Transcript Phillip_Staddon___Winter_Mortalityx

Climate warming will not decrease
winter mortality
Philip Staddon
Research Fellow in Climate Change
Climate change in Europe: high emissions scenario
Source: EC Green Paper, 2007
Simulated Europe temperature rise
Observed temperatures
Simulated temperatures
Source Met Office Hadley Centre
Estimated Mortality Impacts of Climate Change:
Year 2000
Estimated annual deaths due to climate change: malnutrition (~80K),
diarrhoea (~50K), malaria (~20K), flooding (~3K)
14 WHO regions scaled according to estimated annual death rates due
to the change in climate since c.1970.
Estimated Additional Cost of Weather Related
Impacts on Health in Cornwall by 2050-2080
additional cost 2050-2080 £15Mn
additional cost 2050-2080 £93Mn
assuming no climate change impact
including climate change impact
£13Mn
£16Mn
new &
emerging risks heatwaves &
skin cancer
assuming a 30% increase in population and
no change in climate
By 2050-2080, the increase in
weather related health costs will be 6
times higher as a result of climate
change than they would otherwise
have been.
direct health costs (incl.
treatment and lost life years)
indirect health costs (incl.
economic and property losses)
resp. disease
£26Mn
flooding,
storms &
FBD & wildfires
WBD
£24Mn
VBD
£8Mn
£5Mn
assuming a 30% increase in population and impacts
of climate change (warming of 2-3 °C)
SUMMARY OF CLIMATE CHANGE IMPACTS ON HEALTH
•
Globally, climate change is negatively impacting human health and
wellbeing via altered weather and increased frequency of extreme
events
•
Direct impacts:
Heatwaves
Flooding
Storms
•
Indirect impacts:
Manultrition (including as a result of drought)
Water quality and availability
Food and water borne diseases
Vector borne diseases (e.g. via ticks and mosquitos)
COULD CLIMATE CHANGE BRING HEALTH BENEFITS?
•
It is widely assumed that the harmful health impacts of climate
change will be partially offset by a decline in excess winter deaths
(EWDs) in temperate countries, as winters warm.
•
UK government reports from 2012 state that winter warming will
decrease EWDs:
- HPA: "the number of cold-related deaths will likely
decrease due to milder winters"
- CCRA : "increased winter temperatures may lead to
decreased levels of mortality and morbidity due to cold".
HOW ROBUST IS THIS ASSUMPTION?
•
Over the past few decades, the UK and other temperate countries
have simultaneously experienced:
- better housing,
- improved health care,
- higher incomes,
- greater awareness of the risks of cold,
- increased government initiatives to tackle risks from cold.
•
The link between winter temperatures and EWDs may therefore
no longer be as strong as it was in preceding decades.
Excess winter deaths – definition & key points
•
EWDs = the number of deaths from December to March minus the
average number of deaths in the preceding August to November,
and the following April to July.
•
EWDs are causally attributed to seasonal variations in temperature,
with low temperatures causing death directly, and by altering
vulnerability to communicable or non-communicable diseases, such
as influenza and myocardial infarction.
•
Despite harsher winters, there are fewer EWDs in northern than
southern Europe.
Relative excess winter mortality for England and Wales over
the past 60 years presented alongside key determinants
Decreasing variability in relative excess winter mortality in
England and Wales over the past 60 years
EXCESS WINTER DEATHS DATA DESCRIPTION
Three distinct periods in EWDs changes were apparent:
(1) 1951-1970, where EWDs exhibited very high year-to-year variation,
and a strongly decreasing overall trend;
(2) 1971-2000, where year-to-year variation EWDs halved compared to
the preceding period, and the decreasing trend continues, albeit less
strongly;
(3) 2001-2011, where year-to-year variation is very small and the EWDs
rate is flat.
Detrended data showing the
year-to-year variation in relative
excess winter mortality
compared to the number of cold
days and to the activity level of
influenza like illness
Over the whole period, the number of cold days
and flu activity were highly significant, explaining
ca. 43% of the variation. Before 1976, most of the
variation is explained by the number of cold days,
but with a proportion explained by flu activity. After
1976, only flu activity accounts for any of the yearto-year variation in EWDs.
Rolling 10 year correlation between relative excess winter mortality
and the number of winter cold days
EWDs remained
strongly correlated with
the number of cold days
only up to the mid- to
late-1970s, after which
the correlation was
weak to non-existent.
Rolling 10 year correlation between relative excess winter mortality
and the activity of influenza like illness
For most of the period from
1951 to the mid 1990s
EWDs were correlated
weakly with flu activity, while
a strong correlation between
the two has emerged in
recent years.
