Jonathan Edwards
Senior Member (Voting Rights)
Quite. Which is why I feel that where there are documented cases of suspected harm, albeit anecdotal at that point, if a more rigorous study were to follow those up, and try to apply some scientific investigations into them, maybe something might come of that. I mean, what happens if a drug starts getting widely reported as causing severe subjective side effects? Is that just ignored? What is done in such cases, that could not be done here?
There is a particular difficulty if a potential side effect of a treatment is exacerbation of the pre-existing problem. For most drug side effects the effect is something rather rarely encountered in the treated population. Most people with a cough do not have a widespread itchy rash suddenly appear so it is relatively easy to suspect that an antibiotic was the cause. It is even easier if the side effect is rare, like bone marrow failure.
If the unwanted effect is an exacerbation of a fluctuating condition then the only way to gather evidence is to show that the pattern of the worsening is different in a treated population from a non-treated population, using statistical analysis. That need not be confined to mean levels. A different pattern of variance would be important. Thus a scatter plot of the untreated group might show after a certain period of time that the individuals' levels had wandered around but the spread of levels was the same. For the treated group after a period of time there might be a slight upward trend as whole but 10% of cases might dip way below the previous range.
That is why we really need scatter plots for all these studies, to see changes in variance patterns. Maybe the data for PACE exist somewhere on a computer, but drop outs would need to be addressed too.