Another product from the Framingham study! It's difficult not to be wowed by the video attached to web version of this piece - but we need to check the underlying science is sound. The basic message is that obesity is not just about you, it's about your friends and neighbours. Another reminder of why do public health - man is a social being and influenced by social processes.
Social processes are difficult to measure, which means the statistics can be incomprehensible: and incomprehensible means potentially seriously misleading.
In this case, we have to take comfort in the fact that the analysis is published in the NEJM. It's the world's leading biomedical journal so I guess they can call on some seriously competent referees.
It turns out that my note of caution about the statistics was well founded and my trust in the NEJM less so: see this critique of the paper.
Bottom line - beware complex statistics, always look at the raw data. And look out for critiques. (Incidentally I'm still waiting to see any kind of critique or response to that odd paper on salt intake I blogged a few weeks ago.)
Surveys like this are important to build the case for legislation. Politicians like to do good, and they like it even more if doing good doesn't lose them votes. So it is up to us to create and demonstrate popular support. That's pretty much how it happened for seat belt legislation in the 1970s.
These surveys show that even smokers think smoking should be banned in cars when children are in them.
Also note that some jurisdictions have already enacted this legislation, and when you do so support increases.
I have to start today with a word about the ghastly outbreak of E coli which has claimed many lives in Germany. I am sure the basic processes of outbreak control are in place - molecular epidemiology to distinguish patients with the outbreak strain from the background noise of ordinary E coli, plus case finding to ensure that we know the full extent of the outbreak in time (when was the very first case?) and time (is it really only Germany?). Thus we can avoid jumping to conclusions based on the first few known cases.
Usually, when we run a trial we need adequate statistical power, control groups, randomisation and so on. But sometimes, very occasionally, just a single patient is enough. This is one of those once-in-a-lifetime events. The intervention enabled a patient paraplegic from spinal cord injury to stand almost unaided. Unfortunately the report is almost incomprehensible but the basics are clear from the abstract and introduction, not to mention the videos.
Single patient studies are convincing when we know with certainty the natural history of the disease - in the case of spinal cord injury, permanent paralysis. We would also be convinced by any treatment which resulted in even a single patient recovering from rabies - a uniformly fatal disease once symptoms have developed.In the 1940s penicillin allowed recovery from what at the time were almost uniformly progressive staphylococcal infections.
So the hierarchy of evidence - case reports, case control studies, randomised trials - is appropriate 99.9% of the time but you also need to know when to lay it aside.