In my first post on medicine I talked about the various dangers that can arise from medicine that’s neither effective nor particularly harmful. In my second post, I talked about one particular alternative medicine, PSE, showed that it’s not based on any evidence whatsoever and most likely does not work, though that can’t be confirmed because in it’s 15-year runtime it has yet to be properly and fairly tested. That’s fairly common in alternative medicine.

In this post, I want to tackle an argument you can often hear from advocates of alternative medicine: “Why should I want evidence when all the “evidence-based” medicine isn’t based on any actual evidence?” I’ll also give an introduction into evidence-based medicine, albeit a very brief one.

For the purpose of this post, I’ll recommend two excellent books:

Goldacre, B. (2012) Bad Pharma – How drug companies mislead doctors and harm patients

Evans, I., Thornton, H., Chalmers, I. & Glasziou, P. (2006) Testing Treatments – Better research for better healthcare

About 90% of my examples come from these two books, mainly because they give nearly all the relevant sources. (Meta studies and so on)

As Evans et al. document: (P.13-14) As a parent, the most horrible thing that can happen to you is your child dying. Naturally, parents will immediately do what a doctor tells them if it’s said to reduce infant suffering and death. In 1946, Dr. Benjamin Spock‘s book “Baby and child care” suggested babies sleep on their backs. In 1956, he revised that statement, saying:

There are two disadvantages to a baby’s sleeping on his back. If he vomits he’s more likely to choke on the vomitus. Also he tends to keep his head turned towards the same side… this may flatten the side of the head… I think it is preferable to accustom a baby to sleeping on his stomach from the start.

That sounds like incredibly sound advice, which is why many parents (millions?) opted to do so. However, there was one serious flaw: This practice, never evaluated, led to an estimated 50,000 deaths. A first study suggesting harm was published mid 1960’s, a second in the early seventies, two further ones in the mid 80’s. It was only properly acted on in the mid 90’s. (In some cases until 2002!)

Now one might argue that this is a case of “evidence-based medicine” gone wrong. I disagree, it’s precisely the opposite. Here, no evidence or very shoddy evidence was provided, which eventually led to tens of thousands of deaths. Had the evidence been appraised correctly right away, many if not most of these deaths might have been avoided.

There are dozens of examples like this in the two books, so I won’t go through them. The authors have gone to great pains to document these cases and I recommend you read about them yourselves. I just want to bring this back to the point I was trying to make: Evidence-based medicine can only function if we look at the evidence carefully and without bias, and if we then act on the evidence as swiftly as possible.

This system is by no means perfect, because humans will always find a way to either intentionally interfere with the process (for political or monetary reasons) or to be so stupid as to fuck up. However, it’s better to have a system that’s based on evidence we can theoretically gather (evidence-based medicine) than basing it on a system that involves nothing more than guesswork. (alternative medicine)

At this point, I should probably point out how incredibly fragile this review process is. Ben Goldacre documents: (P. 69)

Imagine you’re conducting a study on a particular drug. Let’s call it pregabalin, a drug to help diabetics handle pain when their nerves have been affected by the disease. You gather test subjects, you double-blind the study, you do everything you’re supposed to. But some subjects drop out of the study, due to various reasons. It might be that they still feel pain and don’t want to continue or the side effects might be too severe. Whatever the reason, you’re now faced with a problem. You’ve got half a dataset for the drop-out patient. How can you incorporate the results in the study?

The researcher, let’s call him Pfizer, does a clever thing: He uses the “last observation carried forward”-method. That’s a fancy sounding name for “I’ll put down the last result I got into every column. Basically, the patient had pain levels of 10, 9, 8, 7, 6, 5, 5, 6, 5, and then dropped out. So Pfizer fills in the last six data-points with all 5’s. Before the study, you will notice, the patient had pain levels of 10.

The research then gets published and your drug looks quite good indeed. But wait! what’s that? Some researchers disagree with your method? But why? Well, the reasoning is this: The patient dropped out, so their pain level is back to 10. Obviously, the side effects were worse than the pain and they stopped using the drug. So we have to put all 10’s for the last six data-points instead of all 5’s. That’s called the “baseline observation carried forward”-method.

The difference between the two methods is staggering: With the LOCF the drug gets overestimated by nearly 25%!

That, you will agree, is quite a lot. Once again it shows: Medicine must be based on the best possible evidence, not on guesswork and shoddy methods.

I’ll leave you with a few points:

• Biased testing will leave you with incorrect data, thus putting patients at risk.
• All medicine should be tested and carefully evaluated. Failing to do so puts patients at risk and should be persecuted.
• Trusting doctors blindly is not always a good choice. Ideally, your doctor should be able to show you what research (s)he’s basing his/her decision on.
• Basing therapies on theories instead of real-world outcomes is again putting patients at risk.
• As suggested in my first post, using treatments that are neither directly harmful nor beneficial is also not recommended.
• Established treatments are not necessarily beneficial: Check and re-check. I explained that partially in my second post.
• Tests should be as fair as possible, that is to say: Randomised (double-blinded) Controlled Tests (RCT’s), comparing like with like, meta-studies, etc. This requires that potential researchers and future doctors be trained in the best available methods.

I’ll leave it at that and once again recommend the two above mentioned books. Remember, all science needs evidence and we need laws to ensure the best standards.  We need institutions like the Cochrane Collaboration to appraise the available evidence without interference from commercially or politically driven institutions. Read “Bad Pharma”, it’ll really shock you.