“We now have seven independent studies showing mortality is higher for patients admitted at weekends.”
JEREMY HUNT, ALL THE TIME.
You can view these seven (or rather eight) ‘studies’ here:
On the basis of this evidence Jeremy Hunt and the Department of Health have put forward the argument for sweeping changes to the NHS to create ‘Seven-Day’ services.
Juniordoctorblog deconstructs the Hateful Eight.
DISCLAIMER: this is written for a lay person. Further details on all the papers available on request.
1. Increased mortality associated with weekend hospital admission: a case for expanded 7 day services? by Freemantle and Sir Bruce Keogh, published in the BMJ in 2015.
This is the most recent and most quoted paper, and where the soundbites “11,000 excess deaths” and “16% increased probability of death” come from. The study was performed by a group of researchers which included Sir Bruce Keogh, and was commissioned on his request, which makes the claim “independent” rather dubious. The study was an update of a 2012 paper (see below) and therefore 2 of the Hateful Eight are actually the same paper for different years.
This study pulled numbers from Hospital Episode Statistics, which records patient information from the discharge summaries written by junior doctors when you are discharged from hospital. If you have ever been to hospital you would know this is not always 100% accurate. The study identified the day of admission for every patient admitted to hospital in 2013/4, and then counted how many patients had died at 30 days after admission.
Overall just over 1.5 in 100 patients died in the study. They found patient deaths were LOWEST on Sunday, and HIGHEST on Wednesday, but for those ADMITTED on a Sunday or a Saturday they found a small increase in the risk of death at 30 days, an absolute increased risk of 0.07%* for admissions between Friday and Monday, compared to those admitted on a Wednesday.
The study also found 1/3 of patients died after being discharged from hospital, and the majority died after 7-8 days in hospital. For the first time the study tried to work out how sick patients were and found a higher proportion of the very sickest patients were entering the hospital on Saturday and Sunday compared to the weekdays. The authors conclude ‘to assume these excess deaths are avoidable would be rash and misleading’. At no point did this study measure staffing levels, rota cover or hospital resources, and the figure “11,000 excess deaths” is a statistical guess based on the numbers the study cranked out – they are NOT real identifiable cases.
BOTTOM LINE: Patients admitted at weekends are sicker, and they have a very tiny increased risk of death compared to the weekday admissions. “To assume this is avoidable is rash and misleading.”
2. Weekend hospitalisation and additional risk of death: an analysis of inpatient data by Freemantle/Sir Bruce Keogh published in 2012
This was the original paper as described above, by the same group from the same data using broadly the same methods. The only thing to add for this paper is it actually found patients in hospital on a Sunday were 8% less likely to die than those on a Wednesday.
BOTTOM LINE: 2 papers from the ‘Eight’ are written by Bruce Keogh of NHS England and are actually the same paper repeated.
3. The Global Comparators Project: international comparison of 30 day in-hospital mortality by day of the week by Ruiz, published in BMJ Quality and Safety 2015
The authors for this paper work for the Dr Foster Unit, sponsored by Dr Foster Intelligence: a former Department of Health co-owned patient safety monitoring company. They looked at the same data as the above from the Hospital Episode Statistics warehouse, and compared this to other countries: USA, Australia, the Netherlands and several more. This study looked at emergencies and routine surgery only for 2.8 million patients, 1.3 million of which came from the UK. For surgery, the UK had the lowest risk of death at 30 days. Emergency admissions were sicker than planned admissions. The results were similar for all countries studied, suggesting that this is an international phenomenon. UK planned surgery patients who had procedures on a Sunday, before adjustment**, were 0.7% more likely to die than those on a Monday. For emergency admissions the risk was 0.4% higher on a Sunday compared with a Monday. The effect was seen in nearly every country. Again this study performed no measurement of staffing levels on each day and the authors conclude themselves “we are not able to determine the reason for these findings.”
BOTTOM LINE: The ‘weekend effect’ is seen across the world in varying health systems.
