3
$\begingroup$

I'm not involved in drug testing, but it is my understanding that most trials are performed using a control and a treatment group (or perhaps a few discrete treatment groups). This data is then used to determine effectiveness of a drug, and hopefully uncover any deleterious side-effects before the drug is released on the population.

I'm wondering why it would not be more sensible to use a continuous distribution of drug dosages, rather than a discrete distribution? For example, give 20% of patients the control, and the give the other 80% a randomised (uniform?) dosage between 0% and ~110% of the expected optimal dosage. It would seem to me that this would provide more information than data that is inherently discretised, as well as potentially making it easier to more accurately fit models like logistic functions for both effects and multiple side-effects that occur at different rates?

The obvious draw-back would be the increased cost of manufacture for the trial, but that would be fairly minimal in comparison to potential benefits, especially for large trials. Also, this method could just as easily be applied to any "treatment" experiment (e.g. in agriculture, metalurgy, etc.). So why aren't more trials conducted like this?

Improve this question
$\endgroup$
3
  • 2
    $\begingroup$ In medicine dose-response relationship is not straightforward and can differ between subjects, decreasing the power of test. Additionally, patients are ill and often require some treatment. The low dose of drug can expose the life of patients to unnecessary risk. $\endgroup$
    – O_Devinyak
    May 15, 2013 at 10:52
  • $\begingroup$ @O_Devinyak: ok, but more than the control group? $\endgroup$
    – naught101
    May 15, 2013 at 12:55
  • 2
    $\begingroup$ For sure, there are many issues where control is placebo. But in modern medicinal chemistry to launch a new (original) drug in market the manufacturer should prove that this drug overcomes traditional treatment at least in some situations (certain disease, or its combination, or the disease with associated physiological pecularities). So more frequently the control is a reference drug (in other words, control group is treated with reference drug). Placebo is used when treatment can be advantageous, but is not necessary. $\endgroup$
    – O_Devinyak
    May 16, 2013 at 12:17

2 Answers 2

Reset to default
2
$\begingroup$

There are several types of clinical trials, including dose-finding trials. I have not personally been involved in clinical trials but as far as I know, the dose administered in the largest phase III trials is typically determined using those earlier trials.

Among other reasons, it is important to obtain data on effectiveness and side-effects at the dosage that will actually be manufactured and prescribed later on. I also suspect that manufacturing is not the main draw-back or cost driver, finding enough qualified patients, managing the trial and getting results in a timely manner would be much bigger issues. Attaining sufficient power with one treatment group is already very costly so you don't want to expand resources on other questions without very good reasons. At this stage, the clock is already ticking on the patent so more conditions means more time to recruit patients and ultimately less profits.

Furthermore, physiologically speaking, the effect of dosage is complex but we do know a few things about it so there would be no point and a lot of ethical difficulties in experimenting blindly with all sorts of doses with a large number of people.

Generally speaking, most trials are part of a long process with far-ranging regulatory, financial and industrial consequences so what would seem to make sense in the context of an isolated study is not necessarily the most sensible thing to do from a business or from a medical perspective.

Even for academic studies with no patented drugs (e.g. psychotherapy), finding enough patients can be a big challenge, compared to simple experimental research with healthy people. Not only are some diseases/afflictions uncommon but there are usually other stringent screening criteria. Adding conditions therefore involves significant costs and delays and is not a decision to be taken lightly.

Improve this answer
$\endgroup$
0
$\begingroup$

I know nothing about medical trials, but when your treatment variable is a dummy, you can just run a univariate OLS regression/t-test. When the treatment is continuous you need to pick a functional form or use complicated non-parametrics that many people don't understand. Alternatively, you can do a multivariate OLS regression on a bunch of dummies, with less power than you'd have if you had everybody in the treatment group getting the same treatment.

So, power and parsimony.

Improve this answer
$\endgroup$
1
  • $\begingroup$ Couldn't you discretise the data afterwards, and then run any tests that would be available for discrete data? $\endgroup$
    – naught101
    May 15, 2013 at 12:57

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.