- Email: [email protected]
Associations between high ambient temperatures and heat waves with mental health outcomes: a systematic review
p u b l i c h e a l t h x x x ( 2 0 1 8 ) 1 e2 1
Available online at www.sciencedirect.com
Public Health journal homepage: www.elsevier.com/puhe
Themed Paper e Original Research
Associations between high ambient temperatures and heat waves with mental health outcomes: a systematic review R. Thompson a,*, R. Hornigold a, L. Page b, T. Waite a a b
Public Health England, London, United Kingdom Sussex Partnership NHS Foundation Trust, Brighton, United Kingdom
article info
abstract
Article history:
Objectives: The objective of this study was to describe the mental health effects of high
Received 1 December 2017
ambient temperatures and heat waves, determine whether heat-related morbidity and
Received in revised form
mortality are increased among people with known mental disorders and identify knowl-
28 May 2018
edge gaps to inform targeting of future research.
Accepted 8 June 2018
Study design: Systematic literature review and narrative synthesis.
Available online xxx
Methods: A systematic review of published literature using MEDLINE, Embase and PsycINFO without geographical or temporal limits was conducted. We included studies that explicitly
Keywords:
linked high ambient temperatures and/or heat waves to mental health outcomes. Evidence
Heat
was graded using a National Institute of Health framework.
Hot weather
Results: A total of 35 studies were included in the review, among which 34 were observa-
High temperature
tional studies and one was a caseecontrol study. Six broad mental health outcome cate-
Mental health
gories were identified: suicide and heat; bipolar disorder, mania and depression and heat;
Public health
schizophrenia and heat; organic mental health outcomes and heat including dementia; alcohol and substance misuse and heat; and multiple mental health outcomes/mental health service usage and heat. The studies included were heterogeneous in terms of study design, population, setting, exposure measures, outcomes and location making metaanalysis unfeasible. Fifteen studies showed an increased suicide risk with heat (relative risk [RR] 1.014e1.37 per 1 C, P < 0.05; r ¼ 0.10e0.64, P < 0.05). Increased risks of mental health-related admissions and emergency department visits at higher temperatures were also found. No evidence of change in community care use was found. Evidence for the impact of heat on other mental health outcomes was mixed. Conclusions: High ambient temperatures have a range of mental health effects. The strongest evidence was found for increased suicide risk. Limited evidence was found for an increase in heat-related morbidity and mortality among people with known mental health problems. Knowledge gaps exist on the impact of high temperatures on many common mental health disorders. Mental health impacts should be incorporated into plans for the public health response to high temperatures, and as evidence evolves, psychological
* Corresponding author. E-mail address: [email protected] (R. Thompson). https://doi.org/10.1016/j.puhe.2018.06.008 0033-3506/Crown Copyright © 2018 Published by Elsevier Ltd on behalf of The Royal Society for Public Health. All rights reserved. Please cite this article in press as: Thompson R, et al., Associations between high ambient temperatures and heat waves with mental health outcomes: a systematic review, Public Health (2018), https://doi.org/10.1016/j.puhe.2018.06.008
2
p u b l i c h e a l t h x x x ( 2 0 1 8 ) 1 e2 1
morbidity and mortality temperature thresholds should be incorporated into hot weather ewarning systems. Crown Copyright © 2018 Published by Elsevier Ltd on behalf of The Royal Society for Public Health. All rights reserved.
Introduction Global average temperatures are increasing due to climate change. 2016 was the warmest year on record; the 10 warmest years on record occurred in the past 15 years1 with consensus that this warming is anthropogenic.2 Environmental hazards and disasters are associated with a range of adverse mental health impacts.3 Understanding this risk has been highlighted as a priority in a landmark UN agreement, the Sendai Framework for Disaster Risk Reduction.4 A number of direct impacts have been reported, including suicide epidemics in agriculture workers linked to prolonged drought,5 posttraumatic stress disorder (PTSD), depression after flooding in the UK6 and increased violent suicides at high ambient temperatures.7 Indirect impacts include mental health consequences of migration and displacement after disasters and conflict and reduced health infrastructure spending.7 Common mental disorders such as anxiety and depression are among the greatest health burdens globally.8 Severe mental health problems such as schizophrenia and bipolar disorder are less prevalent8 but still significant in terms of their impact on health. Treatment options for mental illhealth range from primary care interventions to admission and treatment in specialist facilities. The importance of environmental factors on mental health outcomes is beginning to receive attention.9 The fifth assessment report of the Intergovernmental Panel on Climate Change predicted rising temperatures and an increase in the frequency and intensity of heat waves by the mid-21st century.10 This suggests that extreme weather events may happen with greater frequency, such as the 2003 European heat wave that resulted in an estimated 35,000 excess deaths.11 The association between high ambient temperatures and physical illness is well documented;3,12e16 mental health has received less attention although the global evidence base is growing.7,15 The UK Climate Change Risk Assessment 201717 highlights risks to health, well-being and productivity from high temperatures as priority areas for action. It is apparent that a summary of the impacts of heat on mental health would be useful to guide planners and healthcare providers and also highlight evidence gaps. We therefore undertook a systematic review of such impacts to inform guidance on, plans for and response to extreme heat. The primary objectives of the review were to 1. determine the mental health effects of high ambient temperatures and heat waves on populations;
2. identify the geographic, sociodemographic and other contextual factors that contribute to mental disorder during hot weather; 3. determine whether heat-related mortality is increased among people with known mental health disorders; and 4. identify gaps in knowledge to support targeting of future research.
Methods A protocol following Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidance was registered with PROSPERO (2017 CRD42017076839). The search was limited to human studies without date limit. All English language studies were considered from all countries, languages and contexts.
Data sources Following advice from subject matter experts, MEDLINE, Embase and PsycINFO were searched on August 1, 2017. The search strategy is available as Supplementary Material (SM1). Additional articles were sought from reference lists of included studies and by consulting with topic experts to identify all relevant studies, regardless of publication status and hand searching for additional references.
