I shall be going into the Cumberland Infirmary tomorrow afternoon for a minor dental operation.
The fact that I am having to go to Carlisle for this procedure is an indication of the number of services which are no longer available at the West Cumberland hospital in Whitehaven, which is a very sore point in West Cumbria, but that is not the issue this post is about.
My operation tomorrow is not a particularly serious procedure, which is just as well, as the British Medical Journal has just published the results of a study which shows that patients are 44% more likely to die within 30 days of an operation performed on a Friday in an NHS hospital than within 30 days of one performed on Monday.
Knowing that the many journalists are - how can we put this - not particularly well versed in statistics, my first thought was to check how likely it is that this result is statistically significant.
However, when one checks out the details of the study, it rapidly becomes clear that the study was large enough and sophisticated enough that its' results are indeed statistically significant - the details and significance have been published on the British Medical Journal website.
The "p value" for the linear trend for the mortality rate by day of procedure was less than 0.001 - which is statisticians' language for "there is less than one chance in a thousand that this result could have been generated by random chance."
Just to be clear how many deaths we are talking about, when I looked at the published data and checked what would have happened if the mortality rate for operations on the other days of the week had been the same as on Monday, I calculated that we are talking about precisely 5,000 fewer deaths.
The study quotes odds ratios which are adjusted for the statistical impact of age, gender, deprivation, ethnic group, and five other variables. If you apply these adjusted figures, that suggests that the impact of higher mortality rates for operations later in the week is equivalent to an extra 4,407 deaths over those three years.
The study, the results of which you can read for yourself on the BMJ website
here, was conducted at Imperial College London and looked at more than four million inpatient elective procedures conducted in NHS hospitals in England between 2008 and 2011.
In that time 27,582 of patients died within 30 days of their operation - an overall mortality rate of 0.67 per cent.
Death rates were lowest for patients having operations on Monday, increasing by around 10 per cent for each subsequent day of the week, so that for those having surgery on Fridays the risk of death was 44 per cent higher than at the start of the week.
Previous research had already established that the chance of dying after surgery is far higher at weekends, compared with weekdays, but studies have been unable to establish whether this is mainly because surgery carried out at weekends is more likely to be urgent and carry higher risks.
The new study is the first to examine the death rates during the week, and found that patients’ chances suffered dramatically as the week went on. In line with previous studies, this one suggested that death rates were highest at weekends - 82 per cent higher when adjusted for the patient's age and other risk factors than on a Monday - but the number of operations carried out on Saturdays and Sundays was small (a little less that 5% of all elective surgery), and might represent a different mix of patients, accordiny to the authors of the study.
However, they thought it more likely that the findings could reflect differences in the quality of care at the weekend. Dr Paul Aylin, the study’s lead researcher and clinical reader in epidemiology and public health at Imperial College London, said the findings suggested that a lack of staff, resources and diagnostic tests on Saturdays and Sundays were increasing the risks to patients who were recovering from surgery over the weekend.
“
The first 48 hours after an operation are often the most critical period of care for surgery patients,”
said Dr Aylin.
“So if the quality of care is lower at the weekend as some previous studies have suggested, we would expect to see higher mortality rates not just for patients operated on at the weekend, but also those who have operations towards the end of the week, whose postoperative care overlaps with the weekend. That is what we found.”
As well as looking at the data for all operations, the researchers studied death rates for several specific high risk procedures, and found the same trend for higher mortality close to the weekend. They said that
“We tried to account for the possibility that different types of patients might have operations at the end of the week, but our adjustment made little difference. This leaves us with the possibility that the differences in mortality rates are due to poorer quality of care at the weekend, perhaps because of less availability of staff, resources and diagnostic services.”
The argument is that the increase in death rates for those undergoing surgery towards the end of the week is likely to be due to “a failure to rescue the patient” during the most critical 48 hours following surgery, when complications are most likely.
My immediate reaction is that this may well be a major part of the explanation for death rates being higher for patients who had operations on Thursday, Friday or at the weekend than on Monday to Wednesday. However, it does not explain why there is also a rise in mortality from Monday to Tuesday and from Tuesday to Wednesday - and these results, though less marked, are still statistically significant.
The death rate within 30 days for all inpatient elective surgery carried out on a Monday was 5.5 per thousand, on Wednesdays 6.7 per thousand. Adjusting for age and other risk factors, mortality rate was 15% higher on Wednesdays than Mondays - representing an extra 685 or so deaths over the three years of the study for patients who had surgery on Wednesday.
The pattern of increasing deaths through the week from Monday onwards is far too consistent and too strong for it to be reasonable to conclude either that it is due to chance, or that the quality of care at the weekend, though it may well be a factor, is the only explanation.
Something is going on here, and it needs attention.
The study was conducted by the Dr Foster Unit at Imperial, funded by Dr Foster Intelligence, an independent healthcare information company, and the National Institute for Health Research.
It did not cover day surgery such as the dental procedure for which I am going into hospital tomorrow, so I am resisting the tempation to ring up and cancel my appointment and try to reschedule for a Monday.
But if I or any of my family have to go into hospital for planned surgery any time soon, at least before the issues driving this have been identified and any necessary corrective action taken, you can bet that I will be trying to make sure the operation is scheduled at the beginning of the week !