CORRELATION BREAKDOWN ANALYSIS
By performing rolling correlations between EWDs and factors exhibiting
year-to-year variation, namely the number of cold days and the magnitude
of flu activity, it emerged that the correlation between EWDs and the
independent variable, when significant, was not stable over time.
By analysing more recent data (than in previous published work), and
performing rolling correlation analysis on time detrended data, we show
unequivocally that the correlation between the number of cold winter days
per year and EWDs, which was strong until the mid 1970s, no longer
exists.
CONCLUSIONS
•
We found that the association of year-to-year variation in EWDs with
the number of cold days in winter (< 5ºC), evident until the mid 1970s,
has disappeared, leaving only the incidence of influenza-like illnesses
to explain any of the year-to-year variation in EWDs in the last
decade.
•
Whilst excess winter deaths evidently do exist, winter cold severity no
longer predicts the numbers affected.
•
We conclude that no evidence exists that EWDs in England and
Wales will fall if winters warm with climate change.
•
These findings have important implications for climate change health
adaptation policies.
NOTE ON FUTURE TEMPERATURE VOLATILITY
From 1991 to 2011, temperature volatility shows a moderate increase.
Whether this is a real trend or not is debatable…
Nonetheless it is something worth considering in the light of climate change.
Future increases in day-to-day temperature volatility could cause
relative EWD to rise again, despite moderate general winter warming
SIMILAR CONCLUSIONS REACHED BY OTHERS
•
Ebi KL & Mills D. Winter mortality in a warming climate: a reassessment. WIREs Clim Change (2013):
“climate change may alter the balance of deaths between winters and summers, but is
unlikely to dramatically reduce overall winter mortality rates.”
•
Huang C & Barnett A. Winter weather and health. Nature Clim Change (2014):
“cold outdoor temperatures are unlikely to continue to have the same significant effect on
excess deaths in winter as they have had in the past.”
•
Woodward A. Heat, cold and climate change. J Epidemiol Community Health (2014):
“As the global climate becomes more volatile it will be more important than ever to guard
the health of vulnerable populations against weather extremes of all kinds.”
FUTURE RESEARCH
AIMS
To determine whether increased winter temperature volatility as a
result of climate change could lead to increased winter mortality
despite generally warmer winters in the future.
OBJECTIVE O1, Determine the impact of winter periods of high
temperature volatility on mortality.
OBJECTIVE O2, Determine the impact of sudden temperature drops on
mortality.
OBJECTIVE O3, Determine the impact of timing of cold spells on
mortality.
OBJECTIVE O4, Quantify the impact of future winter temperature volatility
scenarios on mortality.
FUTURE RESEARCH - HYPOTHESES
Objective O1
HYPOTHESIS H1. Winter periods of high temperature volatility exhibit increased
mortality.
HYPOTHESIS H2. A sudden increase in winter temperature volatility (compared to
previous period) leads to increased mortality.
Objective O2
HYPOTHESIS H3. Sudden drops in temperature lead to increased mortality.
HYPOTHESIS H4. The relative magnitude of sudden drops in temperature is
correlated with mortality.
Objective O3
HYPOTHESIS H5. Early cold spells (i.e. in autumn / early winter) exhibit increased
mortality.
HYPOTHESIS H6. Late cold spells (i.e. mid-late winter) exhibit lower mortality.
HYPOTHESIS H7. The magnitude of cold spells compared to previous
temperatures determines mortality.
Objective O4
HYPOTHESIS H8. Predicted increase in winter temperature volatility with climate
change will lead to increased winter mortality despite average warmer winters.
Reference:
Staddon, P.L.,
Montgomery, H.E.,
Depledge, M.H.
Climate warming will
not decrease winter
mortality.
Nature Clim Change 4:
190-194 (2014).
BACK UP
Table S1| List of data sources
Data description
Data source
Population data
Hicks, J. & Allen, G. A. Century of Changes: Trends in UK Statistics since 1900.
(1999). www.parliament.uk/documents/commons/lib/research/rp99/rp99-111.pdf
(accessed 17/01/13).
ONS (Office for National Statistics). Mid-1971 to mid-2011 population
estimates: England and Wales; quinary age group; estimated resident population.
(ONS, 2012).
Excess winter deaths ONS. Excess winter mortality in England and Wales, 1950/51 to 2009/10 (ONS,
2010).
ONS. Excess winter mortality in England and Wales, 2011/12 (provisional) and
2010/11 (final). (ONS, 2012).