Study selection Retrieved studies were loaded into the Endnote X7 reference management system,18 and duplicates were removed before title and abstract screening was undertaken in duplicate (R.T. and R.H.). Full-text screening was performed in duplicate by three reviewers of the review team (R.T., R.H. and T.W.). We included studies that explicitly linked high ambient temperatures and/or heat waves to mental health outcomes as follows: Population: All age groups, with or without pre-existing physical and mental health conditions, and psychiatric inpatients; Interventions: Exposure to high ambient temperature indoors or outdoors and heat waves/hot weather; Comparators: Moderate temperatures, between population groups at high temperatures; Outcomes: Morbidity: Alcohol dependence, Alzheimer’s disease, anxiety, behaviour disorder, bipolar disorder, dementia, depression, eating disorders, mania, learning disability/
Please cite this article in press as: Thompson R, et al., Associations between high ambient temperatures and heat waves with mental health outcomes: a systematic review, Public Health (2018), https://doi.org/10.1016/j.puhe.2018.06.008
p u b l i c h e a l t h x x x ( 2 0 1 8 ) 1 e2 1
autism, mood disorder, obsessiveecompulsive disorder (OCD), neuroses, panic, personality disorder, psychoses, PTSD, schizophrenia, self-harm, somatoform disorder, stress, substance misuse and suicide; Mortality: Rates of suicide and changes in mortality in those with pre-existing mental disorder. Studies that did not examine the impact of heat and/or high temperatures were excluded.
Search results and quality assessment Data were extracted independently by the reviewers (R.T., R.H. and T.W.) using a customised evidence table (Supplementary Material - SM2). Studies selected for inclusion were appraised independently by two reviewers to reduce bias (R.T., R.H. or T.W.) using a National Institute of Health (NIH) framework for Observational Cohort and Cross-Sectional Studies.19 Differences between reviewers were resolved through discussion or consultation with a subject matter expert (L.P.). Studies were grouped by disease area or principal setting and synthesised into a narrative analysis using Economic and Social Research Council guidance.20
Results A total of 9300 records were identified, of which 8716 remained after removal of duplicates. One additional article was supplied by topic experts which outlined hospital admissions during heat waves in Vietnam.21 Of 8728 total
3
records, 8640 records were excluded at title and abstract review as they were not within the scope of this review. A total of 88 full texts were reviewed, of which 35 were kept (Fig. 1). Thirty-four observational studies and one caseecontrol study were included. The association between heat and six broad mental health outcome categories was described. Seventeen studies described the association between heat and suicide;22e38 five described bipolar disorder, mania and depression;39e43 five described schizophrenia;44e48 five described organic mental health outcomes including dementia, Alzheimer's disease and senility;21,48e51 two described alcohol and substance misuse;48,51 and seven described multiple mental health outcomes and mental health service usage21,48,51e53 (Table 1). The studies varied in setting, population, country of origin and exposure metrics used. Included studies came from 19 different countries (Switzerland, Austria, Finland, Belgium, Republic of Ireland, England, Wales, Spain, Canada, United States of America, Mexico, Israel, Kazakhstan, Korea, Taiwan, Japan, Vietnam and Australia). Twenty-one studies used routinely collected health data; 15 studies used admission data to psychiatric hospitals or emergency departments. The remained used a variety of data sources including outpatient data and research and experimental data. Study quality, as assessed using the NIH framework, ranged from poor (3 studies) to good (1 study), with the majority rated ‘fair’ (31 studies) [Table 2]. The quality of many studies was affected by omission or lack of detail of the potential factors confounding the exposure/outcome relationship.
Fig. 1 e PRISMA flow diagram of search method. Please cite this article in press as: Thompson R, et al., Associations between high ambient temperatures and heat waves with mental health outcomes: a systematic review, Public Health (2018), https://doi.org/10.1016/j.puhe.2018.06.008
4
p u b l i c h e a l t h x x x ( 2 0 1 8 ) 1 e2 1
Table 1 e Number of studies and headline findings of review by outcome group. Outcome group Suicide
No. of studies
Main findings
17
49% of all included studies examined the association between suicide and temperature 15 of 17 studies found a positive and significant association between increasing temperatures and suicide frequency 4 studies found a positive and significant association between violent suicides and increasing temperature A positive and significant association between admissions due to bipolar disorder and increasing temperature No significant association was found between mania or depression and increasing temperature A positive and significant correlation between temperature and exacerbation of schizophrenic symptoms Risk of mortality more than doubled during episodes of heat wave Risk of admission for organic mental disorders increased significantly during periods of heat wave Agitation and disruptiveness of nursing home residents with known dementia increased significantly with temperature Risk of mortality due to alcohol and substance misuse increased significantly during episodes of heat wave A number of outcomes were covered by only one study with mixed findings Risk of admissions to psychiatric hospitals increased significantly during episodes of heat wave Mortality of those diagnosed with mental health illness increased significantly during periods of heat wave
Bipolar disorder, mania and depression Schizophrenia
5
Organic, dementia, Alzheimer's disease and senility
5
Alcohol and substance misuse Other mental health outcomes and service use
2
5
7
Several temperature metrics were used including maximum (TMax), minimum (TMin) and mean (TMean) temperature. Thirty-two studies used external temperature observations as the exposure. Two used internal temperatures, and one was conducted in a temperature-controlled chamber. Exposure period included daily, monthly and seasonal metrics. No meta-analysis was undertaken because of heterogeneity of study characteristics.
Suicide Seventeen articles described the relationship between suicide and temperature.22e38 Fifteen found a positive association between increasing temperature and suicide frequency22,23,25e28,30e38 with significant (P < 0.05) correlation estimates between r ¼ 0.10 and 0.6422,27,32 and relative risk (RR) estimates between 1.01 and 1.37 for increase per 1 C.23,28,30,31,36e38 One study estimated the absolute risk increase (suicide risk estimate 0.001e0.044 increase per 1 C).33 Two articles reported but did not quantify a positive relationship.25,26 Two articles found no significant relationship between suicide and temperature.24,29 Two articles found a positive association between violent suicides and temperature (r ¼ 0.20 or 0.32, P < 0.05),34,35 and one found that RR increased with temperature (RR 1.05 per 1 C, P > 0.05).37 One study reported but did not quantify a positive relationship.33 All four studies that examined nonviolent suicide rates found no association with temperature (P > 0.05). Four studies also examined the difference between male and female risk of suicide at higher temperatures. Three of these found an increased risk in male suicides compared with female suicides,22,29,31 whereas one study found that risk of suicide was comparable for males and females.37
Bipolar disorder, mania and depression Five articles described the relationship between bipolar disorder, mania or depression and temperature.39e43 Three described a positive relationship between admissions due to bipolar disorder and temperature40,42,43 (b coefficient 0.0022e0.0044, P < 0.001; r ¼ 0.27, P < 0.05),40,42 (RR 1.10 at >24 C, 1.51 at >30.7 C [reference 19.8 C]).43 Two articles described admissions due to mania; neither found a significant association with temperature;39,41 one also described no significant association between admission for depression and heat.41
Schizophrenia Five studies described the relationship between temperature and disease outcomes of individuals diagnosed with schizophrenia.44e48 One described the positive correlation between temperature and both the incidence and outcome score of schizophrenia (incidence r ¼ 0.95, P < 0.01; outcome score r ¼ 0.75, P < 0.05).44 Another study described experimentally induced temperature increase in medicated patients with schizophrenia compared with healthy controls. The authors found that the rate of increase of temperature in patients was significantly higher than among controls as environmental temperature was artificially increased (P < 0.001).45 Two studies described the correlation between temperature and exacerbation of schizophrenic symptoms.46,47 One found positive correlation between temperature and symptomology among inpatients (r ¼ 0.52e0.64, P < 0.0002);47 the other found a correlation between temperature and admission rates for acute exacerbation of schizophrenia (r ¼ 0.35, P < 0.001).46
Please cite this article in press as: Thompson R, et al., Associations between high ambient temperatures and heat waves with mental health outcomes: a systematic review, Public Health (2018), https://doi.org/10.1016/j.puhe.2018.06.008
[24]
[23]
[22]
Ref#
e Evidence against a
classic hypothesis
Eich, D. Gostynski, M.