Now is a good time to pause and discuss mortality. Imagine if you will two hospitals. Hospital A has a 90% mortality rate at 30 days – 90 in 100 people die within 30 days of admission, while at Hospital B the rate is only 2%, or only 2 in 100. Which would you rather be treated at? On the face of it, the answer would be Hospital B, because the obvious logic is: all illness should be curable, therefore I go to hospital to get better, therefore I choose the hospital where I have the greatest likelihood of getting better, ie not dying. Which makes sense: except if I told you Hospital A is a hospice, for end-of-life terminal cancer patients, and Hospital B is a community minor treatment unit for children, for scrapes and bruises and runny noses. Now this changes your perception of the figures: Hospital A has a surprisingly low mortality rate, considering everyone admitted is there to die peacefully, and Hospital B has a worringly high rate – considering no one should be dying at all. Now what if I told you Hospital C had a 1% chance of death for a procedure, and Hospital D had a 1.1% chance? Would you be bothered which hospital you went to? I wouldn’t. But if I told you that Hospital D had a 10% higher probability of death than Hospital C, you might change your mind. This illustrates the problem with superficially accepting statistics and why it’s so important to properly scrutinise the figures. Anyway, back to the papers.
This is not a scientific report at all, but a consulting report from ATOS. The same ATOS that the Department for Work and Pensions recently dropped for the ‘poor quality of their work’. The report is from a group of executives from the East Midlands. It’s really dull, and not scientific at all – all of the numbers come from the other ‘studies’ here in regards to weekend and weekday working. Of 10 clinical standards for ‘seven-day’ services it found all were already being met 50-60% of the time. The biggest fail areas were ‘mental health’ and ‘transfer, discharge to social care’. Both budgets of which have been cut in the last ten years. However, here are some favourite quotes
“It is likely unsustainable and unnecessary for all trusts to provide all services 7 days a week”.
“There may be a need to drive funding for the whole system to deliver 7 day services”.
Here is a good time to remind readers the last eight years have been the worst funded decade for the NHS in its history (including the recently announced ‘extra’ £3 billion). Again no measure of staffing levels and no mention of junior doctors.
BOTTOM LINE: 7 day a week routine services require proper funding and are not necessary or sustainable in all areas. A good proportion of 7-day emergency services are already available.
This is a policy document from NHS England and, again, Sir Bruce Keogh’s office. Also, again, not a scientific ‘study’ at all. Interestingly the focus is nearly entirely on emergency services – no mention of ‘routine’ care at all. The review notes that doctors and nurses are present on the acute medical unit 100% of the time weekday or weekend, the importance of diagnostic services being available 24/7, and lots of case studies- all of which achieved better cover without changing work conditions for staff. Interestingly in the annex it notes that many more weekend admissions are end-of-life patients compared to the weekdays- suggesting an increased need for community hospice and palliative care services.
BOTTOM LINE: Bruce Keogh and friends re-hash other research in this list- but importantly define the need for seven-day services as emergency care improvements, not routine services.
Again, not a unique scientific study but a review of many other studies. Produced by the Academy of Medical Royal Colleges to look into the necessity and feasibility of increasing consultant presence on the wards for emergency unscheduled patients. Again, not routine services and again, nothing about junior doctors or staffing levels.
BOTTOM LINE: Consultant presence is important for emergency admissions, not routine services.
Here is an ACTUAL scientific study, another from the Dr Foster Unit at Imperial College London (which was 50% part owned by the Dept of Health at the time of writing). This is the fourth study in this list that uses the Hospital Episode Statistics warehouse: again discharge letter information. This paper focused only on emergencies. They reached the same conclusion as the papers above, with an absolute increased risk of death at the weekend vs the weekday to be 0.12%*. They didn’t take into account how sick patients were, or their method of admission, and again no explicit measure of staffing levels were made.
BOTTOM LINE: A fourth study from the same data, showing a very small increased risk of death in weekend vs weekday emergency admissions, and no accounting for how sick patients were or staffing levels.
