Including Comorbidities Posted on October 19, 2007 Deciding what inclusion/exclusion criteria to use is tricky, states Helena Kraemer (bio).

This is very controversial. I mean it’s a very difficult decision to make. The FDA has no requirements for example on the representativeness of the sample, at least the last time I checked on this. And so the drug companies tend to set very narrow inclusion and exclusion criteria. It’s not unusual for example for an FDA type study to exclude, if it’s doing a research project on Alzheimer patients for example, to decide to exclude all those with major comorbidities. Now anybody over the age of 50 knows that there aren’t too many people who don’t have multiple problems at the same time. So excluding people with comorbidities from such a study means that the sample that they actually do the study on is not representative of all Alzheimer’s patients, and the results don’t generalize.

In recent years, there’s been a lot of pressure to try to get samples that are really representative of the people the clinicians have to deal with, which means very heterogeneous samples. It means fairly wide inclusion, exclusion criteria. You take them as they come for the most part, and the only people you would have to exclude are I guess two classes. One of them for ethical reasons, that is if any of the procedures that you are going to use during the study might potentially harm a patient of a certain type. Those people should not be included in the study. And the other is a more research oriented one and that is that if you know from previous studies that certain kind of people will not be benefited by the study, it doesn’t make any sense to include them in. I guess that’s an ethical limitation as well.

And so the issue of what your inclusion, exclusion criteria are going to be becomes important for several reasons. One of them is the narrower they are, the harder it is to recruit an adequate sized sample. On the other hand, the less generalizeable it is, the wider the inclusion, exclusion criteria, the easier it is to get the sample but then on the other hand, the larger the sample has to be in order to get enough power to test your hypotheses. So it’s a tricky question.

« Back to Article