Gutzwiller, F. Rossler,
variation on suicide in
five U.S. counties, 1991
e2001
A. N. McDonald &
K. N. Scheitlin &
J. E. Stapleton &
exposure
county used to represent
weather station in each
Monthly TAve, externaldone
(Orange NY, Pierce WA,
within five US counties
series, population level
Observational time
(1995e2000)
M. R. Holley & D. D.
KA and Ventura CA)
Effects of temperature
P. G. Dixon &
weather stations in Tyrol
Austria population data
&
with suicide
P. Parson (2003)
externaldderived from eight
Richland SC, Sedgwick
meteorological factors
G. Kemmler,
Time series of Tyrol,
between 1877 and 1968
Lugano) population
(Geneva, Zurich and
Time series, Switzerland
Study design and setting
J. S. Allen &W. M. Carter
Association of
E. A. Deisenhammer,
Daily TMean, TMax, TMin,
and southern locations
conditions and suicide
C. Sansossio, R. Bopp, M.
W. (2007)
externaldfrom three
locationsdeastern, western
between weather
Lauber,
Monthly Tmean,
Exposure
Seasonal associations
Title
Ajdacic-Gross, V.
of publication
Authors and year
Table 2 e Results summary table.
Total suicides
Statistics
National Centre for Health
Registered suicides via CDC
The positive association between
methods not provided
very large geograph-
record
correlation with suicide for mean, maximum and minimum temperatures. In a model that considered sociodemographic and geographical variables and
per 10 C TMean RR ¼ 1.13 (P ¼ 0012) Tmax RR ¼ 1.12 (P ¼ 0.011) Tmin RR ¼ 1.13 (P ¼ 0.017)
excluded if the county of
study. Records were
not specified within the
and females. Ages were
records of both males
consisted of 3355
Study population
temperature, and discriminant
R < 0.05
identifying months with
rates is not a useful tool for
departure from mean annual suicide
analysis shows that monthly
correlation between suicide and
Simple linear regression shows no
sentative of actual
etc.
FAIR
FAIR
FAIR
score
grade
Agreed
(continued on next page)
population density,
parable in terms of
Each county are com-
country
settings in one
Examines multiple
Strengths:
No lag analysis
exposures experienced
tions may not be repreof suicide risk by 12%.
different external loca-
derived from eight
Use of temperature
Weaknesses:
sures of exposure
Use of multiple mea-
sures of exposure
The use of daily mea-
temp by 10 C resulted in an increase
suicide. On average, an increase of
correlation.
Linier regression
There was a significant positive
Relative risk of suicide
general and direct nature.
fluctuations of suicide is not of a
temperature on monthly
implies that the effect of
particularly relevant in winter. This
Strengths:
changed over the some suicide methods, this issue is
on agreeable temperatures, and in
and frequency
Reporting method seems that outdoor suicides depend
associations with temperature. It
run over by a train, have the highest
ical area
measure to cover a jumping from high places and being
Two suicide methods, that is,
One temperature
experienced
actual exposures
representative of
perature may not be
mean external tem-
Use of unweighted
temperature
of abnormally high
temperatures.
to the lack of low
but is mainly due
due to warm temperatures
and suicide is not
between temperature
Essentially, the association
intermediate peak in summer.
supplemented by an
winter months,
misses shorter spells
Monthly mean temp
suicide emerges mainly during the
Weaknesses:
method of suicide
Investigates the
Large sample size
temperature and
was the only significant predictor of
Suicide cases who lived
Total suicides
and in women
Values for specific
meteorological factors, mean temp
specified in the study.
residence and place of death
temp
found in men
(P < 0.001)
excluded
were included but not
and correlation with
frequencies was
Female r ¼ 0.100
temperature
and suicide
Male r ¼ 0.197 (P < 0.001)
suicide and
Long data series of
Strengths:
weaknesses
Strengths and
between temperature
positive association
A preliminary
Results summary
suicide
coefficient, r for total
Cross correlation
Measure of effect
outside Tyrol were
184 females. All ages
day of death, place of
included; 518 males and
702 suicide cases were
specified
demographics not
female. Age and other
(128,322), both male and
All registered suicides
Population
whose records included the
between 1995 and 2000
All suicide deaths in Tyrol
Specific method of suicide
Total suicide frequency
Registered suicides
measured
Mental health outcome(s)
p u b l i c h e a l t h x x x ( 2 0 1 8 ) 1 e2 1
5
Please cite this article in press as: Thompson R, et al., Associations between high ambient temperatures and heat waves with mental health outcomes: a systematic review, Public Health (2018), https://doi.org/10.1016/j.puhe.2018.06.008
[27]
[26]
[25]
Ref# measured
Mental health outcome(s)
Population
and its possible
association with suicide
in the population of Baja
California Sur (BCS) 1985
e2008
Martagon, Vanessa
Diaz-Castro, Sara Cecilia
Zenteno-Savin, Tania
(2013)
approach
Jesus Celis Labrado-
Discovery in Database
Diaz, Y. (2016)
temperature increase
using the Knowledge
Narvaez, L. Hernandez-
de la Rosa, Alfredo de
Mexican population,
Rojop, I. E. Lopez-
Environmental
burns
temperature in a
Gaxiola-Robles, Ramon
from height, asphyxia and
environmental
from one or more locations
externaldunclear if data
seasonal average TMax,
Monthly averaged TMax and
and yearly
exposureddaily, monthly
death used as
station to the location of
1985e2008
California Sur, Mexico
population level, Baja
Ecological study at
Mexico 2005e2012
leveldTabasco State,
Total suicides
All registered suicides
Total suicides
firearms, poisoning, jumping
cutting blades, submersion,
hanging, using knifes or
specific methods of suicide:
suicidesdbroken down into
Castro, T. B. Juarez-
series, population
Fresan, A. Gonzalez-
from the closest weather
completed suicide and
Tovilla-Zarate, C. A.
