Much has already been said and written about the weekend effect, particularly in the context of a rather polarised debate about the merits of seven day working in the NHS. At times it has felt that these discussions have generated more heat than illumination. Rather than return to old arguments about whether or not mortality rates are higher for patients admitted to hospital at weekends compared with during the week, how can we develop practical approaches to tackling difficult problems like the weekend effect?
The evidence for the weekend effect is conflicting, complicated and has a lot of problems
Trying to make sense of all the studies published into the weekend effect is not an easy task: not only have there been a very large number of them (at least over 100), but there are a lot of technical details that make it challenging to interpret them. The studies are almost exclusively observational in design, studying patterns and associations in data harvested from the real world rather than from controlled experiments. Just like the real world, this type of research is messy and complicated. For example, these studies vary in how well they take account of differences in the characteristics of patients admitted at the weekend.
Adam Steventon from the Health Foundation and I explored these issues in more detail, and what they mean for our understanding of the weekend effect, in a recent editorial in BMJ Quality & Safety.
The term 'weekend effect' is, in itself, not very helpful
Because of some of the technical problems described above, it’s possible to take differing views about whether or not the higher mortality rates that exist at the weekends reflect the quality of the care provided, or the characteristics of the patients admitted. Regardless, the comparison between outcomes at weekends versus weekdays risks blinding us to a bigger, and more significant, picture: that variation also occurs over hours, days, weeks and seasons.
This should not be surprising to anyone who has worked in a hospital: a typical emergency department feels very different at 7 o’clock in the morning compared to 7 o’clock at night. For example, in a study of acute stroke care services in the NHS, colleagues and I found that variation in care quality varied in complex ways across the whole week, and not just between weekdays and weekends. Some aspects of care were poorer overnight, some ebbed and flowed across the week and some even changed between mornings and afternoons. At the same time, what happened (or rather didn’t happen) at the weekend affected people admitted during the week, with patients admitted on Thursdays and Fridays waiting the longest for assessments that were not being provided at the weekend. Encouraging us to think that weekends are the problem has closed minds to the need to look for other time-based inequalities in care and address those.
Context is key
When one study finds that that the weekend effect does or does not exist for a particular group of patients in one setting, this tells us very little about whether it exists elsewhere. Studying the weekend effect is not the health care equivalent of measuring something universal and unchanging like the Planck constant or Cosmic Microwave Background Radiation, but a phenomenon that is deeply rooted in local context. It is entirely plausible that one hospital could have poorer quality care at the weekend while the hospital down the road does not, and that the pattern could be very different six months later.
In some senses, this makes the very question 'Does the weekend effect exist?' unanswerable: all we can say is really 'Does it exist in this particular setting, at this particular time and when measured in this particular way?' Moving to the latter question is not only more intellectually honest, but also encourages us to ask more productive questions, such as 'If the weekend exists here but not there…. Why is this and what do we need to change to improve?'
Quality improvement methods can help tackle problems we don’t fully understand
If the evidence for the weekend effect is messy and conflicting, how do we go about tackling it? This is in fact a very common problem in health care, since we very rarely have all the data and knowledge required to design a 'precision engineering' solution to problems in quality. Health care systems are exceedingly complex: as an illustration, think of all the people, facilities, resources and organisations involved in a single post-take ward round or GP clinic. However, many of the methods from quality improvement can be helpful in this kind of situation.
One of the central insights of quality improvement is that systems do not have to be perfectly understood for them to be improved. Instead, problems can be tackled by working through an iterative process, whereby teams come together to develop a shared understanding of the available (imperfect) evidence, implement incremental changes, and measure their impacts. This is exemplified in the concept of the Plan-Do-Study-Act cycle. Sometimes these changes will be ineffective or be mis-steps, but by sustaining successful changes and accumulating small improvements over time, even complex and poorly understood systems can be improved.
We need to work together across organisational boundaries
Approaches to tackle the weekend effect also need to think beyond hospital walls, since the flow of patients through the hospital depends on features of the wider health and social care system. Again, this is a challenge that is not confined to the weekend effect, and the causes of many problems in health care extend beyond single organisations (and so do the solutions). Improvement requires understanding the local context, working collaboratively across organisational boundaries and building networks for sharing learning and tackling common problems. One example of such a network is Q (find out more about Q, and how you can join, by visiting the Q website).
In summary, as the heat of the controversy cools, I think we can look ahead to developing a more sophisticated understanding of the weekend effect. I hope this will include better ways of describing how care quality varies across time, unpicking the real reasons why this variation occurs and spending our energies on improvement rather than argument.
Ben Bray (@benthebray) is a member of Q, a Public Health Registrar and Research Director for the Sentinel Stroke National Audit Programme