In today’s media (28th July) it has been widely reported that a team of academics from the London School of Economics have done research showing that ‘competition saves lives’. Quoted in the telegraph, the lead author Zack Cooper says ‘“We find that competition between hospitals in a market with fixed prices led to very strong improvements in patient outcomes. Going forward, if the NHS is going to be successful, it can’t shy away from competition.”
Wouldn’t it be great if this were the case? If we’d found a simple solution to drive improvements in the NHS and save lives? We’ve been looking for over 60 years now to find the best way of organising things. And there is was all the time – competition. And we missed it. Sadly though, the research doesn’t show this.
What Cooper’s team have shown is that people who live in urban areas have better 30-day outcomes from Acute Myocardial Infarction (AMI) than those who live more rurally. That’s really about it. So how does he make the claim that he does?
First, the work presumes that the greater the density of provision, the more competition there is. That simply isn’t the case. The more dense the provision the more trained staff services can draw from, the more likely it is areas will include substantial medical training facilities, the more specialist services are likely to be. Of course, it may well also be that services are simply collaborating better to get people with AMI to those who can care for them more quickly. Who knows?
We don’t, because the research doesn’t really check that there has been an increase in competition since 2006. It assumes that the difference in improvement between urban and rural outcomes it finds is down to the fact that, at the beginning of the period under study, the government tried to launch its extending patient choice policy. The research didn’t check that this had led to more choice or more competition (in face, the National Patient Choice survey suggests that neither was the case).
Next we have to ask if 30 day AMI is a good proxy for overall care improvement, and that differences between hospitals are down to patient choice rather than treatment choices. Why would either be the case? One measure of clinical care does not reflect what is going on in an entire hospital, and there is high quality clinical research from the US suggesting that AMI mortality differences are about differences in treatment regimes. To prove their point, the LSE team would have to show that treatment regimes are changing as a result of choice and competition. But again, they don’t. There are other obvious problems with the data, but I hope to go into those in other work I’ll publish later.
Finally, there are confounding factors. There were treatment changes in AMI that were more likely to feed through to urban areas during the period, as they were more likely to have specialist care. No attempt was made to disaggregate those from the data.
So if the paper is so flawed, how come it is being published in the Economic Journal? I can only guess that it has got through peer review because it provides the ideologically correct answer (markets save lives) and that its regressions have been done correctly. However, as the economic historian Diedre McCloskey reminds us, there is a difference between statistical and substantive significance – just because you’ve found what looks like a statistical relationship it doesn’t mean you’ve found something important. In other terms, there is a difference between correlation and causation. I’m afraid this work looks rather more like the first than the second.