Observational time
Association between
Fernandez-Arteaga, V.
from 14 years
both sexes and all ages
A total of 582 records,
more representative
higher during the hot weather than
R ¼ 0.64 (P < 0.01)
effect on suicide
temperature and its
miss short peaks in
Monthly temp may
exposure was derived
No clarity of how
ple size
Relatively small sam-
Weaknesses:
temperatures (P 0.01).
ambient defined with correlation R2 ¼ 0.68
commonly has high
Area of study
Strengths:
confounders
Unable to find info on
actual exposure
may not represent
perature as exposure
Use of external tem-
not considered
economic cofounders
Potential socio-
Weaknesses:
methods of suicide
Investigates different
rate and temp relationship well
during the cold. Hot season suicide
The suicide rate was significantly
is not stated.
hanging. Strength of this association
of exposure
station likely to be a 40 C (86e104 F) were related to males completing suicide by
the closest weather temperatures between 30 C and
Use of temp data from
Strengths:
objectives
Does not meet all
size or density
terms of population
not comparable in
The two settings are
actual exposure
may not represent
perature as exposure
Use of external tem-
Weaknesses:
Adjusts for sex
settings
Examines multiple
Strengths:
perature as exposure
Use of external tem-
ple size
Relatively small sam-
Weaknesses:
etc
Adjusts for education,
weaknesses
Strengths and
suicide: days with no rain and
environmental temperature and
An association was found between
Correlation coefficient
daysdmeasure unclear
males (n ¼ 1145) and females (n ¼ 212).
hot and very hot
age range of 10e76, both
A total of 1357 with an
External TMax & TMin, data
Number of suicides on
temperatures (32 C) rises approx. 2 the median temp (~22 C).
risk of suicide at warmer
Mississippi (1980e2006) data. 3355 total records.
All registered
females, and they were
(1986e2009) & Jackson,
and Hawkins Field Airport.
types
Suicides were more likely to occur
(22þ), males and
leveldToronto, Ontario
July, Aug and Sept.
higher than normal suicide totals for
RR but not stated in text.
during warmer conditions. Relative
derived from Toronto
in two different climate
Sheridan, S. C. (2014)
consisted of adults only
Total suicides
records were missing
TMin And monthly TAve
anomaly, all externally
temperature anomalies
Haney, C. R. Ellis, K. N.
series, population
‘Warm’ weeks were associated with
UNCLEARdchart with
warmer than the annual average
temperatures that are colder or
Results summary
only excluded if their
weekly TMax anomaly, weekly
suicide rates with
Schaffer, A. Levitt, A.
Study population
Registered suicide deaths
Measure of effect
Pearson International Airport
Weekly TAve anomaly,
Association of weekly
Dixon, P. G. Sinyor, M.
was different.
Observational time
Study design and setting
J. B. Roberts (2007)
Exposure
suicide and residence
Title
Inman &
of publication
Authors and year
Table 2 e (continued )
FAIR
FAIR
POOR
score
grade
Agreed
6 p u b l i c h e a l t h x x x ( 2 0 1 8 ) 1 e2 1
Please cite this article in press as: Thompson R, et al., Associations between high ambient temperatures and heat waves with mental health outcomes: a systematic review, Public Health (2018), https://doi.org/10.1016/j.puhe.2018.06.008
[30]
[29]
[28]
Temperature in East
Asian Countries: A
Time-Stratified Case-
Y. Guo, Y. L. Chen, B. Y.
Woo, J. M. Ebi, K. L.
(2016)
Crossover Analysis
Suicide and Ambient
Kim, Y. Kim, H. Honda,
number of stations per city
hourly data from unknown
externaldderived from
Daily average apparent temp,
specified geographical
within 25 km of
Japan and Taiwan
population level: Korea,
crossover study at
Multicountry case-
e2010
airport
stations at each regional
Health Prev Med
rates in Finland from
Partonen (2014)
study, population level
Ecological time-series
2005e2010
Total suicides
three East Asian countries
All registered suicides in
Total suicides
All recorded suicides
Total suicides
population level, Astana, Kazakhstan,
All registered suicides
Ecological time series,
regions of Finland 1974
seasons, and suicide
Kirsi Suominen, Timo
taken from local weather
Daily TMean, externalddata
from one location in Astana
TMin, apparent TMax, taken
Daily external TMean, TMax,
1974 to 2010. Environ
temperatures, thermal
Astana, Kazakhstan
(2013)
Haukka, Reija Ruuhela,
daily suicide counts in
Kosbayeva, A. Menne, B.
Local daily
air temperature and
Kozhakhmetova, G.
Laura Hiltunen, Jari
Associations between
Grjibovski, A. M.
included.
above 10 years were
and females, all ages
the study, both males
numbers not specified in
Total suicide record
within the study.
were not specified
and females. Age groups
study period, both males
recorded suicides over
A total of 10,802
in study
Other info not specified
A total of 685 records.
other weather factors.
TMean ¼ 1.37 investigated
temp to be
short-lived peaks in
associated with ambient temp in East Asia.
suggesting that suicide is positively Japan ¼ 4.5% per 4.2 C
differ significantly with sex or age, Korea ¼ 6.8% per 4.7 C
per 2.5 C
temperature and suicide did not
Taiwan ¼ 7.8% increase
in the study. Association between
suicide in all but two cities included
significantly with higher risks of
Higher temps were associated
Multiple variables tested.
inconsistent.
For women, the results were more
year analysis).
