Understanding the infamous hospital ‘weekend effect’ can be achieved by looking not at weekends but weekdays, write Rachel Meacock and Matt Sutton

The weekend death rate is the number of deaths occurring among patients admitted to hospital on a weekend divided by the total number of admissions on a weekend.

Rachel Meacock

The infamous ‘weekend effect’ is the extent to which this is higher than the equivalent rate among weekday admissions. The weekend death rate has repeatedly been shown to be higher than the weekday death rate. This had led the NHS in England to extend hospital services at weekends to combat what has been interpreted to be excess mortality.

However, a higher mortality rate can arise for two reasons. Weekend mortality rates may be higher because the number of deaths is higher. This would be a cause for concern.

Alternatively, the weekend effect may be driven by a reduction in admissions at the weekend. Insufficient attention has been paid to whether it is the number of deaths or the number of admissions that is driving the elevation of mortality rates at weekends. Understanding which is driving the weekend effect is critical for determining whether it is a cause for concern and, if so, how it can be tackled.

#### What drives the weekend effect?

Matt Sutton

To examine this, we used data from national hospital episode statistics for all of the 140 non-specialist acute trusts in England for the 2013/14 financial year. In common with most of the literature, we focused on emergency admissions and deaths within the following 30 days.

In Box 1, we show how the weekend effect is calculated. It is the ratio of the death rate following a weekend admission and the death rate following a weekday admission.

To unpick what is driving the weekend effect, it can be re-expressed as the product of two ratios.

The first is the number of deaths following a weekend admission divided by the number of deaths following a weekday admission. The second is the number of weekday admissions divided by the number of weekend admissions.

We have populated this calculation with average figures per day for weekdays and weekends. The weekend effect equals 1.10, as reported in the literature. The ratio of deaths is 0.88, indicating that there are fewer deaths following a weekend admission. The number of admissions on a typical weekday is 1.25 times the number of admissions on a typical weekend day. Hospitals admit 25 per cent more patients on weekdays. It is this difference between weekends and weekdays that drives the weekend effect.

Focusing only on admissions that come via accident and emergency (around 80 per cent of all emergency admissions enter hospital through this route) then allows us to examine the importance of hospital admission decisions further. Table 1 shows there are more A&E attendances on an average weekend day than on an average weekday. But there are fewer admissions, as 27.5 per cent of attending patients are admitted at weekends compared to 30.0 per cent during the week.

#### Sicker patients

When we relate the numbers of deaths to the numbers of admissions, we find that the death rate is higher at the weekend. But when we relate the numbers of deaths to the numbers of patients attending A&E, the death rate is lower at the weekend. The weekend effect is only apparent in the subset of patients who are admitted to hospital. This raises the possibility that the weekend effect is a statistical artefact caused by admissions being restricted to sicker patients at weekends.

The figures shown here are not adjusted for patient characteristics that may affect the probability of admission and the risk of death. The conclusions are unchanged when we adjust for a wide range of risk factors in our full analysis (available from 9am, Friday May 6). There is a 7 per cent relative reduction in the probability of being admitted via A&E at weekends which is not explained by patient characteristics. The risk-adjusted mortality rate for patients who attend A&E at the weekend is not different from the weekday rate.

Current initiatives to move towards seven-day hospital services will only reduce the number of deaths if reduced quality of care on admission is the major cause of the weekend effect. Our findings suggest that weekends are not unusual because there are more deaths, but because hospitals apply a more stringent admission threshold at weekends.

#### Excess deaths

Increasing hospital capacity at weekends will likely increase the number of admissions, particularly for patients with less severe illness. This will lead to lower mortality rates but this would be a statistical phenomenon rather than a clinically meaningful improvement. It would be achieved by admitting a less severe mix of patients at weekends rather than by saving additional lives.

The only real impact of the planned service extensions would then be increased strain on hospital capacity and an increase in costs.

Rather than being concerned about excess deaths at weekends, the debate about weekend versus weekday care should instead be focused on the number of admissions during the week. The weekend effect is not driven by more deaths at weekends but by more admissions of less severe patients during the week.

Rachel Meacock and Matt Sutton are from the Manchester Centre for Health Economics, Manchester University

### Box 1: Unpicking the “weekend effect” in hospital mortality rates

Weekend effect = Death rate after weekend admission

= 414 x 14764

469    11843

1.10                   = 0.88 x 1.25

Greater than 1 = Less than 1 Much greater than 1

Notes: Average daily counts across all 140 non-specialist acute rrusts in England, 1 April 2013 to 28 February 2014. Admissions are emergency admissions. Deaths are within 30 days of admission.

Table 1: Daily A&E attendances, admissions via A&E and deathsAverage weekdayAverage weekend dayDifference

A&E attendances

37,812

38,254

442

11,360

10,526

-834

30.0%

27.5%

-2.5%

Deaths within 30 days of admission

389

378

-11

3.42%

3.59%

0.17%

Death rate among attending patients

1.03%

0.99%

-0.04%