A series of articles previously published in this journal has sought to highlight the role of nearness to death upon the marginal changes in health service demand and costs (Jones, 2019a; 2019b; 2019c; 2019d; 2019e). Ambulance demand is sensitive to nearness to death (Pentaris and Mehmet, 2019). Most deaths occur during the winter and the excess winter mortality (EWM) calculation (Johnson and Griffiths, 2003) is a good way to measure which winters create the biggest capacity problems.
In this study, the excess winter mortality starts with the average deaths in the 8 months to August versus the average in the 4 months to December. Move forward 1 month and repeat the calculation.
As has been discussed, a rolling 12-month total of deaths acts as a frequency filter which removes the usual seasonal cycle to reveal hidden patterns in the trend in deaths (Jones, 2019f). Recall that in a rolling total, a sudden shift up/down leads to a ramp up/down in the rolling total. To this end, Figure 1 shows the rolling 52-week total trend in deaths at regional level in England and Wales since 2011. As can be seen, deaths are following curious previously ignored patterns (Jones, 2019f), which show regional specificity. Note the large shift-up in deaths for the calendar year ending 2015 and a seeming (delayed) compensating shift-down, which commenced toward the end of the 2018 calendar year. No explanation is currently available.
Also note that the 2013, 2015 and 2018 calendar years show evidence for influenza epidemics imposed upon the less well understood behaviour. In an influenza epidemic, the ensuing spike in deaths creates a table-top or plateau-shaped feature. Clear is the fact that the other trends can act to increase the apparent mortality from influenza.
However, a shift down (howsoever triggered) creates the opportunity for a compensating shift-up; hence, the shift-down starting in mid-2018 has now finished and a large shift-up for the winter of 2019/2020 may ensue.
Regarding the ensuing winter capacity pressures, Figure 2 shows the trend in EWM across local authority areas in England and Wales. Data have been summarised as the median plus the first and third quartile. Firstly, note that the shape of the EWM calculation varies considerably since the totality of winter infections interacting with temperature create opportunity for unique profiles (Keatinge et al, 1997; Johnson and Griffiths, 2003; Shaw Stewart, 2016).
The shift-up in deaths, which commenced earlier in 2014 (Figure 1), led to the highest summer mortality as measured by the rolling EWM calculation in 19 years. EWM in the ensuing winter is likewise the highest in 19 years. EWM then drops back to near the average and shows an unusual trend up to the second highest EWM in 19 years for the winter of 2017/2018.
The shift-down in deaths during 2018 then creates the second lowest EWM in 19 years for the winter of 2018/2019. Record deviations are being set in the space of a few years—all with no seeming official explanation or communication to the NHS regarding possible implications for capacity planning. Government departments are somewhat loath to communicate the unexplained for fear of attracting unwanted attention. Having summarised the national picture, it is now apposite to look more deeply at local variation which is partly observable in the upper and lower quartile lines in Figure 2.
Figure 3 therefore calculates the percentage point difference between the upper and lower quartile—the interquartile range (IQR)—over the past 19 years. The IQR is a measure of the variability in winter demand (and costs) seen between different parts of the country. Figure 3 shows that the IQR demonstrates a minimum rolling EWM at October 2009 and a multimodal maximum in December 2014 (before the influenza outbreak) and February 2015 (after the influenza outbreak) (Public Health England, 2015), i.e. workload pressures show highest differences between locations in 2014/2015.
Note that deaths lag hospital admissions since illness precedes death. The funding formula does not allow for these large differences in capacity pressure to properly compensate for differences in costs between locations (Jones and Kellet, 2018). The current assumption is that cost variations are a result of inadequacies within NHS organisations, which should be ‘punished’ for failure.
The interquartile range during 2019 remains very low, i.e. everyone has jumped together. This high synchrony creates the possibility of a very bad winter in 2019/2020, and this will be discussed in Part 2 to be published in the next issue of the Journal of Paramedic Practice (Vol 11, Iss 12).