References

Chen R, Yin P, Wang L Association between ambient temperature and mortality risk and burden: time series study in 272 main Chinese cities. BMJ. 2018; 363 https://doi.org/10.1136/bmj.k4306

Estimating excess winter mortality in England and Wales [Office for National Statstics]. 2003. https//tinyurl.com/yyplgchl (accessed 29 March 2019)

Jones R, Sleet G, Pearce O, Wetherill M. Complex changes in blood biochemistry revealed by a composite score derived from Principal Component Analysis: Effects of age, patient acuity, end of life, day-of week, and potential insights into the issues surrounding the ‘Weekend’ effect in hospital mortality. Br J Med Med Res. 2016; 18:(5)1-28 https://doi.org/10.9734/BJMMR/2016/29355

Jones RP. End-of-life demand is highly volatile and shows unexpected trends. J Para Pract. 2019; 11:(4)122-124

Office for National Statistics. Excess winter mortality in England and Wales. 2015. https//tinyurl.com/y7c2ab2b (accessed 29 March 2019)

Office for National Statistics. Deaths registered monthly in England and Wales. 2019. https//tinyurl.com/ybfk4pk2 (accessed 29 March 2019)

Snowed under: understanding the effects of winter on the NHS [Nuffield Trust]. 2018. https//tinyurl.com/y5fodc93 (accessed 29 March 2019)

Thorn J, Turner E, Hounsome L Validating the use of Hospital Episode Statistics data and comparison of costing methodologies for economic evaluation: an end-of-life case study from the Cluster randomised trial of PSA testing for Prostate cancer. BMJ Open. 2016; 6:(1) https://doi.org/10.1136/bmjopen-2016-011063

NHS winter workload and on/off switching of deaths

02 April 2019
Volume 11 · Issue 4

This winter, the pressure was on (Scobie, 2018). Ambulance crews had a get-together and those from area X were complaining about what a hard winter they were experiencing. Those from Y (just down the road) were sniggering in their cups of tea at those lightweights up the road.

The concept of on/off switching in deaths (and hence end-of-life acute and ambulance demand) was introduced in Part 3 of this series (Jones, 2019). In on/off switching, deaths and associated end-of-life demand suddenly jump to a new and higher level; they remain high for around 12 months and then switch-off and revert back to the usual baseline position. In view of this series on the role of nearness to death in the marginal changes in NHS workload, it may be useful to investigate just how variable winter workload can be in adjacent districts. Hospital admissions peak in the last month of life (Jones et al, 2016; Thorn et al, 2016); hence monthly deaths become a good measure of workload.

Excess winter mortality

To this end, Figure 1 presents a calculation of excess winter mortality (EWM) for England and three local authorities in Berkshire. EWM is defined as the average deaths in the 4 months from December to March versus the 4 months either side (August to November; April to July) (Johnson and Griffiths, 2003). Figure 1 takes the EWM calculation and converts it into a rolling proportion, e.g. start with the standard calculation at March 2002, move forward 1 month and recalculate. This approach is useful because it helps to visualise the shape of the trend. Of special interest is that EWM in the same county does indeed vary significantly even in adjacent areas. There are marked differences in both magnitude and timing, with Bracknell Forest experiencing a surge in deaths for the EWM calculation centred around October 2018. In some years, the trend for England is bi-modal which is the result of timing and magnitude behaviour in local authorities across the whole of England.

Figure 1. A rolling monthly calculation of excess winter mortality (EWM) for local authorities in Berkshire versus England

For the whole of England, the winter of 2014/15 had the highest EWM since 1999–2000 (Office for National Statistics (ONS), 2015), although this picture was not shared across Berkshire with only 10% EWM in Bracknell Forest compared with 32% in Reading. However, Bracknell Forest reaches its peak in April rather than March.

While temperature has a profound effect on deaths and cause of death (Chen et al, 2018; Scobie, 2018), the data for a single county (Berkshire) in Figure 1 was specifically chosen to exclude the effects of temperature gradients. Hence the most likely cause for the wildly different EWM between local authorities in Berkshire will be outbreaks of common winter viral infections (coronavirus, cytomegalovirus, influenza(s), norovirus, parainfluenza, respiratory syncytial virus, rhinovirus, etc), which can then trigger more severe pneumonia and other conditions, especially in the young and elderly. With this in mind, Figure 2 shows the range in deaths by month of the year for England from 2001 to 2018.

Figure 2. Monthly deaths in England (2001–2018), relative to the minimum deaths for each month

Figure 2 is summarised by the word ‘chaotic’. Deaths for each month can bounce between a minimum and a maximum depending on the year. Temperature variation (away from the usual monthly average) is unable to account for this extreme variation. So how may on/off switching be contributing to this chaos?

To illustrate, Figure 3 shows the relative frequency of switch-on events for local authorities/counties in Arizona, Berkshire and Scotland during the last two decades. Arizona is included to show that switch on/off is international in scope. As can be seen, switch-on most commonly occurs in late winter/early spring but can have an out-of-season peak in August. At local authority level, switch-on can create a typical 20% increase in deaths, which endures for around 12 months; hence frequency of switch-off follows the same pattern. Events of this magnitude are enough to enhance the chaos seen in Figure 2. Around 33–42% of switch on/off occurs in the winter months (as per EWM) and this will add to the variability in EWM seen in Figure 1.

Figure 3. Relative frequency of switch-on events in Arizona, Berkshire and Scotland

On/off switching would seem to be the hidden hand enhancing or diminishing winter demand. Fascinating, that something with no official recognition is having such profound effects on NHS demand and costs.