An “open” letter to the editor of the BMJ

Fiona Godlee


British Medical Journal


Dear Fiona

Imagine that you are about to board a long haul flight on a Sunday morning and you read in your copy of The Telegraph that the airline runs on a skeleton staff at the weekend and as a result your aeroplane is much more likely to crash.

Most people would cancel the trip, or rebook with a different carrier. The consequence for people with no option but to fly would be a very uncomfortable journey. The increase in stress and anxiety for nervous passengers could be very significant.

And the newspapers publishing this sort of allegation had better be confident that it is true. You can be sure the airline will use the full weight of the law to sue for reputational damage and loss of income if there is doubt about the veracity of the story.

Sadly emergency NHS patients do not have the option of delaying their treatment or changing provider in the face of similar news stories and indeed they could come to real harm if they try to do so. There is clear evidence that patient outcomes can be influenced by expectations. Increasing fear and anxiety in emergency patients could cause real distress and possibly lead to actual harm.

Which brings me on to the BMJ paper on hospital mortality rates published this weekend and the press brouhaha surrounding it.

What does this paper add to our current knowledge to justify the publicity surrounding its publication?

The paper shows an increased death rate for weekend admissions, which has been well known for many years. It is completely in line with previous findings – so no news there.

Similar results have been obtained in other health economies including the USA, Canada, France, Germany and Australia so this is not a story about the NHS per se.

It is very likely patients admitted at the weekend are sicker. The number of admissions on a Saturday and a Sunday is lower than the rest of the week and it seems probable that the missing emergencies are from the less severe end of the spectrum.

I disagree with the author’s assertion that the results have been fully corrected for disease severity – an assertion made in the paper and also in the BMJ press release. The results have been corrected for predictors of death such as comorbidity and social deprivation but these are just factors inherent in the patients. They are not indicative of the severity of the illness itself. We do not usually code for disease severity and the surrogate “admission urgency” does not really make up for that.

This is important because if disease severity had been corrected for then the differences would be entirely due to process issues.

The bottom line is that this paper provides relatively little new information about weekend death rates and or their underlying causes.

So what is the problem with the paper and the way it was released?

This study essentially just confirms and updates earlier findings. It does not tell us why there differences between weekend and weekday death rates or indeed if any of them are avoidable. A big launch with a press release and a media flourish is surely not justified by the scientific merit of the findings.

I believe the authors have overstated the importance of their findings and presented them in a sensationalised way. The BMJ has missed an opportunity to provide moderation here.

The authors seem to assume that death in hospital is always a bad thing. The vast majority of hospital deaths are unavoidable and many are expected. Patients with terminal illness may be more likely to be admitted to hospital for a crisis at the weekend when community services are thinly stretched for example. They may be quickly discharged to die in the community within 30 days but are still included in this analysis.

So the headline grabbing figure of 11,000 excess deaths is disingenuous. It compares the weekend with the best day and assumes that everything else is excess. At least the discredited “unnecessary deaths” claims in Mid Staffs were calculated using the national average – not by comparing Mid Staffs with the best performing trust in the country.

The 11,000 figure is then further inflated by extending the weekend to include Friday and Monday on the flimsy rationale that people are winding down on a Friday and overwhelmed on a Monday. This is in effect adjusting the analysis based on prejudice about the true cause of the difference. If you want to really examine the effect of staffing levels then Saturday and Sunday are definitely different to weekdays – but staffing on all weekdays is the same.

It is notable that there are big differences between death rates on Saturday and Sunday when in most trusts consultant staffing levels are identical – further undermining the argument that consultant presence is important.

It is reasonable for authors to speculate on the possible causes for their findings in the discussion and clearly medical staffing levels are a possible cause and should be mentioned. It is pushing things a little to add “a case for expanded 7 day services” to the article title though as the data does not support this unequivocally.

And now the awkward part

This is good, solid, helpful data which updates earlier findings but which really does not take our understanding of causes of weekend excess mortality very much further forward.

The major issue is the way that this paper was launched by the BMJ, with a press release and a fanfare of publicity, into the midst of a very high profile consultant contract negotiation and the debate of 7 day working. The press release talks nearly as much about this than it does about the merits of the paper.

It is clear that although these results were embargoed until Sunday morning the government had advance notice of the findings. Jeremy Hunt has already caused considerable anger amongst hospital consultants buy implying that excess hospital weekend deaths are caused by their work practices and the completely nonsensical assertion that removing the largely unused opt out from ELECTIVE work will somehow improve the situation.

I have argued that Mr Hunt’s approach – using weekend death statistics in order to strengthen his negotiating position – is unjustified and damaging to NHS patients and staff. It prompted me to call for his resignation and start the #weneedtotalkaboutjeremy campaign on social media which has resulted in the debate in parliament on Monday.