FAIR
FAIR
FAIR
(continued on next page)
actual exposure
rately represent
region may not accu-
from one station per
Use of external temp
Weaknesses:
cal power of results
set adding to statisti-
Relatively large data
long period of time
Data availability over
sex
Stratified by age and
populations
countries and
Examined multiple
Use of daily data
Strengths:
actual exposure
rately represent
region may not accu-
from one station per
Use of external temp
analysis
included in the
treatment not
tors or access to
Socio-economic fac-
Weaknesses:
Adjusts for sex
cal power of results analysis and autumnewinter half-
set adding to statistitemperature was (both year-round
Relatively large data
long period of time
Data availability over
Use of daily data
Strengths:
and temperature
ship of female suicide
examine the relation-
statistical power to
providing enough
therefore not
tion by male suicides,
ple size and domina-
Relatively small sam-
actual exposure
may not represent
perature as exposure
the lower the daily mean
rate ratio for men's suicide rate was,
the Helsinki region: the lower that
temperature and suicide in men in
association in daily mean
mean temperature
increase of each city's
corresponding to SD/2
Increase of suicide
P > 0.05
Apparent TMax ¼ 1.19
No clear picture. There was an
with outcome after controlling for
other weather variables:
perature allows
Use of external tem-
temp only significant relationship
months, holidays and
Apparent TMean ¼ 2.11
analysis. Mean and max apparent
Adjusted for years,
The use of daily tem-
Strengths:
Weaknesses:
with suicide numbers in crude
increase in temperature
TMax ¼ 0.91
All measures of temp are associated
RR of suicide per 1 C
p u b l i c h e a l t h x x x ( 2 0 1 8 ) 1 e2 1
7
Please cite this article in press as: Thompson R, et al., Associations between high ambient temperatures and heat waves with mental health outcomes: a systematic review, Public Health (2018), https://doi.org/10.1016/j.puhe.2018.06.008
[34]
[33]
[32]
[31]
Ref#
association with
climate: A population-
based study
Herng-Ching Tsai,
Shang-Ying Li, Chung-Yi
Chen, Chu-Chieh
temperature and suicide
mortality in Japan in the
Yasushi Honda, Masaji
Ono (2011)
associations with
violent and non-violent
suicide: A population-
Chin-Shyan Xirasagar,
Sudha Lee, Hsin-Chien
(2008)
based study
Seasonality and climatic
Lin, Herng-Ching Chen,
approach
series at the population level in Taiwan
weather data from 19
weather stations
Observational time
Monthly TAve,
externaldderived from daily
aggregated to monthly values
violent, non-violent
All daily suicides classed as
Non-violent suicides
analysis with a
semiparametric
Violent suicides
deaths and homicides.
comparators of accidental
violent, non-violent and
All daily suicides classed as
Total suicides
suicidesdMonthly data
Total suicides
e1995
level in Japan, 1972
series at the population
Observational time
Taiwan 1997e2003
population level in
All registered
Total suicides
All registered suicides
measured
Mental health outcome(s)
using time-series
to each region)
monitoring station assigned
Daily TMax, external (one
areas.
excluding mountainous
monitoring stations,
externaldaveraged over 19
Time-series study at
e2005.
level in Korea 2001
series, at population
Observational time
Study design and setting
confounding factors
presence of other
Relation between
Victoria Likhvar,
(2006)
Huang, Chung-Chien
Suicide rates and the
Lee, Hsin-Chien Lin,
e2005)
Monthly TAve,
station sites
temperature and suicide
mortality in Korea (2001
averaged from 64 weather
daily environmental
S. (2011)
Daily TMean, external
Exposure
Association between
Title
Kim, Y. Kim, H. Kim, D.
of publication
Authors and year
Table 2 e (continued )
temperature. (not clear at what temp the RR increases from) Males >65 years and those with less education had higher risk than females <65 and with higher
Suicides from Islands off the south coast were excluded due to significantly warmer climate
correlated with suicide rate.
R ¼ 0.376 (P < 0.001)
violent suicides were not significantly associated with temp.
Violent R ¼ 0.2012 (P < 0.001)
ages and both sexes.
(P > 0.05)
Non-violent R ¼ 0.0110
violent suicides, whereas non-
was significantly associated with
time trends, ambient temperature
After adjusting for seasonality and
discernible pattern.
methods of death did not display any
stronger than for non-violent. Other
violent showed a clear association,
violent and non-violent methods,
temperature.
and non-violent
temperature and violent
Correlation of
non-violent not given
Values for violent and
0). When this is broken down into
suggesting a short-term effect (lag of
between 0.0010 and 0.0435
estimated risk of suicide mortality
Total suicide range
found between increasing temp and
term trends, an association was
1 C increase in temperature per city.
After allowing for seasons and long-
Risk estimate for every
Use of external temp
rate.
exposure
represent actual
as exposure may not
Use of external temp
records
illness in suicide
No data on mental
miss peaks
aggregated data may
Use of monthly
Weaknesses:
Adjusts for sex
Large sample size
Strengths:
actual exposure
rately represent
region may not accu-
from one station per
Use of external temp
Weaknesses:
Use of daily data
Large sample size
Strengths:
effect on suicide
temperature and its
miss short peaks in
Monthly temp may
actual exposure
rately represent
region may not accu-
from one station per
Weaknesses:
age
Adjusts for sex and
ple size
Relatively large sam-
Strengths:
Unclear at what temp
actual exposure
not accurately reflect
External temps may
significantly associated with suicide
and unemployment, temp was still
Controlling for seasonality, trends
was significantly and positively
coefficients
included in the study, all
non-violent) were
(11,633 violent and 6497
A total of 18,130 suicides
included.
and sexes were
the study period, all ages
deaths recorded over
A total of 501,950 suicide
both males and females
groups 15e64 and >64,
separated into two age
recorded suicides,
Cross correlation showed that temp
education attainment
every 1 C increase in daily mean
attainment all included.
Cross correlation
creases above 1.0
increased daily suicide risk with
and education
A total of 18,083
mortality risk in-
that Koreans may have a 1.4%
both males and females,
Weaknesses:
(95% CI ¼ 1.010e1.017), indicating
groups, <65 and 65þ,
increasing temp
Estimate % risk with
Use of daily data
Strengths:
weaknesses
Strengths and
among all populations was 1.014
temperature and daily suicide count
association between daily
1 C increase above mean temperature
The relative risks (RR) of the
Results summary
% increase in suicide per
Measure of effect
stratified into two age
the study period,
recorded suicides over
A total of 49,451
Population
FAIR
FAIR
FAIR
FAIR
FAIR
score
grade
Agreed
8 p u b l i c h e a l t h x x x ( 2 0 1 8 ) 1 e2 1
Please cite this article in press as: Thompson R, et al., Associations between high ambient temperatures and heat waves with mental health outcomes: a systematic review, Public Health (2018), https://doi.org/10.1016/j.puhe.2018.06.008
[38]
[37]
[36]
[35]
obtained from one location
temperature and the
valuesdexternal. Weather stations were assigned to
weather on suicide rates
in the elderly in British
Columbia
Agbayewa, M. Wiggins,
S. (1999)
(Central England Temp). Two HW episodes identified
daily suicide counts and
temperature in England
and Wales
Kovats, R. S. (2007)
each geographical location.
drivers and suicide in
Australia: Bayesian
K. Tong, S. (2014)
spatial analysis
Externaldmean values for
Socio-environmental
Qi, X. Hu, W. Mengersen,
Monthly TMean,
within the study period
TMax and TMin), external
Relationship between
Page, L. A. Hajat, S.
