Results
A total of 723 patients were eligible for the study and 413 (57%) consented to participate. Six patients in the work-focused intervention and 2 patients in the control intervention were incorrectly randomised (Figure 1). 45 patients admitted with neck pain were included, and these were evenly distributed between the work-focused and control group. Two patients dropped out of the control intervention, but none from the work-focused intervention. Compliance was defined as accomplishing at least 50% of the treatment and 2 patients in the control intervention and 6 in the work-focused interventions were non-compliant (Figure 1).
(Enlarge Image)
Figure 1.
Flow chart.
No significant differences were found in the baseline characteristics between the participants in the work-focused intervention and the control intervention (Table 2).
Patients lost to follow-up at 12 months had higher baseline disability scores (mean difference 3.60, p = 0.018), reported higher baseline pain (mean difference 0.52, p = 0.039) and higher baseline FABQ-P scores (mean difference 1.57, p =0.015). There were also a significantly higher number of men, smokers, patients with a foreign mother tongue, and patients with low education in patients lost to follow-up. The response rate was 74 % at the 12-month follow-up. There were a similar number of patients lost to follow-up in both groups (Figure 1). Concurrent treatment (e.g. physiotherapy, manual therapy, acupuncture) was reported by 61% of patients after 4 months and 66% of patients at 12 months, there were no differences in the rate of concurrent treatment between patients in the work-focused intervention and control intervention (Chi-square 0.36, p = 0.551 at 4 months, and Chi-square 0.03, p = 0.858 at 12 months).
Only 180 (60%) of the patients who completed 12 month follow-up had complete FABQ scores at both baseline and 4 months, and some had only completed one of the subscales (10 missing on FABQ-P subscale and 11 missing on FABQ-W subscale) (Figure 1). Subjects attending a 12-month follow-up with missing FABQ at 4 months did not differ from the subjects with complete FABQ regarding any baseline characteristics except for a higher number of males (66% and 39% respectively, p < 0.001) and blue collar workers (44% and 32% respectively, p = 0.027).
Change in Pain and Disability at 12-month Follow-up
The mean reduction in pain was 1.59 (SD 2.70) points on NRS in the work-focused intervention and 1.36 (SD 2.88) in the control intervention. For disability, the reduction in ODI/NDI was 8.80 (SD 15.55) in the work-focused intervention and 9.02 (SD 14.67) in the control intervention. The differences in change between the two groups were not statistically significant (Table 3). Analyses with only patients who had complete scores at both measure points (casewise deletion of missing) did not change these results (Table 3).
Association Between Improvement in Fear-avoidance and Decreased Pain, Disability and Return to Work at 12-month Follow-up
FABQ-P and FABQ-W scores decreased in both groups after intervention (4-month follow-up). Improvement in FABQ-P scores after intervention were achieved in 22% (N = 20) of patients in the work-focused intervention and 18% (N = 14) of patients in the control intervention. Improvement in FABQ-W scores were achieved in 26% (N = 24) of patients in the work-focused intervention and 20% (N = 15) in the control intervention. The differences between the groups were not statistically significant (Chi Square; p = 0.362 for FABQ-W, and p = 0.569 for FABQ-P).
Univariate logistic regressions with RTW, pain and disability as response variables were performed and are shown in Table 4. Subsequent multiple logistic regression models including variables with p < 0.2 from the univariate analysis were calculated. The results from the multiple regressions are shown in Table 5, Table 6 and Table 7. All the logistic regression models had acceptable goodness of fit (p-values ranged from 0.118 to 0.952), and none of the predictor variables had a Spearman's rho above 0.7.
Age and improvement in FABQ-P and FABQ-W scores at 4 months were identified as possible predictors for reduction in pain scores at 12 month follow-up in the univariate analysis. None of these remained significant in the multiple regressions models (Table 5).
Age, mother tongue, and improvement in FABQ-P and FABQ-W scores at 4 months, were identified as possible predictors for reduced disability (p <0.2). In the multiple regression analyses, younger age and improvement in FABQ-P remained positive predictors for improvement in disability (Table 6). Controlling for FABQ-P score at baseline lowered the OR for FABQ-P to 2.7, p = 0.056.
From the univariate analyses with RTW as response variable, age, anxiety score (HADS-A), improvement in FABQ-P and FABQ-W scores at 4 months were possible predictors (p < 0.2). Younger age, low anxiety score and improvement in FABQ-W remained positive predictors in the multiple regression analyses (Table 7). Controlling for baseline values of FABQ-W did not change this result, and the OR for RTW increased to 4.0 (p = 0.015) for the group with improvement in FABQ-W scores.