Introduction. The STarT Back trial demonstrated that targeting back pain treatment according to patient prognosis (low, medium or high-risk subgroups) is effective. However, the mechanisms leading to these improved treatment outcomes remain unknown. This study aimed to identify which psychological variables included in the study were mediating treatment outcome for all patients and within the low, medium and high-risk subgroups. Methods. Secondary analysis was conducted on 466 patients randomised to the active treatment arm with 4-month follow-up available. Psychological variables included depression (HADs), fear (TSK), catastrophising (PCS), bothersomeness and illness perception constructs (IPQ brief) e.g. personal control. Treatment outcome was characterised using change in disability score (RMDQ) at 4-months. Residualised change scores were calculated for each variable and Pearson's correlations were calculated overall and at the subgroup level to determine potential mediating variables for disability improvement. Results. Overall, correlations with RMDQ change were .62 for change in bothersomeness, .56 for change in catastrophising, .51 for change in fear, .48 for change in anxiety, .58 for change in depression, −.32 for change in personal control and .40 for change in symptom identity. The strength of correlation generally increased from low to high-risk subgroups, e.g. bothersomeness (low=.54, high=.70). However, the predominant variables mediating treatment outcome were common across risk-groups. Conclusion. The psychological variables which were highly correlated with improvements in disability were bothersomeness, depression and catastrophising. This finding was consistent across low, medium and high-risk subgroups. This study is ongoing and further mediation analysis using structural equation modelling is in progress. Conflicts of Interest. None. Source of Funding. NIHR Spinal Pain Programme grant

Background. A wide array of measures has been developed to assess the role of psychological factors in the development and persistence of pain. Yet there is likely to be considerable conceptual overlap between such measures, and consequently a lack of clarity about the importance of psychological factors. Purpose. To investigate whether conceptual overlap exists within psychological measures used in back pain research. Methods. An observational cohort study of 1591 back pain primary care consulters provided data on anxiety, depression, self-efficacy, coping, illness perceptions, fear avoidance and catastrophising. Exploratory and confirmatory factor analysis (EFA and CFA) was carried out at the sub-scale level (n=20) to investigate factor structure. Derived factors were then tested using hierarchical linear regression in relation to clinical outcomes (pain intensity and disability). Results. EFA derived 4 factors from the sub-scale measures. Factors were termed ‘distress’ (comprising depression, anxiety, pain self-efficacy, fear avoidance, catastrophising), ‘cognitive’ (cognitive coping strategies), ‘causes’ (psychological, immunity, risk factors) and ‘future’ (expectancies of potential treatment and recovery), and accounted for 65.5% of the data. CFA confirmed the validity of these factors. The distress factor was found to have the strongest association with patients' outcomes, accounting for 28.7% of the variance in pain intensity, and 53.6% in disability. Conclusion. Considerable overlap exists in psychological measures used in back pain research. Most measures tap in to patients' emotional distress; this factor was shown to have the greatest association with clinical outcomes. These findings have implications for the way psychological measures are used in back pain research. No conflicts of interest. Source of funding: Programme Grant awarded by the Arthritis Research UK [13413] and Dr KM Dunn's Wellcome Trust Grant [083572]. This abstract has not been previously published in whole or substantial part nor has it been presented previously at a national meeting