Influenza incidence
Goddard, N.L., et al. Influenza surveillance in England and Wales: October 1999
to May 2000. Commun. Dis. Public Health 3, 261-266 (2000).
ONS. Excess winter mortality in England and Wales, 2011/12 (provisional) and
2010/11 (final). (ONS, 2012).
Fleming, D.M. & Elliot, A.J. Lessons from 40 years' surveillance of influenza in
England and Wales. Epidemiol. Infect. 136, 866-875 (2008).
Griffiths, C. & Brock, A. Twentieth Century Mortality Trends in England and
Wales. Health Statistics Quarterly 18. (ONS, 2003).
Daily temperature
Met Office. Met Office Hadley Centre Central England Temperature Data. (2012).
www.metoffice.gov.uk/hadobs/hadcet/data/download.html (accessed 24/10/12).
Parker, D.E., et al. A new daily Central England Temperature Series, 1772-1991.
Int. J. Clim. 12, 317-342 (1992).
Heating expenditure Kramper, P. From economic convergence to convergence in affluence? Income
growth, household expenditure and the rise of mass consumption in Britain and West
Germany, 1950-1974. LSE Working Paper 56. (2000).
ONS. Consumer Trends quarter 2 2012. (ONS, 2012).
Utley, J.I. & Shorrock, L.D. Domestic energy fact file 2008. (Department of
Energy and Climate Change, 2008). www.bre.co.uk/filelibrary/pdf/rpts/fact_file_
2008. pdf (accessed 07/08/13).
Cold weather payment Kennedy, S. Cold Weather Payments. (UK Parliament, 2010). www.parliament.uk/
briefing-papers/SN00696.pdf (accessed 17/01/13).
Winter fuel payment Ghazali, Y. Changes to the amount of Winter Fuel Payment for winter 2008/09.
(Department of Work and Pensions, 2008). www.dwp.gov.uk/docs/publishedwinter
fuelpayments-080814.pdf (accessed 17/01/13).
Warm Front
Hough, D. & Bolton, P. Warm Front Scheme. (UK Parliament, 2012).www.
parliament.uk/briefing-papers/SN06231 (accessed 17/01/13).
Housing quality
Halifax. 60 years - The changing face of the UK Housing Market. (Lloyds Banking
Group, 2012). www.lloydsbankinggroup.com/media/pdfs/Halifax/2012/2805_
Housing.pdf (accessed 21/01/13).
Palmer, J. & Cooper, I. United Kingdom housing energy fact file. (Department of
Energy and Climate Change, 2012). www.gov.uk/government/uploads/system/up
loads/attachment_data/file/201167/uk_housing_fact_file_2012.pdf (acc. 07/08/13).
DCLG. English housing survey. (Department for Communities and Local
Government, 2011). www.gov.uk/government/uploads/system/uploads/attachment
_data/file/6733/19372481.pdf (accessed 07/08/13).
DECC. Energy Consumption in the UK. (Department of Energy and Climate
Change and Office for National Statistics, 2012). www.decc.gov.uk/assets/decc/
Statistics/ publications/ecuk/file11250.pdf (accessed 21/01/13).
Randall,
C.
Housing.
Social
Trends
41.
(ONS,
2011).
1
Table 1| Multivariate regression analysis of the relationship between excess winter
2
deaths and independent variables for selected periods between 1951 and 2011
Significance level (p) and standardized coefficient (β)
3
R-square
4
Total
HQ
HC
Pol
CD
TD
FA
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
ON UNMODIFIED DATA
40
HQ housing quality; HC heating costs; Pol policy initiatives; CD number of cold days; TD
41
number of cold days with strong temperature drop; FA flu activity; NA not applicable; X
42
removed; a analysis performed on smoothed data (10 year moving average) - to identify the
43
variables behind the decreasing trend;
44
component removed) - to identify the variables behind the year-to-year variation.
