The NHS: when statistics and politics collide

There are two key pieces of evidence to consider with regard to Jeremy Hunt’s recent handling of NHS management:

1) The original paper, which can be found here.

2) Jeremy Hunt believes in homeopathy. The man ultimately in charge of the NHS believes in homeopathy. (!)*

Combine the two and ask yourself if Jeremy Hunt has read the paper let alone understood it. There is a reason why people spend lives studying statistics. It isn’t straightforward. Let us consider the primary conclusion as stated in that paper:

“Patients admitted at the weekend are more likely to be in the highest category of risk of death.”

That is the first sentence of the conclusion. Consider this alongside the fact that the two days with the lowest number of deaths are… Saturday and Sunday. It is simple to understand why total number of deaths is not the most important measure. Do you think Jeremy Hunt would understand why?

A GCSE maths student might have heard the well worn truism ‘correlation doesn’t imply causation’. It seems self evident to me that mortality rates (the likelihood of dying within thirty days of admission but not within three days (as a toy problem ask yourself why the authors of the paper chose that seemingly bizarre measure)) will be dependent on patient behavior surrounding the causes of admission. To what end do we move elective surgeries from weekdays to weekends? If the aim is not to improve mortality rates for a given type of hospital admission (i.e. for a given patient, their likelihood of a positive outcome) then we have entered in to a world of truly Carrollian logic.

My central criticism of the paper is that removing those deaths within three days does not as they claim ‘confirm the robustness of the model.’ Rather it is arbitrary and statistically simplistic.

The second conclusion of the paper is:

“Patients admitted on Saturday or Sunday face an increased likelihood of death even when severity of illness is accounted for.”

That is of course more pointed, but ask yourself how the authors determined severity of illness. How would you do that without measuring outcome? Do you not accept as the ultimate measure of severity of illness that of likelihood of death? Alternatively, it is not a measure of severity of illness, it is severity of illness.

Crucially the author’s go on to state quite plainly:

“It is not possible to ascertain the extent to which these excess deaths may be preventable; to assume that they are avoidable would be rash and misleading. From an epidemiological perspective, however, this statistic is ‘not otherwise ignorable’ as a source of information on risk of death and it raises challenging questions about reduced service provision at weekends. Similar to our previous analysis, we have found that patients already in hospital over the weekend do not have an increased risk of death.”

My summary of this statement would be ‘its complicated, further study needed.’ Have you ever read an academic paper that didn’t have that conclusion? One final point about academic papers is that people churn them out. They are professionally obliged to publish as much as they can. Consider the volume of literature generated by the medical profession and wonder why one paper with vague conclusions surrounding widely available data has received so much attention. The attempt to fix a problem which is not understood by changing doctors contracts is therefore wilfully simplistic to the extent that one must question what other aims are being followed. Is there any meaning at all in the phrase ‘a truly 7 day NHS?’ It clearly doesn’t mean a 7 day NHS because that already exists. This whole debacle represents a new low in government debate.

Finally, and most contentiously, let us truthfully ask ourselves what the motives of government are. If you wished to dismantle the NHS would you perhaps intentionally instigate a strike as a means to devalue a profession and turn public opinion against it. The British people hate strikers and the government knows that. The doctors have made the crucial error of trying to argue rationally with politicians. They are bound to fail. This might explain their desperate lashing out, but only allows us to criticise their tactics not their argument. This final paragraph betray’s my personal prejudice but don’t lose your intellectual rigour and let it prejudice you against the principle arguments above.


*An exclamation point doesn’t seem enough.

As a further note regarding the paper, I wonder if the authors are broadly Bayesians or frequentists. At the risk of being offensive to doctors and biologists everywhere (one can only hope), perhaps you should have asked a mathematician to do this research.