More information about text formats
Russell et al report on the risk of neurodegenerative disease (a composite outcome including dementia, motor neuron disease (MND) and Parkinson’s disease) among former international rugby union players. Using a matched retrospective cohort design they show not only that the rugby players group had two and half times the risk of developing neurodegenerative disease, they were also more likely to die from, be hospitalised due to, or be prescribed drugs related to neurodegenerative diseases. These are important findings.
The risks for Dementia, Parkinson’s disease and MND are also estimated. It is regrettable that many media outlets (The Times, Independent, Mirror, Irish Times and the Guardian to name but a few) have focused on the secondary motor neuron disease outcome, with its odds ratio point estimate of around 15, given the considerable uncertainty attached to this figure (95% confidence interval 2.10 to 178.96). Although the other outcomes have more events, and are consequently more likely to be robust, attention has been drawn to the most eye-catching 'risk' figure, even though this is not one of the primary findings of the paper.
What seems to be missing from this report is information about how many players were diagnosed with MND, or what these findings mean in terms of the absolute risk. The authors do not report how many people had MND in the rugby group, but they do say "In the analysis of MND/ALS, zero events were recorded among the...
What seems to be missing from this report is information about how many players were diagnosed with MND, or what these findings mean in terms of the absolute risk. The authors do not report how many people had MND in the rugby group, but they do say "In the analysis of MND/ALS, zero events were recorded among the general population comparison group. To accommodate this, one general population observation was re-allocated from no event to event" and then say that the absolute number was "low" without explaining exactly what this means. We think it is likely that 5 events in the rugby group would be consistent with an odds ratio of 15, i.e. (5/407)/(1/1235)=15.17. Although this does appear to indicate an apparent increase in risk, we question the validity of trying to quantify the risk when this can only be achieved by modifying the data, and with such modification the conclusion has very little certainty. It is unfortunate that this may have led to misunderstanding in several media reports.
We reiterate this work is important and points towards an increase in risk for many of the outcomes reported, but the risks in public health terms for relatively rare outcomes should not be exaggerated until they can be replicated in larger samples.