1951-2011
0.78
1951-2011
0.77
1951-1971
0.72
1971-1991
0.61
1991-2011
0.72
1951-1976
0.75
1976-2011
0.65
p
β
p
β
p
β
p
β
p
β
p
β
p
β
0.000 0.002 0.032 0.312 0.000 0.544 0.000
p
β
p
β
p
β
p
β
p
β
p
β
0.000 0.000 0.016 0.040 0.534 0.633 0.712
p
β
p
β
p
β
-0.597
-0.404
0.204
0.000 0.000 0.000 X
-0.523
-0.525
0.000 0.025 NA
-0.377
NA
0.329
0.041
0.621
0.000 X
0.000
X
0.350
0.612
NA
0.002 0.461 0.178
NA
0.606
X
-0.109
0.241
0.022 0.844 0.622 0.437 0.023 0.592 0.003
0.077
-0.171
-0.265
0.560
0.102
0.796
0.003 0.141 0.299 0.290 0.622 0.455 0.000
-0.873
-0.288
0.000 0.011 NA
-0.323
NA
0.623
0.094
NA
0.001 0.654 0.011
0.175
NA
0.510
-0.053
0.765
0.340
0.000 0.170 0.089 0.932 0.092 0.318 0.000
-0.498
-0.521
-0.031
0.221
0.125
0.681
a
ON SMOOTHED DATA
1951-2011
0.92
1951-2011
0.92
1951-1985
0.97
1951-1985
0.97
1986-2011
0.88
1986-2011
0.85
-1.542
-0.351
0.357
0.066
0.000 0.000 0.010 0.009 X
-1.517
-0.320
-0.028
-0.057
X
X
X
X
0.366
X
NA
0.002 0.010 0.241
NA
NA
0.154
0.000 0.000 NA
NA
0.000 0.000 X
NA
0.176
0.000 0.000 NA
-0.952
-0.907
NA
0.130
0.161
-0.068
X
0.000 0.012 0.409 0.058 0.147 0.101 0.058
-1.187
-0.196
-0.239
0.376
0.057 X
X
0.007
X
0.695
X
X
0.489
0.000 NA
NA
NA
0.000 0.982 0.000
NA
NA
NA
0.411
0.000 NA
NA
NA
0.000 0.636 0.025
0.000 0.000 X
-1.202
0.677
0.412
b
ON DETRENDED DATA
1951-2011
0.43
1951-1976
0.62
1976-2011
0.40
-0.002
NA
NA
NA
4.383
0.000 NA
NA
NA
0.207 0.938 0.000
NA
NA
NA
0.181
b
-0.480
0.470
-0.011
2.404
0.627
analysis performed on detrended data (time
Note on EWD as the metric used
KEY ADVANTAGES
• Widely used by government and public health professionals
• Uses season/year metrics rather than daily ones – extension of daily metrics
to answer seasonal questions require the assumption that dose-response is
invariant within and across seasons and that seasonal "cold-related" effects
are simply an aggregation of daily ones
KEY DISADVANTAGES
• Excludes cold-related deaths outside the winter months
• The nominator and denominator are not fully independent, making this
measure less useful going forward (as summer heatwave deaths mushroom)
Development of alternative metrics
Asking the question at a different time scale (i.e. season or year, rather than day) is
fundamentally interesting, therefore research in temperature-health outcomes needs
to move beyond time-series analysis and develop and test metrics for "long-term
cold weather" and for "long-term health outcomes".
This has been attempted in air pollution over four decades, and much can be
learned from those approaches.
Future metrics for exposure and outcome may include variables at different levels,
e.g. days, seasons, years.
But they will have measures of exposure and response at a scale of
season/year as their starting point, as opposed to extrapolating from daily
analyses.
Excess winter deaths
theoretical impact of a warmer but more
variable climate
PL Staddon, ECEHH, University of Exeter
August 2012
Aims, scope and caveat
The aim is to demonstrate theoretically that increased variation
in winter temperature could negate the positive effect of
increased mean temperature on excess winter deaths.
All numbers are simulated and based on average low monthly
mean. The simulation is not intended to accurately reflect day to
day variation.
The results do not attempt to predict the future; they simply
highlight that the assumption that an increased mean average
winter temperature will substantially decrease excess winter
deaths is deeply flawed.
Temperature vs threshold assuming stable
variability
future climate
10
10
9
9
8
8
temperature
temperature
current climate
7
6
5
7
6
5
4
4
3
3
2
2
0
20
40
60
winter days
80
100
120
0
20
40
60
winter days
80
100
When temperature drops below the threshold, excess winter deaths occur.
A 2’C increase in average temperature means the threshold is very rarely breached,
resulting in the elimination of most excess winter deaths.
However the assumption that variability will not change is known to be false.
120
Temperature vs threshold assuming increased
variability
future climate - higher variability
12
10
11
9
10
8
9
temperature
temperature
current climate
7
6
5
4
8
7
6
5
4
3
3
2
2
0
20
40
60
winter days
80
100
120
0
20
40
60
winter days
80
100
When temperature drops below the threshold, excess winter deaths occur.
A 2’C increase in average temperature, along with significantly increased variation,
means the threshold is regularly breached, meaning excess winter deaths do not
disappear.
Note also that, due to increased variation, it is possible that the period where excess
winter deaths occur could increase.
120