Playing political football with the NHS causes real harm and my worry is that launching this research into the current political maelstrom in the way it was done has been harmful. Politics and science do not mix well – and this is a great example of things going very badly.

What next?

There is no doubt that everyone – including consultants – want this country to have an NHS that provides effective and safe emergency care every day of the week. I think we have yet to agree as a country whether we want elective services at the weekend and if we do how we are going to pay for them.

It would be helpful if the government could make clear its top priorities to address the weekend mortality issue and indicate how they might be funded. They should do this in a transparent and dispassionate way rather than using the medical press and the media to spread half-truths and misinformation in an attempt to get political leverage.

The BMJ is in a unique position to influence the future of the NHS. I hope you will take these comments in the spirit they were intended – to stimulate sensible debate and defuse a difficult situation.

I am sending this as an “open” letter to raise awareness of the issues. I have chosen to publish it as a blog and not submitted to the BMJ for publication, as it will get a much broader readership this way.

If you would like to reply I will be very happy to publish your response on the same site.

Yours sincerely

Steve Smith



    1. Is this the full story? I work in theatres and don’t know of anyone who works there who send anyone to a ‘corridor’ post-operatively.


    2. I doubt this story is true. If it is true then please be brave enough to name names and places ‘cos to my knowledge it doesn’t happen in the NHS and a system broken this badly needs to be fixed. I would resign from a hospital that does this so it is important that everyone knows.


  1. You mention of mid staffs is disingenuous in fact mid staffs overall death rate was below average, and Francis only found one yes that’s a single episode that the death “might have been avoidable”, unfortunately the mass media sensationalised the story and instead of it being a story that showed the understaffing of front line staff, all backroom admin jobs were more than adequately staffed, leads to a reduction in the standards of care, it was used to take away desperately needed services at Stafford to pay off the massive PFI debt at Stoke which now has the worst and most dangerous a/e in the country because as we always stated it can not cope with the excess work that it took on at the behest of Jeremy hunt and the Mark Hackett the many times failed chief executive, along with the failure of the CCG’s another insanee government imposed level of management.


  2. Why not let Dear Jeremy make his changes and then ensure that he and all those sensationalists are properly held to accounts for their lack of understanding of a complex problem and are properly pilloried for their actions and views which provoked these actions.


  3. Excellent critique of a complex conflation of issues and the paper’s ‘conclusions,’ for what would appear to be largely a politically (or at least preconception-) driven imperative: measured, rational, logical, and based on facts – well done!

    One factor that I haven’t seen explicitly mentioned anywhere, although implied in your piece, is that even if one could accurately allow for disease severity and co-morbidities [which, as you point out, we can’t yet from the data available], there is still the massively important covariate of frailty, which isn’t yet routinely measured, despite being a widely accepted adverse prognostic factor. For example, it has been shown to predict mortality risk in ITU patients with the same illness severity score. [see ]


  4. Excellent overview Steve. We are at the point where NHS Policy is driven by the knee jerk reaction of the ill-informed. Based on a toxic mix of pseudoscience and misinterpretation of flimsy evidence, fuelled by media hysteria and tainted by puerile party politics the good ship NHS and all those who sail in her is now foundering terminally.

    I believe that similar accusations could be made in respect of the GMC – 17 doctors put through their processing plant daily in their pursuit of a media-inspired McCarthyist witch-hunt of doctors.Who can blame the profession for heading for the hills.


  5. While your wider point has a lot of merit, attacking the BMJ for publishing confirmatory studies is sadly a bit anti-science. Confirmatory studies are a crucial part of the scientific method which are under-published, under-appteciated, and under-funded – leading directly to harm as the failings of one-off studies go uncorrected.

    Criticism of this study and its methods is also an important part of the scientific method, but if you’re seriously suggesting that the BMJ should “hold back” a study which clears their bar of scientific robustness because it doesn’t fit a particular political narrative, then I’m afraid we disagree.


    1. I think you missed mt point. My issue was with the press release, the leak to the govenment press machine in advance, the overstatement of the risk and the media circus that followed. I am not against publishing confirmatory studies. I am against mixing science and politics. The BMJ should be above this.

      Liked by 1 person

  6. Excellent analysis. The original RSM paper showed death rates in hospitals at weekends were LOWER than on weekdays (even though for the obvious reasons stated by Steve Smith those admitted at weekends were more likely to die in the next 30 days). Without increased resources, putting more staff on at weekends will mean fewer staff working on weekdays so Tuesday and Wednesday will become the new Saturday and Sunday. Furthermore,continuity will be disrupted, recruitment made more difficult and morale dented. Locum agencies will flourish like never before and patients will get a worse deal. But no one will know, because the only metric available is a comparative one so if the mortality of patients admitted on Tuesday and Wednesday goes up to the same level as that of patients admitted on Saturday and Sunday the government will tell us how well they have done because there are no longer any peaks and troughs, – just troughs.
    Peter Trewby


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s