Heroes Not Zombies

An article in the BMJ recently repeated the statement made several years ago by a researcher who works in the area of pharmacogenomics for GlaxoSKF, the drug company. He said

“The vast majority of drugs – more than 90 per cent – only work in 30 or 50 per cent of the people,” Dr Roses said. “I wouldn’t say that most drugs don’t work. I would say that most drugs work in 30 to 50 per cent of people. Drugs out there on the market work, but they don’t work in everybody.”

Whilst that observation caused a stir at the time, and is causing a stir again, now it’s been repeated, at the time it didn’t really surprise me. You don’t have to work as a GP for long to discover that there are no drugs which do what the manufacturers and researchers say they do for every single patient who you prescribe for. How many different BP pills does the doctor have to try sometimes to get hypertension under control? How many different painkillers? Different antidepressants, anticonvulsants, treatments for constipation, diarrhoea…..you name it. I really don’t know of any drug on the market which does what it claims to do for EVERY single patient who takes it. What did surprise me were the figures quoted – 90% of the drugs only work in 30 – 50% of the people!

And yet, there are still those who claim there are only two kinds of treatments available – those which work, and those which don’t.

Life just isn’t like that.

But here’s another comment in that BMJ article which really grabbed me, and I don’t know why I didn’t see this so clearly before!

Pain relief is not normally distributed but usually bimodal,being either very good (above 50%) or poor (below 15%). Using averages is unhelpful and misleading, because “average” pain relief is actually experienced by few(if any)patients, and it tells us nothing about how many patients will experience clinically useful pain relief [BMJ 2013;346:f2690 doi: 10.1136/bmj.f2690]

What does this mean? “Bimodal”? Well, here’s another article, referred to in this BMJ article, pointing out the same problem –

Systematic reviews of regulatory trials often pool average data. In acute and chronic pain, however, underlying distributions are commonly not normal, tending to be U-shaped rather than bell-shaped, where the average describes few individuals [PAIN 149 (2010) 173–176]

When you look at the effect of a drug on a research population you don’t get drugs which work, and those which don’t. What you get is two distinct groups of patients – those who get a “good” result, and those who don’t.

By averaging out the results of the entire group, this reality is obscured.

Whilst these articles refer to painkillers, I believe this finding is likely to be found with pretty much any therapy you can think of. There will be a group who really get no benefit, AND a group which get significant benefit.

This is a common problem in health care – there are no average people. Every single person needs to be considered and treated as an individual. After all even the results from the group trials have been obscured by this averaging out.