“It would be unethical to have a control group”

November 20, 2008 at 6:28 pm (Bad Science) (, , )

This is something I’ve heard a few times now. I emailed a health writer, Bill Sardi, regarding a piece he’d written on a miracle anti-cancer drug. His piece referred to a trial published in 2008 in the International Journal of Cancer that had a total of 16 women all given the drug for 4 years. As JCmacc pointed out on the Bad Science forums: There was no control group. All 4 women had a breast tumour removed surgically and were given chemotherapy or radiotherapy to kill off any metastatic cells prior to the trial in question starting. The fact none of these women relapsed during the 4 years isn’t a shock given their conventional therapy. I dropped Bill an email and put some of these points to him (that there was no control group, that the women has conventional treatment…) and Bill responded with an email that included this:


Now, I’ve been thinking this through and I’m not sure how it can be unethical to have a control group for a treatment that may or may not work. It is true that some trials have been stopped early on ethical grounds when it became clear that substance x worked and it would have been unethical to deny the control group this remedy (or when it became clear that substance y was linked to an increased risk of death, as I believe was the case with Beta Carotene in smokers*). However, I think it is generally the case that if you don’t know whether a treatment works then you can’t know whether it is unethical to withold it – and the point of conducting a trial is to find out precisely that. However emotively it is framed, the idea that it is unethical to withold unproven treatments from a group in a trial is a wrong-headed one in my book.

You can Google “BIll Sardi” for more health advice – on resveratrol, on statins not saving lives, on dissolving gallstones naturally and, of course, on antioxidants.

*Beta Carotene and smokers – NCI summary of trial.

Durham CC thought it unethical not to give unproven fish oil pills to children: The Fishy Reckoning.

And finally: someone has done a meta analysis looking at “Placebo-Controls in Short-Term Clinical Trials of Hypertension” Al-Khatib. Also – The Ethical Use of Placebo in Clinical Trials Involving Children (Derivan et al).


  1. dvnutrix said,

    Ethically, you must have control groups, wherever possible and appropriate, precisely because this research is so important.

    Control group does not always imply, no treatment. In many cases, an intervention is assessed against the current best practice. Sometimes, I think some commentators don’t communicate that very clearly.

    Where there is no control group, you have to emphasise that, front and centre, when you are discussing the results. For preference, you might provide a frame of reference by discussing survival times, morbidity levels etc. in comparable groups, but you would frame and phrase that carefully because you are not discussing people in the same study.

    I worry about some of these people passing on their skills in research interpretation – and their influence in making other people think that what they discusses passes as either science or scholarship.

  2. dvnutrix said,

    jdc, you may have more than a passing interest in a new website in Scotland that is even advertised on TV: Get randomised.

    We are all in this NHS together. We need to continually find out how to improve this service…

    The popular view of medical research is that it is something that only goes on in hospitals or in laboratories by people in white coats. That is partly the truth but by no means all of it. What we need to know most is how well do our treatments work in the real world? These treatments could be drugs, surgical techniques, appliances, therapies, medical gadgets or anything else used in a person’s care. This is where we all come in. To be sure how well treatments work, how safe they are and how they can be improved we need to do a lot more studies involving people who already have the sort of conditions that we are trying to treat more effectively.

    This is what the Get Randomised campaign is all about.

  3. stavros said,

    “In many cases, an intervention is assessed against the current best practice.”

    That’s true: “control group” does not always mean “placebo group”. In fact, if I remember correctly, most of the times it is the current best treatment as dvnutrix mentions.

  4. jdc325 said,

    In a further communication, Bill has referred to placebo group rather than control group.
    The trial referred to involved women all being given conventional treatment plus Gc-MAF, an appropriate control would have been “conventional treatment only” which, I guess, amounts to the same thing as “current best treatment”.

  5. jdc325 said,

    dvnutrix – yes, the Get Randomised campaign is of interest to me. I’m impressed with the initiative. I think RCTs could be useful in many areas (e.g. social policy), but medicine is obviously a field in which conducting such trials could save lives and, as such, probably the most important one.

  6. sanjiva said,

    The control arm does not have to be a no treatment arm. In most cancer trials, the control is usually the current standard of care.

    That said, there are trials in advanced cancer patients that will compare an experimental treatment to best supportive care only.

  7. Autism Candles said,

    Reblogged this on Autism Candles Blog.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: