Is the problem of excessive winter deaths unique to the UK?

This content relates to the following topics:

Several commentators have reported on how improvements in life expectancy in the UK have stalled since 2010, with the most recent analysis showing that long-term mortality improvement has stalled in males and females, and at younger and older ages. Views about the causes of this disturbing trend vary widely. We have previously commented on issues that need to be considered in this debate, including whether or not these trends are unique to the UK. Given that some of this slowing has been caused by higher than the normal excess seen in winter deaths, and that data for the recent 2017–18 winter is now available, it is timely to see whether it brings more of the same.

The European Mortality Monitoring Network (EuroMOMO) monitors real-time mortality data for European countries, with the aim of detecting risks to public health from, eg, outbreaks of infectious disease or extreme weather conditions, as early as possible and informing the public health response. EuroMOMO uses a statistical model to identify when mortality rates move out of the ‘normal’ range. 

Figure 1* (from EuroMOMO's latest bulletin) shows pooled, weekly mortality data for 23 European countries from 2014 to week 25 2018, ie, last week. The data for age groups above 15, and for all ages, are adjusted for seasonality, ie, the higher numbers of deaths seen in most winters. There are pronounced spikes in all-age numbers of deaths in the winters of 2014/15, 2016/17 and 2017/18 – these are caused primarily by the spikes in mortality at ages 65+ because most deaths occur at older ages, although excess deaths at ages 15–64 also make a contribution. These spikes are well above the ‘normal’ range expected in winters. Interestingly, the numbers of deaths overall and at older ages don't spike in winter 2015/16 (or in EuroMOMO data for 2013/14 not shown here). 

Figure 1: pooled number of deaths

Figure 2* for all age-mortality shows similar patterns across most European countries. The 2014/15 spike in winter mortality is seen in most countries, and although winter 2015/16 is unremarkable in all countries, excess winter mortality above the normal range returns to most countries in 2016/17. In the recent 2017/18 winter, significant mortality excesses are again seen in many European countries, although the increase is again moderated in Scandinavian countries and not seen in Greece. 

Figure 2: weekly z-score

All EuroMOMO data is standardised so signals between countries and over time are comparable. Understanding what’s causing the similarities/divergences in mortality between European countries over time, and for individual countries over time, is critical for informing the broader debate about current mortality trends in UK.

EuroMOMO's additional analyses conclude that both the 2014/15 and the 2016/17 winter excess mortality spikes (in older people and overall)  were driven mainly by flu, and particularly the strain A(H3N2) in widespread circulation throughout Europe – the variant that most commonly causes flu fatalities among older people. EuroMOMO will report in due course on the above normal 2017/18 winter mortality seen across Europe and the role of flu in it. Meanwhile, the EU’s disease prevention and surveillance agency reports that the flu virus circulated at high levels this last winter, and for longer – from December 2017 to end March 2018.

So what are the implications of these seasonal mortality trends for overall life expectancy? The above normal mortality in winter 2014/15 was followed by a fall in life expectancy in 2015 in many European countries including UK. Thereafter the ‘benign’ winter of 2015/16 was followed by increasing life expectancy in 2016 in most European countries, again including the UK, albeit the rise here was small. How will the return of ‘bad’ winters in 2016/17 and 2017/18 impact on life expectancy across Europe? Time will tell, when the European life expectancy data for 2017 and 2018 become available. 

Meanwhile, ONS’s recent report that the mortality rate in England in quarter 1 2018 was 5 per cent higher than in quarter 1 2017, and the highest rate since 2009, has prompted the Department of Health and Social Care to ask Public Health England to conduct a review. This follows several papers postulating that the stalling of improvements in life expectancy since 2011 are associated with ‘austerity’-driven constraints on health and social care spending following the 2008 recession, and calls for an urgent investigation by government

We agree the reasons why mortality rates have shown little or no improvement since 2010 need detailed scrutiny, given also that life expectancy in the UK is lower than in many comparable European countries, especially among females. While it’s likely that several factors contribute to these trends, including some specific to UK, it does look as though excess winter deaths above the normal range and flu could be part of the story. 

Given the commonalities with European countries in terms of the slow-down in mortality improvements during the post-2008 ‘austerity years’, and the timing and causes of winter mortality spikes, such an enquiry needs a broader lens than just UK. The King’s Fund is working with the Health Foundation on research the Foundation has commissioned to examine recent mortality trends in UK, and their likely determinants, in the context also of what’s happening in Europe. We hope to provide some provisional results later this year and a final, rounded assessment by early 2019. 

  • *. Figures 1 and 2 show the number of deaths and ’z-scores’, a z-score being the number of standard deviations that a data point is from the mean. Put simply, z-scores are a way of showing how different a value is from the average (shown here as red lines). If a z-score is 0, the score as identical to the mean score. Z-scores may be positive or negative, indicating whether the value is above or below the mean, and the extent of the difference. Figure 1 shows z-scores for the numbers of deaths in the normal range (+- 2 z-scores shown as broken lines above and below the mean) and substantial increase (+4 z-scores shown above the mean as a large broken line). Figure 2 shows z-scores in the range -4 to +8. Further details of EuroMOMO's methodology are available at:



Comment date
06 July 2018

Yes, flu could be part of the problem. Or it could be some other sort of infectious agent that lowers the body's resilience and allows underlying conditions to gain the upper hand, terminally in some cases. Clearly it would have to be endemic as you have a large geographical area yet changes in the countries track each other fairly closely over the same period. What agent(s)? Difficult to say unless you're looking. Sometimes the answer is right under your nose but unless you look in the right place or with the right kit you just cannot see it.

Add your comment