NL
In a recent study, Cochrane Rehabilitation* examined the influence of bias on effect estimates in randomized clincial trials (RCTs) in the field of rehabilitation. The researchers included seven meta-epidemiological studies with more than 200,000 participants, and the findings suggest that some forms of bias may overestimate effect estimates. However, the main conclusion is that the current evidence is inconclusive and inconclusive. More research is needed to better understand the relationship between bias and effect estimates within rehabilitation research.
Supporters of evidence-based practice argue that clinical decisions are based on patient preferences, the clinical reasoning of healthcare professionals and on the best available scientific knowledge. Especially the latter, the best available scientific knowledge, is a tricky concept. What exactly is the "best" scientific knowledge? There are researchers who argue that the best knowledge is generated using randomized and blinded scientific research in the so-called RCT. Over the years, there have been agreements that a good RCT must meet. These include proper randomization, blinded participation of patients, practitioners and effect evaluators, presentation of all results, and so on. However, within rehabilitation research, it is not always possible to conduct the perfect RCT. Exactly what impact these shortcomings in RCTs (bias) within rehabilitation research have on effect estimates is still largely unknown.
For this reason, researchers, led by Cochrane Rehabilitation,* conducted a study of the impact of shortcomings in RCTs of rehabilitation interventions on published effect estimates. The study, based on the analysis of seven meta-epidemiological studies published through June 2023, shows that a clear relationship between a deficiency and the over- or underestimation of an effect appears to be lacking. Inadequate randomization seems to result in overestimation of rehabilitation effects. Less clear were the results on the lack of blinding within RCTs. One study found that not blinding in RCTs underestimated the effect, while another study overestimated the effect. They also found conflicting evidence about the risks of dropout in studies and whether or not to perform an intention-to-treat analysis. The researchers did find a relationship between the risk of selective reporting and overestimation of treatment effects.
Overall, the study highlights the complexity of the relationship between domains of bias and effects of rehabilitation. The mixed and often contradictory findings highlight the urgent need for further studies to better understand this relationship and improve the methodological quality of RCTs in rehabilitation medicine. Furthermore, the researchers recommend introducing other types of research within rehabilitation medicine.
*Thomas Hoogeboom, senior researcher at the IQ Health department, was part of the Cochrane Rehabilitation consortium.
Read the article here by Arienti C, Armijo-Olivo S, Ferriero G, Feys P, Hoogeboom T, Kiekens C, Lazzarini SG, Minozzi S, Negrini S, Oral A, Pollini E, Puljak L, Todhunter-Brown A, Walshe M; Participants in the 5th Cochrane Rehabilitation Methodological Meeting. The influence of bias in randomized controlled trials on rehabilitation intervention effect estimates: what we have learned from meta-epidemiological studies. Eur J Phys Rehabil Med. 2023: 82124722.pdf (gcu.ac.uk)