Classification and targeted treatment of patients with Non Specific Chronic Low Back Pain

Non-specific chronic low back pain (NSCLBP) disorders have proven highly resistant to change in spite of enormous resources directed at them. There is lack of evidence for single treatment interventions for patients with NSCLBP despite the substantial amount of Randomised Controlled Trials (RCT) evaluating treatment outcome for this disorder. It has been hypothesised that this vacuum of evidence is caused by the lack of sub-classifying the heterogeneous population of patients with chronic LBP for outcome research. Another reason suggested for the limited evidence is the lack of sub-grouping and managing the disorder from a biopsychosocial perspective. There is growing evidence that NSCLBP is associated with maladaptive cognitive, movement and lifestyle behaviours that act to promote a vicious cycle of pain. Few classification systems reflecting a bio-psycho-social model have been validated and tested in RCTs for the management of NSCLBP disorders. The O’Sullivan Classification System (OCS) has been developed and validated over many years and subgroups patients based on their maladaptive cognitive, movement and lifestyle behaviours. This thesis consists of three papers and the work for these papers was performed from 2005 till 2009. In paper I a systematic review with a metaanalysis was undertaken to determine how the integration of sub-classification strategies with matched interventions in RCTs evaluating manual therapy treatment and exercise therapy for NSCLBP was performed in the literature. A structured search for relevant studies in Embase, Cinahl, Medline, PEDro and the Cochrane Trials Register database, was followed by a hand search of all relevant studies in English up till December 2008. A large number of RCTs (n= 767) was retrieved, and 68 of these focused on the efficacy of manual therapy and exercise and were looked at in more detail. However, only five out of 68 studies (7.4%) sub-classified patients beyond applying general inclusion and exclusion criteria. In the few studies where classification and matched interventions have been utilised, our meta-analysis showed a statistical difference in favour of the classification based intervention for reductions in pain (P=0.004) and disability (P=0.0005), both for short and long-term reduction in pain (P=0.001). Effect sizes ranged from moderate (0.43) short term, to minimal (0.14) for long-term. However, the data should be interpreted with caution, as the numbers are insufficient to definitively quantify the effect of sub-classification strategies for the treatment of NSCLBP. Also, out of the five papers using a classification system in the meta-analysis only three considered all dimensions of LBP within a biopsychosocial perspective. We concluded that a better integration of sub-classification strategies in NSCLBP outcome research was needed. This led us in paper II to examine the inter-tester reliability of clinician’s ability to independently classify patients with non-specific low back pain, utilizing the mechanism-based OCS. Here 26 patients underwent a full examination by four different physiotherapists (O’Sullivan and three others) independently. The therapists underwent a multilevel decision making process, based on disorder classification, primary directional pain provocation and the detection of dominant psychosocial factors. Percentage agreement and Kappacoefficients were calculated for six different levels of decision-making. For levels 1-4, percentage agreement had a mean of 96% (range 75-100%). In the 5th level, deciding the directional pattern of provocation, Kappa agreement could be calculated. For the primary direction of provocation, Kappa and percentage agreement had a mean between the four testers of 0.82 (range 0.66- 0.90) and 86% (range 73-92%) respectively. Increased familiarity with the system increased the reliability scores. In the final decision making level, the scores for detecting psychosocial influence gave a mean Kappa-coefficient of 0.65 (range 0.57-0.74) and a mean agreement of 87% (range 85-92%). Our findings confirmed some of the previous work on the classifications system, suggesting that the inter-tester reliability of the system is moderate to substantial for a range of patients within the NSLBP population. These findings were considered an important step towards implementing targeted interventions programs for subgroups with NSLBP. In paper III we performed a RCT to investigate the efficacy of the intervention called ‘cognitive functional therapy’ (CFT) utilised in the OCS. The intervention aims to address the behaviours often seen in NSCLBP in a targeted, functionally specific and patient focused manner. The RCT was a two-armed study comparing classification based CFT (CB-CFT) with patients receiving traditional manual therapy and exercise (MT-EX). 121 patients with mechanically provoked NSCLBP (>52 weeks) were randomized to either CBCFT (n= 62) or MT-EX (n= 59). A linear mixed model was used to estimate the group differences in treatment effect and also in the change in outcome from 3 and 12-month follow-up. Primary outcomes were change in Oswestry Disability Index (ODI) score and pain intensity measured with numerical rating scale (PINRS) at 12 months follow-up. Secondary outcomes were the Fear Avoidance Behaviour Questionnaire, the Orebro multidimensional questionnaire, the Hopkins symptoms check list and time off work due to their disorder. After adjustment for baseline scores, the CB-CFT group displayed superior outcomes supported by both statistically and clinically significant differences, compared to the MT-EX group. The degree of improvement in the CB-CFT group for ODI score was 13.7 points from baseline (95% CI, 11.4 to 16.1, P<0.001) and for PINRS scores 3.2 (95% CI, 2.5 to 3.9, P<0.001). In the MT-EX group, the improvement for ODI score was 5.5 points (95% CI, 2.8 to 8.3, P<0.001) and 1.5 for PINRS (95% CI, 0.7 to 2.2, P