8. Time for training Report by Professor Sir John Temple from the Department of Health published in 2010.
Unfortunately this the original report has disappeared but in summary this was another policy document from the Dept of Health looking at the issue of training doctors under the European Working Time Directive. It’s main conclusions was that shift work is anti-social and has had an impact on training, and that consultants should be more involved in 24/7 work to support trainees.
BOTTOM LINE: Another non-study, suggesting a larger consultant presence day-to-day would help training. Nothing to do with the ‘weekend effect’.
I’d be remiss for not mentioning the latest papers in the ‘weekend effect’ argument, which haven’t quite made it onto the gov.uk website yet but are already in the briefs and interviews of Mr Hunt and the spin machine.
A fifth paper looking at Hospital Episode Statistics, and the third from the Dr Foster Unit. It is remarkable actually that no single paper has tried to analyse ‘the weekend effect’ in any other way than use the same source. This group tried to identify a weekend effect on seven different measurements associated with giving birth. Overall the stillbirth rate was 0.7%, or 7 in 1000. It actually finds that the stillbirth rate is significantly lower on Monday and Tuesday, which had
‘no association with staffing’.
BOTTOM LINE: No link between mortality and staffing, and no obvious ‘weekend effect’ (Thursday had the highest rate of perinatal mortality.)
This study unsurprisingly also used the Hospital Episode Statistics database, looking at all emergency admissions undergoing surgical procedures or admitted with pancreatitis over five years. The study then cross-referenced these numbers with data about the hospitals it was collected from – e.g. staffing levels, number of beds etc. The methodology in this paper was actually quite good, and they show a very strong association with the number of doctors, nurses and beds and the association with better surgical outcomes- of course this does generally reflect the amount of money a hospital has, and how well-resourced it is overall. The weekend data shows the same bump in mortality at the weekend as all the other studies that looked at the HES data, but didn’t measure weekend vs weekday staffing levels, as many media stories wrongly reported.
BOTTOM LINE: Increasing resources improves outcomes from emergency surgery, regardless of the day of the week.
UPDATE: Following the strike announcement Jeremy Hunt began quoting ‘you are 20% more likely to die from a stroke at the weekend’. Given how stupendously dangerous delaying presentation to hospital is for a stroke I’ve updated this post to add in the following; (Full credit to Prof David Curtis and Ben White for drawing this to public attention.)
This was a study that took the Welsh equivalent Hospital Episode Statistics and looked specifically at patients who were admitted to hospital for a stroke between 2004-2012 and counted how many died at 7 days, 30 days and 1 year. They found less patients were admitted on a weekend for strokes (88) and patients admitted during the week (111), and a small increase in mortality of 1.8% at 7 days between weekend and weekday admissions. There are three really important things to say about this:
- 1) The study notes – stroke mortality fell by 3.1% every year for the 8 years of the study. This reflects the radical improvement in stroke care that has occurred over the last twenty years with the introduction of ’emergency’ stroke pathways and hyperacute stroke units. Here is a nice graph. This study doesn’t really factor in this massive improvement in overall care, and isn’t relevant to today. Also – this improvement was done without changing working conditions for staff.
- 2) Stroke occurs on any day with equal frequency except mondays– where it is slightly higher. Stroke can range from transient weakness or loss of vision which resolves after 24 hours, to permanent loss of power to limbs and face and even death. The authors note that the effect ‘may be influenced by a higher stroke severity threshold for admission on weekends’. If you look at day of stroke, regardless of admission, there is NO WEEKEND EFFECT, as seen here in a study from Japan.
- 3) Stroke is defined as ‘maximal at onset’ – it represents sudden and complete blood loss to an area of the brain. There is only one main treatment, which is to give clot-busting medication. However- this is very dangerous and the list of situations where the risk of the treatment outweighs the benefit is very long. Having worked on-call in a stroke unit and ICU previously I have only seen one patient who met the criteria. Only 15% of strokes were treated this way in 2014. 60% of patients came to hospital too late for treatment. Stroke is now treated as an emergency – the ambulance calls a stroke-centre hospital before the patient arrives, and a specalist team sees the patient as soon as they come in to the door. The limiting factor now is when the patient dials 999.
