Investigating the veracity of a sample of divergent published trial data in spinal pain

Abstract

Evidence-based medicine is replete with studies assessing quality and bias, but few evaluating research integrity or trustworthiness. A recent Cochrane review of psychological interventions for chronic pain identified trials with a shared lead author with highly divergent results. We sought to systematically identify all similar trials from this author to explore their risk of bias, governance procedures, and trustworthiness. We searched OVID MEDLINE, EMBASE, CENTRAL, and PEDro from 2010 to December 22, 2021 for trials. We contacted the authors requesting details of trial registration, ethical approval, protocol, and access to the trial data for verification. We used the Cochrane risk-of-bias tool and the Cochrane Pregnancy and Childbirth group's Trustworthiness Screening Tool to guide systematic exploration of trustworthiness. Ten trials were included: 9 compared cognitive behavioural therapy and physical exercise to usual care, exercise alone, or physiotherapy and 1 compared 2 brief cognitive behavioural therapy programmes. Eight trials reported results divergent from the evidence base. Assessment of risk of bias and participant characteristics identified no substantial concerns. Responses from the lead author did not satisfactorily explain this divergence. Trustworthiness screening identified concerns about research governance, data plausibility at baseline, the results, and apparent data duplication. We discuss the findings within the context of methods for establishing the trustworthiness of research findings generally. Important concerns regarding the trustworthiness of these trials reduce our confidence in them. They should probably not be used to inform the results and conclusions of systematic reviews, in clinical training, policy documents, or any relevant instruction regarding adult chronic pain management.