Here's an RCT (randomised controlled trial) of a public health measure - hand washing. A very full write-up with plenty of detail on methods, how the outcome was chosen, study power and so on - worth a careful read!
New York's attempt to tackle obesity by limiting super size drinks fell to objectors and the Supreme Court, but over in California (where else?) they are having more success.
The best bit of this paper is the introduction - some basic economics about elasticity of price and demand, and an interesting section on why it's better to use excise duty rather than sales tax as the vehicle to increase prices. It depends partly on a habit common in Canada and the US but not Europe: prices posted in cafes are not what you actually have to pay - the sales tax gets added later. I've always thought of that as a kind of quiet revolt against government: "I'm only charging you $2, so don't blame me if you have to pay $2 and 15c - that's the fault of the government." I wonder why that doesn't happen in Europe; are we more passive in the face of taxation?
Every now and then there is a dry period with no interesting articles, so no post last Friday, but here's something interesting. You'll need a quiet moment to understand it.
The grand question is whether it matters who your GP is. But grand questions are not researchable, so we need to be more precise. The first refinement is to focus on cancer. Cancer can't be treated by a GP: he or she must refer to a hospital service. So the focus narrows down to referrals, and how they differ between GPs, or more precisely between practices. Some practices refer more, and some less; some sooner and some later. It's a difficult balance if you're a GP. Refer too much the hospital gets swamped; refer too little and the patient's cancer may advance beyond a treatable stage.
The study compared a practice's 'propensity' to refer with its conversion rate (what percentage of the referrals turned out to be cancer) and survival. Low propensity to refer did indeed correlate with poorer survival but we need to look at the effect size: the difference is not, to my mind, enormous.
We also need to be careful about what to conclude here. The easy conclusion is to blame low referring GPs for missing clues to cancer. But poorer survival may also (as the authors point out) be due to poorer treatment at the local hospital. If you think about it, the best surgeons etc tend not to choose their careers in the sort of places that have large numbers of rubbish GPs. A double whammy.
Breast cancer proves to be an interesting exception.
See also a thoughtful editorial from the excellent Willie Hamilton.
I suppose vaccine trials are the commonest form of randomised controlled trial in public health.
Here is an account of a large (they're all large) trial of dengue vaccine. Dengue isn't the worse virus in the world so the number of countable events (hospitalisation) is remarkably small. That puts a premium of high quality fieldwork: a few missed events and you're in trouble:
'hospitalization for virologically confirmed dengue occurred in 65 of 22,177 participants in the vaccine group and 39 of 11,089 participants in the control group'
The more alert among you will have noticed that they randomised 2:1 vaccine to control. As is common in vaccine trials, both intention to treat and per protocol analyses were carried out.
It seems particularly true of virus infection that they're less severe if you get them young - chickenpox, mumps and rubella spring to mind. So if you're not careful, a vaccine campaign could do more harm than good.