Daily TMean (calculated from
stations were selected
Australia, 1986e2005
the population level in
Time-series analysis at
e2003
England and Wales 1993
the population level in
Ecological time series at
Total suicide
territories
eight Australian states and
All recorded suicides from
and meteorological data and suicide. age was not included.
segments and the association
Analysis was split into 4/5-year
suicide and M3 sociodemographic
suicide; M2 meteorological data and
M1 sociodemographic data and sexes. Information on
which included both
during the study period
Study used three different models:
why however.
on suicide numbers. Unclear as to
HW event did not display an effect
mortality of the same event. 2003
sets
FAIR
FAIR
FAIR
(continued on next page)
Relatively large data
factors
sociodemographic
Investigates effect of
Strengths:
chiatric diagnosis
not specify any psy-
Record of death does
exposures the 10.8% increase in all-cause
reflect experienced 41.5% suicide deaths, compared with
may not accurately
Use of external temps 1995 HW resulted in an increase of
females (3.5%).
encompassing 0), males (3.8%) and
Weaknesses:
extreme heat not significant and 95% CI
tained an episode of (5.0%), non-violent suicide (2.1%d
Study period contotal suicide (3.8%), violent suicide
every 1 C increase above 18 C for
increasing temp Use of daily data
and risk of death with
method of suicide
Investigates the
Estimated % increase in suicide for
Relative risk
temp, whereas non-violent suicides
above 18 C
exposure Strengths:
Non-violent suicide
A total of 45,293 suicides
show significant correlation with
risk for every 1 C rise
did not. Male and female suicides
Total suicides and violent suicides
% increase in suicide
represent actual
as exposure may not season.
group, in which suicide was linked to
short-lived peaks Use of external temp
No significant difference in younger
values may miss elderly suicide rate.
Use of monthly
ple size
Relatively small sam-
Weaknesses:
sex
Adjusts for age and
and temp
lation group suicide
ship of specific popu-
Examines relation-
Strengths:
associated with a 30% increase in the
2.5 C warmer current month is
preceding 3 months together with a
The combination of a 2.5 C cooler
months (RR ¼ 1.12, 95% CI 1.01e1.23).
temperature for the preceding 3
and with lower mean daily
2.5 C change in mean temperature)
1.05e1.28 for each
the current month (RR ¼ 1.16, 95% CI
higher mean daily temperature for
Elderly suicides increased with
suicide count
peaks in temp and
may miss short-lived
show comparable relationships.
and both sexes
which included all ages
during the study period
A total of 53,623 suicides
sample size was 3900.
years. Comparator
raise in temperature
violent suicide occurrence.
the past few weeks are predictors of
with changes in temperature over
Suggests that ambient temp together
Violent suicide
Total suicide
methods of suicide
suicidesddata included
All daily registered
Total suicide
their proximity. In total, 32
suicides in the 10e64 yrs age
males and females >65
suicides included both
>65 yrs age group, with a comparator of all registered
Study population of 758
All registered suicides in
exposure Use of weekly temp
represent actual
ambient temp, past temp (3-wk prior) and RH.
as exposure may not
suicide could be explained by
Use of external temp
Weaknesses:
methods of suicide
of temp with different
Examines correlation
Strengths:
that 16% of the variance of violent
Multiple regression analysis showed
Relative risk for 2.5 C
relationship between non-violent
non-violent P > 0.05
Non-violent suicide
group. Monthly values.
e1991
British Columbia 1981
population level in
and temp (r ¼ 0.32, P < 0.0001); no
(r ¼ 0.32, P < 0.0001)
correlation between violent suicide
Regression analysis showed a sig
violent suicide and temp
Correlation coefficient
suicides or homicides.
of cases not stated.
specified. Total number
sexes and age was not
Study included both
Violent suicide
homicide deaths
with a comparator of
into violent and non-violent
All suicide deaths categorised
each study region based on
aggregated to monthly
The effect of season and
Marion, S. A. Oluwafemi Time-series study at
Belgium 1979e1987
(e.g. TMax) is not stated. Data
suicide rate, ambient
D. Cosyns, P. (1994)
Daily TMax and TMin,
population level in
however the exact metric
rhythms in violent
F. Thompson, P. Peeters,
lightedark span
Observational study at
Weekly average of temp,
Synchronised annual
Maes, Michael De Meyer,
p u b l i c h e a l t h x x x ( 2 0 1 8 ) 1 e2 1
9
Please cite this article in press as: Thompson R, et al., Associations between high ambient temperatures and heat waves with mental health outcomes: a systematic review, Public Health (2018), https://doi.org/10.1016/j.puhe.2018.06.008
[40]
[39]
Ref#
ple size
rately represent
significantly and positively associated with temperature (b ¼ 0.0022, P < 0.001), with no
admission was included
episodes.
association seen for maniac
region may not accu-
unspecified bipolar admissions were
effect on suicide
temperature and its
miss short peaks in
Monthly temp may
actual exposure
from one station per
Use of external temp
Weaknesses:
confounding
Some adjustment for
weeks, only the first
associated with temperature
significantly and positively
depressive bipolar admissions were
Relatively large sam-
admissions occurred within 8
not specified
while age and sex were
selected for the study,
controlling for seasonality,
(b ¼ 0.0044, P < 0.001), and mixed or
depressive episodes and
diagnosis of manic episodes,
coefficient)
confounding Strengths:
episodes. Where multiple
Taiwan, 1999e2003
admissions were
Regression model showed,
mixed or unspecified
mountainous areas.
association with
All admissions with principal
Regression analysis (b
based study
stations, excluding
admissions and the
Ching (2007)
population level in
A total of 15,060
No adjustment for
No detail on diagnosis
isability of results
power and general-
ple size may limit
Relatively small sam-
episodes
short-lived hot
climate: A population-
averaged over 19 monitoring
bipolar disorder
Shang-Ying Lin, Herng-
Bipolar disorder admissions.
did the previous month's temp
catchment area.
