Increased mortality associated with weekend hospital admission: a case for expanded seven day services?

Sir Bruce Keogh's belated response persists in describing the 'weekend effect' in blinkered terms, and persists in claiming it relevant to plans for seven-day working.

For example, his statement "whether or not the finding of increased mortality associated with weekend admission published in 2012 and derived from 2009/10 data still existed." fails yet again to insert the word "rate" after "mortality". He completely fails to recognise or address the fact that mortality ( total deaths) are fewer following admission on any weekend day than after admission on any week-day. If he achieves his aim to re-distribute NHS service ( routine and emergency) equally throughout the seven days, then it seems more likely from all the critique from respondents, that he will merely redistribute the consequent deaths, WITHOUT PREVENTING ANY DEATHS. Unless he knows something we don't ? His hypothesis presumes that equalised seven-day working will lead to a reduced absolute number of deaths, without specifying which services are changed, how and at what cost.

Following the hypothesis-generation of this observational study, the next step should be a local pilot, ie a controlled study of seven-day versus usual-working, to prove the hypothesis, and establish the costs, benefits, and harms.

Instead, we have presumption, wholesale reform, and debacle.

Mr Dean raises concerns about the governance and provenance of our work on weekend mortality,[1] however, documentation available to him in the public domain [2] clearly demonstrates these concerns are unfounded. The HSCIC license referred to by Mr Dean can be seen in row 3 of the relevant document published by HSCIC.[2]

This license refers to work for the replication of Summary Hospital Mortality Indicators (as clearly specified in the document). University Hospital Birmingham are part of the SHMI working group which was set up by the then Secretary of State to facilitate methodological work on that metric. We might note that in row 4[2] there is an identical license given for the same purpose to another Institution. For the analysis of hospital mortality outcomes by days of the week, we did not and could not use the SHMI Group license in row 3.[2]

A careful look at the same documents would have identified several licenses (rows 330-334)[2] for data release to University Hospital Birmingham, which do not have any reference to the Secretary of State. The analysis for mortality by days of the week was conducted under the governance umbrella provided by the latter licenses.

For the avoidance of doubt, our work published in the BMJ [1] was not conceived, executed, or published in collaboration with the Secretary of State for Health.

References

1. Freemantle N, Ray D, Mcnulty D, Rosser D, Bennett S, Keogh BE, Pagano D. Increased mortality associated with week-end hospital admission: a case for expanded seven-day services? BMJ, 2015; 351: h4596
2. http://www.hscic.gov.uk/media/17946/Data-Release-Register-January-to-May... [accessed 19/2/16]

I have previously provided evidence that suggests the Secretary of State had access to this paper’s pre-publication and un-peer reviewed data for his King’s Fund Speech on 16/07/2015 [1]. The authors of this paper have not yet commented on this. When you ask the Department of Health (DH) how Jeremy Hunt got his 6,000 excess deaths figure for this speech, you are directed to this DH webpage [2, Webpage 1, Webpage 2]. This webpage and attached document claims that the 6,000 excess deaths figure was derived from the Freemantle 2012 JRSM paper [3].

I believe this webpage which claims that Jeremy Hunt’s 6,000 deaths data was derived from the 2012 Freemantle paper to be incorrect for two reasons:

Firstly, Jeremy Hunt used the 15% increase in Sunday mortality noted from the 2015 study, not the 16% increase in Sunday mortality noted from the 2012 Freemantle study [see 1].

Secondly, when you submit a Freedom of Information request to the DH about all the correspondence involved in creating that webpage, you get an interesting set of emails between NHS England and the DH. These emails confirm that the 6,000 estimate came from medical research expected to be ‘published by the researchers in the British Medical Journal in due course’ [Email 1].

However because the DH author wants to ‘avoid undue criticism of either DH or of NHS England’, they ask NHS England whether (since the research is unpublished) they can help answer the question posed by the UK Statistics Agency as to 'how best to supplement the information currently available’ so the public can understand the figure.

NHS England dutifully respond with a one-page suggestion of how the figure of 6000 ‘can be arrived at by combining the relative risks from the published 2012 paper with HES data for 2009/10’ [Email 2]. This information was seen as something that could have been ‘as readily generated by DH as NHSE’ [Email 2]. They went on to say that if the ‘solution appeals’ then they should feel free to take it forward. The attached one-page solution [Email 3] obviously appealed. It looks very similar to the explanation that can be found on the DH webpage.

This alleged collaboration between NHS England and the DH to obscure the source of Jeremy Hunt’s data for the King’s Fund speech is very worrying, as (amongst many other reasons) it is an undeclared collaboration before the paper was published. Please can the authors urgently clarify that they were not in any way involved or aware of this?

Jonathan P Sturgeon
Paediatrics Trainee
South London

(1) 'No copy of the paper was shared with the Department of Health by ‘any of the authors’ in advance of publication’ , Rapid Response, 3rd February, Sturgeon J, Dean B. http://www.bmj.com/content/351/bmj.h4596/rr-59
(2) https://www.gov.uk/government/publications/higher-risk-of-death-associat...
(3) Freemantle N, Richardson M, Wood J, Ray D, Khosla S, Shahian D, Roche WR, Stephens I, Keogh B, Pagano D. Weekend hospitalization and additional risk of death: An analysis of inpatient data. Journal of the Royal Society of Medicine, 2012; 105: 74-84.

The paper by Freemantle et al should have been titled “Increased mortality RATE associated with weekend hospital admission..”

The HiSLAC collaborators meeting referred to by Dominic Pimenta, referenced below , gives a very straightforward and revealing table of raw data of the sort requested by many respondents to Freemantle et al. The crucial difference between actual number of deaths, and death rate, is revealed.

The authors have confirmed in Rapid Responses that their ‘weekend effect’ persists amongst ‘emergency’ admissions after the ‘routine’ admissions are removed.
If I now make the reasonable assumption that the population has a propensity to die, and that admission as an emergency in large part reflects perception of this risk, the ‘weekend effect’ can now be readily explained. The data table shows that fewer people are admitted as emergencies at weekend. Freemantle et al agree that “fewer hospital admissions occur at the weekend, patients admitted on Saturday and Sunday are sicker and face an increased likelihood of death within 30 days even when severity of illness is taken into account”. Given that admission is discouraged, and resisted, could it be that those less likely to die do not get admitted on weekends ? Thus, in absolute terms, there are fewer deaths at weekends, but the rate amongst those admitted is higher. Triage works.

Elucidation of risk and response would be well served a prospective trial of current configuration, versus easy admission of all at risk. Who will fund it, with so few doctors available ?

Reference

http://hislac.org/images/docs/HiSLAC_Collab_Meet_13th_April_2014_session...