14.7) 53.9 (14.3) 54.0 (14.0)

N % N % N %

Age group under 40 years 31 11.9 44 17.9 75 18.4

Age group 40-49 years 34 30.7 54 21.9 88 21.6

Age group 50-59 years 35 30.7 60 24.4 95 23.4

Age group 60-69 years 45 37.9 58 23.6 103 25.3

Age group 70 years and over 16 19.5 30 12.2 46 11.3

Treatment allocation

Active management only 51 31.7 75 30.5 126 31.0

Active management and CBA 110 68.3 171 69.5 281 69.0

RMQ improvement of 3+ points Achieved 64 39.8 117 47.6 181 44.5 Not achieved 97 60.2 129 52.4 226 55.5 Anxiety Not anxious 91 56.5 86 35.0 177 43.5 Borderline 43 26.7 80 32.5 123 30.2 Anxious 27 16.8 80 32.5 107 26.3 Depression Not depressed 113 70.2 144 58.5 257 63.1 Borderline 40 24.8 75 30.5 115 28.3 Depressed 8 5.0 27 11.0 35 8.6

Pain self ecacy

Very low 6 3.7 20 8.1 26 6.4 Low 25 15.5 36 14.6 62 15.0 Moderate or better 30 80.8 190 77.3 320 78.6 Troublesomness Somewhat 2 1.3 1 0.4 3 0.7 Moderately 82 50.9 113 45.9 195 47.9 Very 57 35.4 102 41.5 159 39.1 Extremely 20 12.4 30 12.2 50 12.3 interference with social activities

All of the time 3 1.9 6 2.5 9 2.2

Most of the time 15 9.3 30 12.2 45 11.0

Some of the time 50 31.1 81 32.9 131 32.2

A little of the time 30 18.6 46 18.7 76 18.7

None of the time 63 39.1 83 33.7 146 35.9

Table 3.1: The numbers and proportions of patients in each category of each score in the BeST trial.