go to content

Are NHS Weekend Rotas Really Killing People?

People really are more likely to die if they're admitted to hospital on weekends, but the reasons for that are not as straightforward as Jeremy Hunt seems to think.

Posted on

Health secretary Jeremy Hunt claimed on Tuesday that people who have strokes are 20% more likely to die if they get admitted to hospital on a weekend.

Christopher Furlong / Getty Images

He told BBC Radio 4's World at One programme: "At the moment, for example, if you have a stroke at the weekend you are 20% more likely to die. That cannot be right, and that is something every doctor wants to sort out as well."

Hunt said it was a reason to support his changes to junior doctors contracts and to make the NHS a more "seven-day" service.

Doctors, many of whom were striking yesterday over Hunt's NHS plans, have responded angrily. One wrote a blog dissecting the stats of the claim.

Christopher Furlong / Getty Images

Prof Dave Curtis of University College London pointed out that according to one study, fewer people were admitted for stroke at weekends – 88 people per 100,000, instead of 111 per 100,000 during the rest of the week.

He said this suggests that the people being admitted at the weekend are, on average, sicker – that some people with less severe symptoms are not coming to hospital. That would mean that the stats would look worse, because the average stroke patient coming in at a weekend would be more likely to die. He estimated, looking at the numbers, that the risks were about equal. The study itself concluded that the increased mortality “may be influenced by a higher stroke severity threshold for admission on weekends”.

It’s not clear whether this study was the one Hunt was referring to on World at One. On Wednesday morning the health secretary tweeted a link to an NHS report on weekend coverage that also claims a 20% increase in risk, so he may have been referring to that. It bases its claim on an unpublished analysis carried out in 2011, though, so it is impossible to judge how reliable it is.

Hunt has made similar claims before. He once said there were 11,000 "excess" deaths per year due to NHS hospitals being understaffed at weekends, but the article he based it on said it would be "rash and misleading" to draw that conclusion.

Christopher Furlong / Getty Images

He blamed the deaths on junior doctors contracts, which include clauses for increased pay for weekend shifts. "What we do need to change are the excessive overtime rates that are paid at weekends," he said. "They give hospitals a disincentive to roster as many doctors as they need at weekends, and that leads to those 11,000 excessive deaths."

The 11,000 claim came from research published in the British Medical Journal and led by epidemiologist Nick Freemantle.

Freemantle's research – an update of a similar work he did in 2012 – looked at patients' chances of dying within 30 days of being admitted to hospital, and how that changed depending on which day of the week they were admitted.

That article did find that patients were more likely to die if they were admitted at weekends. But it wasn't as simple as Hunt made out.

Christopher Furlong / Getty Images

The authors of the research pointed out that patients who are admitted to hospital at weekends tend to be emergency cases, and are therefore often more gravely ill than those admitted during the rest of the week. The paper said: "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."

Dr Fiona Godlee, the editor of the BMJ, wrote a letter to Hunt accusing him of "publicly misrepresenting" the article and "misusing data to mislead the public". Andrew Dilnott, the head of the UK Standards Authority, also rebuked Hunt, using gentler language.

The paper did say, however, that a definite "weekend effect" – probably partly due to a difference in staffing at weekends – was "unignorable", and recommended a more comprehensive seven-day NHS service.

"Appropriate support services in hospitals are usually reduced from late Friday through the weekend, leading to disruption on Monday morning," it said. "This could go some way towards explaining our finding of a 'weekend effect' extending into Friday and Monday." The weekend effect was visible even after the authors adjusted for the fact that people tend to be more ill if they come to hospital at the weekend.

It is definitely worth noting, however, that according to the figures, although you're more likely to die if you're admitted at a weekend, if you're already in hospital over the weekend, you're actually less likely to die.

Hunt has said "eight studies in the last five years" have shown a weekend effect.

He didn't say which ones, but the Department of Health lists eight studies on its website, including the Freemantle research.

However, one of the eight is the earlier, 2012 version of the Freemantle study – not a separate study – and four of them are in-house reports by medical bodies, rather than studies published in peer-reviewed journals. That's not to say they're less rigorous, but it's harder to judge their value.

The other three articles did find a weekend effect.

They are Aylin et al (2010), Freemantle et al (2015), and Ruiz et al (2015). All of them found that even after controlling for other factors, people were more likely to die if they were admitted at weekends.

There are a few complicating factors though. First, the studies themselves don't all agree with Hunt's analysis that NHS rotas are to blame.

Ben Pruchnie / Getty Images

Freemantle et al explicitly linked the "weekend effect" to a lack of staffing at weekends, although they were careful to say they couldn't be sure. Aylis et al, similarly, said the excess deaths "may reflect differences in quality of care", but said more research was needed.

But Ruiz et al looked at hospitals in Australia, the Netherlands, the USA, and England, and found a weekend effect in all of them – suggesting, at the very least, that this is not an NHS-specific problem. They also said that because of the way their study was carried out, it couldn't tell us what actually caused those weekend deaths.

Second, and perhaps more important, the Department of Health has only presented studies that support its case – but they are not the only studies that exist.

For instance, this study in the journal Health Economics found that while the weekend effect is real, the proposed changes to fix it would be counterproductive. It said a seven-day service would cost between £1.07 billion and £1.43 billion, but would only prevent harm to patients that the NHS would usually spend no more than £731 million to avoid.

Every pound spent on one service in the NHS is not being spent somewhere else, so, the authors argue, the seven-day service would cost lives, by spending NHS money inefficiently.

Third, it's always worth talking about risks in absolute, rather than relative, terms.

Christopher Furlong / Getty Images

Hunt, for instance, says that the risk of death from stroke is 20% higher at weekends. Whether that's true or not – and, as discussed above, it's hard to tell – the figure doesn't mean much on its own.

What it means in real terms is that on weekdays, your risk of dying after being admitted to hospital for a stroke is 11.1%; if you are admitted on a weekend, it's 12.9%. That's obviously worse, but it might not sound as scary.

There are similar stories to be told elsewhere in the data. Aylis et al found that emergency admissions during the week had a 4.9% mortality rate; at weekends, it was 5.2%. Freemantle et al found a total 30-day mortality rate of 1.8%, so the weekend-admissions mortality rate would be somewhere around 2%. These are real increases, but it's important to put them in context.

None of this is to say that the weekend effect isn't real.

But it might be that it's more complicated, harder to fix, and perhaps less dramatic than Jeremy Hunt has been suggesting.

Tom Chivers is a science writer for BuzzFeed and is based in London.

Contact Tom Chivers at tom.chivers@buzzfeed.com.

Got a confidential tip? Submit it here.