Preventive trials biases
Not always, at least. Things have to be done at the right time and, in many cases, it’s of no help to do it ahead of time. However, in medicine we are prone to think that it is better to do everything ahead of time, and that is not always true.
For instance, when we speak about prevention we usually assume that the earlier we diagnose a problem, the better the prognosis will be. Nevertheless, this is not always true and, when it is, it may be difficult to prove.
As it happens, prevention studies usually involve healthy people so, when we value a preventive intervention, we have to wait longer periods of time in order to give the illness that we want to prevent the opportunity to happen. In addition, it may be that, although the intervention controls the risk factor or disease, the patients’ prognosis or survival will not change at all, either due to the studied illness or to other independent factors.
An additional difficulty is posed by usual and characteristics biases affecting observational studies, which may lead us to a wrong conclusion in favor of the intervention that we are studying. And, moreover, to add insult to injury, studies of preventive measures are exposed to three types of characteristics biases.
Preventive trials biases
The first one is the participation bias. It often happens that those who agree to participate in these studies are healthier than those who reject it or who do not have access to them. Thus, any observed benefit may be due to this factor or even to any other uncontrolled factor, all because of the absence of a random distribution of participants between control and intervention branches of the study.
On the other hand, diseases have a latency period since they begin until they become evident and are usually diagnosed. If we study a preventive measure and make a diagnosis during the latency period, survival may look prolonged, but not because the patient lives longer, but because we have made the diagnosed earlier and have started to count time before. This is call lead time bias. It’s not the survival what is increased, but the known time of disease.
Finally, the third source of error is called the length time bias. It might be that, diagnosing the disease in earlier phases, we are detecting cases with a longer presymptomatic period, which may be less severe and with a better prognosis. We thus have the false sense that survival is higher in cases detected earlier than in those who are diagnosed at the usual time.
It is obvious; the way to avoid these three biases is to assign participants randomly to intervention or control branches. Put another way, we must do a randomized controlled clinical trial if we want to safely demonstrate the efficacy or lack of efficacy of any preventive measure.
And that’s all for now. We have not discussed other characteristics of preventive trials regarding follow-up and sample size calculation. Being healthy subjects, the number needed to participate or the necessary follow-up period to observe the effect under study may be higher or longer than those for other types of clinical trials. But that’s another story…