When I wrote recently that Medicare bases its reimbursement decisions on trials that largely exclude the elderly population (Tested on the young, used on the old), Cor Doc commented –quite rightly– that there may be good reasons for this:
There are difficulties performing clinical trials in the elderly population. This is most often due to multiple comorbid illnesses in the older patients. Trials are usually designed to exclude such subjects lest their other illnesses cloud the outcome of the study and cast into doubt the significance of the results. Performing clinical trials is challenging to begin with because we are constantly assessing a sample of the general population and then applying these results to the world-at-large. The same statistical flaws can therefore theoretically affect clinical trials as they can New Hampshire poll results.
While we would love to study outcomes in a target population with every intervention, researchers hold sacred the validity of their results. Having results as iron-clad as possible, therefore, means that one must remove any confounding variables.
Three years ago I looked into a related topic –the proliferation of exclusion criteria in pharmaceutical clinical trials. Many of these trials were designed in a sloppy manner: exclusion criteria were added without a clear rationale, in a way that made the trials hard to conduct and the results less useful.
In a 2002 article in the American Journal of Psychiatry, Posternak et al. studied exclusion criteria in drug trials and made this conclusion:
Many of the standard exclusion criteria currently used in antidepressant efficacy trials may not be achieving their intended goals of maximizing drug-placebo differences. The practice of excluding subjects with particular clinical profiles, which greatly reduces the generalizability of antidepressant trials, appears to lack empirical support.
Posternak’s study was based on trials in the 1990s, but the phenomenon was alive and well a decade later. As one researcher said:
Companies are making exclusion criteria too narrow to the point that it’s very difficult to meet enrollment expectations. Oftentimes they are also vaguely defined, such as excluding those with borderline personality disorder without specifying how such patients are to be identified.
Let’s hope designers of trials are being more thoughtful these days.February 5, 2008