More information about text formats
We congratulate O’Keeffe et al.  for their research on the comparative efficacy of Cognitive Functional Therapy (CFT) and physiotherapist-delivered group-based exercise and education for individuals with chronic low back pain (CLBP). Their study shows that “CFT can reduce disability, but not pain, at 6 months compared with the group-based exercise and education intervention”. The CFT approach is very promising and has caught the attention and interest of a number of clinicians worldwide in the management of non‐specific disabling CLBP. The study by O’Keeffe et al.  has methodological strengths compared to a previous clinical trial by Vibe Fersum et al. [2,3] such as a higher sample size which means it is less vulnerable to type-II error. Nonetheless, some shortcomings threaten substantially the risk of bias and type I error that are worthy of further discussion.
The first is the choice of three physiotherapists for delivering both interventions in this trial. This aspect was considered by O’Keeffe et al.  as a strength of the study because it arguably minimized differences in clinicians’ expertise and communication style. Notwithstanding, this fact could also have decreased the treatment effect on the control group. It is important to remember that the trial was performed by the research group that not only developed CFT but also has trained the physiotherapists on such an approach, and thus the enthusiasm and motivation to apply the intervention on the CFT...
The first is the choice of three physiotherapists for delivering both interventions in this trial. This aspect was considered by O’Keeffe et al.  as a strength of the study because it arguably minimized differences in clinicians’ expertise and communication style. Notwithstanding, this fact could also have decreased the treatment effect on the control group. It is important to remember that the trial was performed by the research group that not only developed CFT but also has trained the physiotherapists on such an approach, and thus the enthusiasm and motivation to apply the intervention on the CFT group could have been considerably greater. Likewise, if the physiotherapists providing the group-based exercise were not involved in any kind of CFT training and had a strong belief in their intervention, one can argue that the performance of the latter group could have been better. Also, the CFT group received the intervention for an average (SD) of 13.7 (10.9) weeks, while the comparison group received treatment for just 4.4 (2.4) weeks. The same therapist applying both interventions combined with the longer period of the CFT group treatment may have generated performance bias.
The second is the unblinded assessment of the outcomes immediately postintervention. Since CFT therapists had more one-to-one time with each patient and therefore had more opportunities to enhance the therapeutic alliance, this could have influenced patients’ behavior when filling the outcome assessment questionnaires in their presence, arguably not only in the postintervention assessment but also in the blinded follow up periods at 6 and 12-months.
Third, apparently, the authors were not concerned with the multiple primary outcomes (n = 2; pain and disability) and endpoints (n = 4; at 8-14 weeks; 6, 12 and 36 months). The problem of multiple testing is that the overall type-I (false positive) error is much greater than 5%. Assuming a true null hypothesis for all the 6 (independent) effects being tested, the probability that no false positives occur in 6 tests equals (0.95)^6 and hence the probability that at least one false positive occurs is 1–(0.95)^6 = 0.26. This also highlights a discrepancy regarding the primary outcomes between the pre-registered trial protocol on ClinicalTrial.gov (https://clinicaltrials.gov/ct2/show/NCT02145728) and the published protocol by the same authors in 2015 . Notably, O’Keeffe et al.  gave up on the postintervention follow up results, which was a primary outcome according to the pre-registered protocol. Therefore, it is mandatory to discuss this aspect and to reduce the number of primary outcomes in future studies, if possible, to 1; in this case, disability at 6 months after randomization would have been recommended.
Finally, 37% of the participants were lost to follow up in the first primary outcome endpoint (postintervention), 28% of loss of follow up on the second primary outcome endpoint (6 months), and 31% on the third (12 months). Although the unmeasured bias was recognized by O’Keeffe et al. , it seems that there is a systematic high loss of follow up in CFT clinical trials conducted by the research group of the developers of the method. Similarly to the first trial of Vibe Fersum et al. , this trial did not provide high quality evidence about the efficacy of CFT and should be considered as an exploratory study, not confirmatory enough to generate recommendation. Therefore, the next CFT trials should focus on improving the methodological quality. It is recommended that future studies avoid a loss of follow up higher than 15%, blind the assessors, establish up to 2 primary outcomes, and special attention be paid to the quality and duration of the treatment provided to the comparison group.
1 O’Keeffe M, O’Sullivan P, Purtill H, et al. Cognitive functional therapy compared with a group-based exercise and education intervention for chronic low back pain: a multicentre randomised controlled trial (RCT). Br J Sports Med 2019;:bjsports-2019-100780. doi:10.1136/bjsports-2019-100780
2 Vibe Fersum K, O’Sullivan P, Skouen JS, et al. Efficacy of classification-based cognitive functional therapy in patients with non-specific chronic low back pain: a randomized controlled trial. Eur J Pain 2013;17:916–28. doi:10.1002/j.1532-2149.2012.00252.x
3 Fersum KV, Smith A, Kvåle A, et al. Cognitive Functional Therapy in patients with Non Specific Chronic Low Back Pain A randomized controlled trial 3-year follow up. Eur J Pain 2019;:ejp.1399. doi:10.1002/ejp.1399
4 O’Keeffe M, Purtill H, Kennedy N, et al. Individualised cognitive functional therapy compared with a combined exercise and pain education class for patients with non-specific chronic low back pain: study protocol for a multicentre randomised controlled trial. BMJ Open 2015;5:e007156. doi:10.1136/bmjopen-2014-007156