Time-series analysis at
admission rates in either period nor
addresses in the unit's
Monthly TMean, external
Monthly tempera-
significantly with monthly
participants had
Seasonal variations in
investigated Weaknesses:
variations did not correlate
and 27 males). All
e1984
Lee, Hsin-Chien Tsai,
temp to be the analyses. However, temperature
included (39 females
Galway, Ireland 1980
tures do not capture
short-lived peaks in significantly with day length in both
both sexes were
psychiatric unit
variation.
perature allows only. Monthly temp correlated
admissions. All ages and
of mania or hypomania to the
The use of daily tem-
Strengths:
medications
nosis or prescribed
mental health diag-
psychiatric unit in
study period (1980e1984) and 1984
P < 0.05
patients, with 104
of patients with a diagnosis
exposure No information on
admitted to a
Analysis carried out twicedover full
cide incidence.
factors in the spatial pattern of sui-
important roles than meteorological
demographic factors played more
The results indicate that socio-
suicide
sociation between temp and
but there was no substantive as-
tors was similar to that in Model 1,
Impact of sociodemographic fac-
M3:
2.34% (2001e2005)
average of 2.27% (1996e2000) and
increased suicide rate by an
A 1 C higher yearly mean temp
reflect actual
may not accurately
with suicide from 1996 to 2005 M2:
Use of external temp
ployment positively associated
miss peaks
Monthly data may
Proportion of Aboriginal and Torres Strait Islanders and unem-
Weaknesses:
confounders
Adjusts for
weaknesses
Strengths and
M1:
in two models.
between temp and suicide assessed
Results summary
Correlation coefficients
Measure of effect
Sample size was 66
Population
Monthly rates of admission
measured
Mental health outcome(s)
county or city of Galway
series of patients
Observational time
Study design and setting
from one location in the
externaldUnclear if data are
the Prevalence of Mania
Fitzgerald, C. E.
Monaghan (1988)
Daily TMean,
Exposure
Influence of Climate on
Title
P. A. Carney, C. T.
of publication
Authors and year
Table 2 e (continued )
FAIR
POOR
score
grade
Agreed
10 p u b l i c h e a l t h x x x ( 2 0 1 8 ) 1 e2 1
Please cite this article in press as: Thompson R, et al., Associations between high ambient temperatures and heat waves with mental health outcomes: a systematic review, Public Health (2018), https://doi.org/10.1016/j.puhe.2018.06.008
[52]
[43]
[42]
[41]
which RR increases
(increasing). Neither temp nor humidity had a sig effect on predicting treatment for depression
depression or anxiety; history of depression; history of anxiety
One unit increase in temperature
and anxiety.
actual exposure modifying the effect of temp currently treated for
population
representative of relationship, with humidity only female. psychological distress; being
likely to be more suggesting that high temp drives this
representative of the
the last four weeks, presence
significant when combined
observations
study. Large cohort
over, both male and
generated from in situ
sets), all aged 45 and of high or very high levels of
study was
gridded daily temps
ple size
GOOD
FAIR
FAIR
FAIR
(continued on next page)
perature as exposure
Use of external tem-
Weaknesses:
confounding
Adjusts for
Spatial resolution however together heat alone is
from Australia
psychological distress over
participant: general
distress when considered separately,
Mental Health: Evidence
cohort study 45 and up
subgroup of a large
(53,144 complete data
external. Derived from
Relatively large sam-
Strengths:
actual exposure
may not represent
Use of external temp
provided in the data
No medical history
60,404 participants
interview/assessment),
predictors of high and very high
outcome study consisted of
Heat and Population
health were taken for each
Relationship between
randomly selected
weeks (at the time of
Humidity in the
Heat and humidity both significant
Increase % chance of Subgroup of cohort
during extreme heat conditions.
bipolar disorder episodes, especially
exacerbating the recurrence of
Weaknesses:
with temperatures above 24.0 C and heat plays an important role in
mate temperatures at bipolar disorder admissions increase
Bennett, C. M. (2016)
Five measures of mental
missing data Establishes approxi-
vulnerability to admission Results indicate that the risks of
limits effect of Females indicated greater
bipolar diagnosis.
enced exposure and
accuracy of experitemperature exceeded 30.7 C.
comparison group of non-
The Importance of Participants were a
1.23e1.84) when the daily mean
depressive), with a
potentially increases
exposure assessment (55th percentile), and 1.51 (95% CI:
unspecified maniac-
polation methods for
Use of spatial inter-
Use of daily data
sets
Relatively large data
when daily mean temp was >24 C
disorder admissions significantly
admissions with RRs of bipolar
ambient temperature and bipolar
was found between increasing
unspecified bipolar,
temperature points
admission at specific
ple size Strengths:
increased to 1.10 (95% CI 1.01e1.16)
comparators.
and 58,921 non-bipolar
with 5054 bipolar cases
ages and both sexes,
A positive correlation with 1-day lag
Relatively small sam-
peaks
miss short-lived
bipolar, mixed bipolar,
maniac bipolar, depressed
recurrent maniac disorders,
subcategories (single and
disorders, including
Population contained all
Ding, N. Berry, H. L.
Average TMax from previous 4
Psychiatric Inpatient
from the 1996e2007
from Taiwan.
disorder identified using Medical Claim data set
geographically.
temperature and bipolar
series analysis of data
psychiatric inpatients
interpolate estimated values
between ambient
H. J. (2013)
Retrospective time-
a national cohort of
stations used to spatially
A positive relationship
Sung, T. I. Chen, M. J. Su,
Daily TMean, externald372
stations
one or multiple weather Relative risk of
significantly correlated with temp.
above.
measures were derived from
Admissions of bipolar
depressed patients were not
sexes of 18 yrs and
Unclear if exposure
absolute minimal and
seasonal data may
Admission rates of unipolar
temperature
Use of monthly and
temperature (r ¼ 0.27, P ¼ 0.002)
which consisted of both
environmental
included in the study
maximal monthly temps
correlate with maximal
(2004)
maximal monthly environmental
Weizman, Abraham
of temp Weaknesses:
correlated significantly with mean
unipolar disorder)
maximum monthly temp,
minimum and mean e1991
disorder and 1036
spring/summer and
Popper, Miriam
hospitaldmonthly
Used multiple metrics
bipolar depressed patients
admitted to a psychiatric
patients increase during
Oded Hermesh, Haggai
admissions (4117 bipolar
setting in Tel Aviv, 1981
bipolar depressed
Shiloh, Roni Potchter,
Mean monthly temp, mean
Admission rates of
Shapira, Avraham unipolar depression
Strengths:
Mean monthly admission rates for
whole region
perature across the
rately represent tem-
region may not accu-
weather station per
Use of only one
exposure
represent actual
as exposure may not
Use of external temp
Weaknesses:
set
psychiatric hospital
Correlation
Association not statistically
specified.