- 4) Lastly, a recent study found the presence of a consultant or doctor had no effect on a patients survival after stroke, whatever day of the week they were admitted. However, the presence of adequate nurses had a huge impact: increasing nursing numbers from 1.5 nurses/10 beds to 3 nurses/10 beds reduced mortality by 4%. This reflects the fact that stroke patients are very vulnerable in the immediate period after the event, and it’s good nursing care, not junior doctors, that directly influence this. However – Jeremy Hunt has so far suppressed the NICE recommended safe staffing levels for nurses- and the NHS student bursary to incentivise nursing training has been cut.
BOTTOM LINE: This study took place 12 years ago, in Wales, during a time of rapid improvement in stroke care overall. It shows a reduced number of strokes admitted on the weekend – and likely increased severity of those admissions resulting in a small 1.8% bump in mortality overall. Jeremy Hunt’s scaremongering has previously led patients to delay coming to hospital – in this particular case this could lead to devastating loss of function and even life. Time is the single biggest factor in survival in stroke, and has nothing to do with weekend doctor staffing or junior contracts.
Now time to look at things differently. You hear a lot about the studies showing a ‘weekend effect’. But did you know there are many studies that show no effect? The fact that you don’t is an example of something called publication bias – the government only wishes you to think the ‘body’ of evidence all points one way. It doesn’t.
Here are some studies that show NO WEEKEND EFFECT.
Weekend mortality in paediatric patients in Scotland – published by the Royal College of Paediatrics, Turner 2015. 
Byun 2012 (small study compared to the others) 
Kazley 2010 (US study) 
Kevin 2010 (Canada) 
Myers 2009 (USA) 
- Of the ‘Hateful Eight’ studies only four represent actual research
- Two are the same paper and co-authored by Sir Bruce Keogh,
- The other two are from Dr Foster, formerly owned by the Dept of Health.
- All of the studies come from a single source of data.
- None of them show any link to staffing levels, and none of them show any link to junior doctors working patterns.
- Much research exists disputing the weekend effect
- Research shows that increasing resources improves outcomes. Which is obvious.
And here is the pièce de résistance. When there is a finite amount of money the logical management of resource is to put money where it will do the most good. The National Institute for Health and Care Excellence, NICE, have a recommended money spent vs benefit formula for approving treatments. The cut off is currently about £20,000 to buy a year of quality life. This is how all new medications are decided if they are value for money or not.
Meacock in 2015 sat down and worked out the cost of a ‘seven day NHS’ and then tried to work out if NICE would approve if it were a medicine. Needless to say the money spent (estimated for emergency services to be £1-1.4 billion) is 2x-3x as much as the ‘recommended’ cut off.
BOTTOM LINE: This isn’t even good value for money.
Finally – some context. Every year in the UK 25,000 people will die of a blood clot to the lungs, 60,000 people will die of a heart attack, 30,000 people will die from chronic lung diseases, mostly smoking related. Improving research and treatment pathways for any of those conditions would save more lives than this endless politically driven ‘seven day’ debacle. I dread to think how much money has already been spent on the ‘seven-day’ services problem – but if it is real, it is a tiny relative problem and a problem no country anywhere has been able to solve.
All doctors would want to have the entire gamut of services on hand every day of the week – but the first lesson of practicing medicine is learning to prioritise. So far, the ‘studies’, simply don’t add anything useful to the debate – we need to know where and how to spend our money, whether that’s in the community, in social care, in improving hospice care, or in expanding emergency departments or increasing perioperative care. The list goes on. It’s not clear there is a truly avoidable ‘weekend effect’, but more importantly it’s not clear if it’s worth the vast amounts of money, damaging publicity, time and general consternation being spent on it.
This is a classic situation of political meddling in the NHS creating harm. We have a government and media who prefer soundbites to sound decision-making and spin doctors to actual doctors. THIS is the true threat to the safety of patients.