Do you read the small print about side effects? Does your doctor tell you? What are the chances that taking a medicine will kill you or make you ill in an entirely new way?
This last question is one reason why trials are run before a medicine is approved for use. However, there is an inherent flaw in the system. No matter how many people are in a trial there is a chance that a side effect will be missed. Consider this: Imagine choosing one school class out of all the classes in the country to check the prevalence of albinism. Given only about 1 in 17,000 people have albinism then you can imagine that it is unlikely that you will find an albino person. However, you may find a red haired child because the prevalence is much higher. If, though, you check a whole school you may still not find albinism. Can you, then, conclude there is none? No, because even if the school has 2000 pupils there is only a small chance of finding the condition. Quite simply, the rarer the condition the more pupils need to be checked. In terms of drugs, the rarer the side effect the more people need testing.
However, the difference between new drugs and our analogy is that trials are looking for unknown side effects. What this means, is that a statistician can turn things around and say that for a trial of a given size if a side-effect is not seen what is the maximum prevalence of that side effect. If, for example, you only cared if the drug increased the risk of a side-effect by more than 5 times (Relative Risk = 5) compared with those not taking the drug, and the event was relatively rare (say 1 in 5000), then you would need to have a trial with 14,707 people taking the drug and another 14,707 in a control group (e.g. taking a drug already used to treat the disease) in order to be reasonably certain that the new drug did not increase the risk by more than 5 times (see the table). The side-effect you are interested in may be serious (eg kidney cancer), but if the drug is saving many lives in the first place (eg a drug that suppresses the immune system and allows transplants to take place), then it may be an “acceptable” risk. The point is that trials must be of sufficient size to measure an “acceptable risk.”
In a recent article published in PLOS Medicine (see here) researchers looked at the number of participants in trials of drugs approved by the European Medicines Agency over the last decade. Quite shocking is that for medicines intended for long term use (Chronic diseases) nearly 20% were approved even though they did not meet the Agency’s own criteria for numbers of patients in studies, and these were very very low numbers indeed (300 over 6 months)! Only about 10% of studies had sufficient participants to pick up on a risk of greater than 5 fold with an incidence of more than 1 in 1000. This highlights why it is absolutely imperative that there are further ongoing studies monitoring side effects of drugs once the drug has been approved. Anyone who has read Ben Goldacre’s book “Bad Pharma” will know such studies are often poorly done if done at all. How do we ensure such studies are done? Do we legislate that drug companies do them (potential bias here), or do we make sure we have a well trained, adequately funded independent group of scientists able to do this? If you think the latter, let your MP know! In the meantime, let the medicated beware.