The UK’s Secretary of State for Health, Jeremy Hunt, spent part of last week annoying those working in the NHS. He quoted a couple of numbers in support of his contention that the contracts of consultants should change. One was from a 2012 paper by Fremantle et al. in the Journal of the Royal Society of Medicine. In his speech he said: “You are 15% more likely to die if you are admitted on a Sunday compared to being admitted on a Wednesday.” Compare this with: “Thus for every 100 deaths among patients admitted on a Wednesday, we would expect 116 among otherwise similar patients admitted on a Sunday”, from the discussion section of Fremantle et al.. The number refers to deaths within 30 days of admission to hospital. So Hunt’s sentence is a fair reflection of something Freemantle et al. said.
But Freemantle and coworkers said more. Their very next sentence is: “However, for every 100 deaths among patients in hospital on Wednesday we would expect to see 92 among similar patients already in hospital on a Sunday.” Intriguingly, in our NHS hospitals which is where according to Jeremy Hunt, “we do not have a proper 7-day service”, patients are 8% less likely to die on a weekend than on a week day. Taking that statistic at face value, and in isolation, we could reduce mortality by 8% by running the NHS Monday through Friday, exactly as we currently run it on Saturday and Sunday.
Of course, this just shows that you can take a scientific paper that analyses a complex problem with many factors, and cherry pick statistics from this to support pretty much any position you want, if you are prepared to select single statistics and use them in isolation. Such a paper will have many stats in it, and the odds are good that one taken in isolation will apparently support something you have already decided to argue for.
But if you want to work out why a higher fraction of patients admitted on a weekend die within 30 days and use this to improve the NHS, you need to analyse the data more carefully*. You should also, as Freemantle and coworkers do, know when to admit that you don’t know. They put forward a couple of speculative suggestions for why the mortality rate is higher, one of which is fewer senior doctors on hand. But they also point to some findings that suggest that despite their efforts to control for this, patients admitted on a weekend are different from those admitted to hospital on a weekday, in a way they do not have a good grasp on. Clearly, if patients admitted on a weekend are on average sicker than those admitted on week days, then renegotiating contracts will not address the difference in mortality rates.
* Freemantle et al. also analyse data for a large set of US hospitals and find the same result, so it looks like whatever is causing this is not unique to the NHS.
Chance & Necessity 