New research demonstrates the importance of negative results, says Torsten Rackoll

Doing fitness exercise research in neurorehabilitation is very satisfying, especially for early career scientists. We deliver therapies which, at least in part, follow recommendations for several other chronic diseases, and therefore address comorbidities as well. Unfortunately, with the results of our new study, published today in The BMJ, the point in time came, when a beautiful idea was quashed by facts, followed by uncomfortable questions of what to do when negative results arise. 

In our study we compared aerobic body weight supported treadmill training with relaxation therapy in moderately to severely impaired patients following subacute stroke. The rationale to perform aerobic treadmill training during the early phase of neurological recovery seemed quite strong. So we were surprised when our statistician told us that the intervention had no effect compared to relaxation sessions. After we had put so much hope into our intervention, we were stunned. 

We discussed these findings with our participating study sites to get more information from the therapists about the perceived clinical course of the patients in our study. It turned out that there was also a discrepancy between the statistical results of the trial and the perceived success of our intervention from the therapist’s viewpoint. These discussions convinced us once more of the relevance of our data. They show that it is difficult to estimate the real effect of a therapy in everyday clinical practice as well as the effect seen in smaller studies. Still, it became obvious that our results were not at odds with all previous studies, but rather, cast doubt on a “one size fits all” approach of aerobic exercise.  

Another unanticipated hurdle was the resistance from some journals and reviewers alike regarding the publication of null results. Not being able to convince some reviewers of the importance of our results despite long and exhausting discussions left me puzzled. Don’t our results help to assess the real effect of aerobic exercise as well as the suitability for patients? Thus, we were glad about the fruitful discussions held with the reviewers involved in the publication at The BMJ. The academic community would benefit from further discussion about the relevance of null results. 

In general, in terms of our intervention, there is good evidence on using aerobic exercise in post stroke rehabilitation. But it seems not to be the best choice in moderately to severely impaired subacute stroke patients. Some positive effects of a more active lifestyle might still be present, but the risks in this sample outweigh any beneficial effects e.g. metabolic changes. Aerobic exercise in less affected subacute stroke patients could be valuable, but the best therapeutic strategy in severely impaired patients still needs to be developed. 

Torsten Rackoll is a PhD candidate at the Charité Universitätsmedizin Berlin and a research associate at the Kliniken Beelitz neuological rehabiliation clinic.
Twitter: @TorstenRackoll
Competing interests: See research paper