bipolar)
A total of 6189
Tmin) in a small number of regions.
both sexes. Age not
depression (unipolar or significant for any measure.
mania and depression AND Tmax and
P > 0.05
study period including
diagnosis of mania or
All patients with bipolar or
e2002
Relatively large data
Use of daily data
admission frequency and temp (both
coefficient)
Republic of Ireland 1971
Strengths:
Only weak association seen between
Regression analysis (b
admissions during the
hospitals recorded with a
population level in the
A total of 34,465
All admissions to psychiatric
Time series at the
Time-series analysis in a
A number of temp metrics:
region selected
admissions to
psychiatric hospitals
externaldone station per
and affective disorder
A. O'Callaghan, E. (2014)
Daily TMax and TMin,
Daily weather variables
McWilliams, S. Kinsella,
p u b l i c h e a l t h x x x ( 2 0 1 8 ) 1 e2 1
11
Please cite this article in press as: Thompson R, et al., Associations between high ambient temperatures and heat waves with mental health outcomes: a systematic review, Public Health (2018), https://doi.org/10.1016/j.puhe.2018.06.008
[48]
[54]
Ref#
and psychiatric
emergencies
Collsamata, A. Joaquim,
I. Perez, Y. Tricio, O.
The effect of heat waves
on mental health in a
temperate Australian
city
Hansen, A. Bi, P.
Nitschke, M. Ryan, P.
Pisaniello, D. Tucker, G.
(2008)
Ibiza, C. (2009)
Meteorological factors
Title
Garcia, F. i Boada, S.
of publication
Authors and year
Table 2 e (continued )
significance. The main conclusion of the study is that there is a direct relationship
alcohol) and other nonspecific mental health illnesses.
four continuous days exceeding 40 C. The highest recorded maximum temperature was 44.3 C. There appeared to be a threshold temperature of 26.7 C (P ¼ 0.001),
with principal discharge diagnosis or cause of death was attributed to mental, behavioural and cognitive disorders
e1.102)
Schizophrenia ¼ 1.034 (0.969
misuse ¼ 1.005 (0.913e1.105)
due to psychoactive substance
Mental and behavioural disorders
Dementia ¼ 1.174 (1.017e1.355)
Organic ¼ 1.213 (1.091e1.349)
MBD ¼ 1.073 (1.017e1.132)
below:
days compared with non-HW days
IRR (95% CI) of hospitalisation on HW
Morbidity-specific conditions and
markedly.
disorder (MBD) admissions increased
above which mental and behavioural
January 2006 with temperatures over
admissions and mortalities
and 30th June 2006
including an extreme heat event in
ranging in duration from 3 to 8 days,
31 and 29 heat waves, respectively,
periods
deaths
admissions and 2599
psychiatric hospital
Australia. Both counts of
mortality data for south
Adelaide Metro area and
between 1st July 1993
metropolitan area
series, Adelaide
During the study periods, there were
in regards to humidity
relationship could also be observed
hospital emergency department. A
number of patients attending the
temperature increase and the
Study also incorporates HW
representative, external.
weather station deemed to be
A total of 171,614
Incidence rate ratios
comparisons did not reach statistical
stress, substance misuse (not
All hospital admissions in
(non-linear form). The remaining
psychoses, schizophrenia,
Observational time
Relatively short study
was also statistically significant
personality disorders,
Daily TMax from one central
investigated Weaknesses:
humidity and number of daily visits
mood disorders, neuroses,
between environmental
temp to be
(TMIN). The relationship between
anxiety, eating disorders,
sent actual exposure
not accurately repre-
from one station may
Use of external temp
statistical power.
nosis may limit
comes in some diag-
Small number of out-
Weaknesses:
Use of daily data
diagnosis
ship of multiple
Examines relation-
Strengths:
clear
Objective of study not
exposure was derived
No clarity of how
period
short-lived peaks in
ages
(TMAX) and minimum temperature
perature allows
The use of daily tem-
Spain. Clinical diagnosis of
endpoints of medium temperature
population Strengths:
2004e2005
in text, only significance
male and female and all
significant for the meteorological
The results were statistically
late to younger
sults may not trans-
stations
no coefficients provided
were included, both
questionnaire
effect of heat more than doubled to
Study population re-
self-reporting via
percentile within the study, the 0.5% (P < 0.001)
tures usually occur Potential bias from
When humidity rose to the 99th
before high tempera-
very high distress, respectively.
collected in spring,
(P < 0.001) increased risk of high or
Majority of data
weaknesses
Strengths and
with a 0.2% (P < 0.001) and 0.1%
and vapour pressure was associated
Results summary
(TMED), maximum temperature
Correlation e however
Measure of effect
A total of 3048 patients
Population
Reus (Psychiatric hospital)
the Hospital Pere Mata de
emergency department of
All attendees of the
measured
Mental health outcome(s)
department in Spain,
hospital emergency
series in a psychiatric
Retrospective time
Study design and setting
derived from a number of
from one weather station or
TMin, unclear if data is taken
Daily external TMean, TMax,
Exposure
FAIR
FAIR
score
grade
Agreed
12 p u b l i c h e a l t h x x x ( 2 0 1 8 ) 1 e2 1
Please cite this article in press as: Thompson R, et al., Associations between high ambient temperatures and heat waves with mental health outcomes: a systematic review, Public Health (2018), https://doi.org/10.1016/j.puhe.2018.06.008
[21]
[51]
Admissions for Mental
Disorders in Northern
Vietnam
Giang, Gunnar Kullgren,
Maria Nilsson (2016)
and substance misuse
(2012)
Joacim Rockl€ ov, Kim Bao
psychosis, dementia
Howard, Louise M.
Heat waves and Hospital
deaths in people with
Shakoor Kovats, R.
Phan Minh Trang,
Temperature-related
Page, Lisa A. Hajat,
at both national and
nationally available mortality data
England and for those with a primary
compared with non-heat wave periods. Relative risks were
spectated by ICD-10 codes (F0-3; F4-6; F7-9;
somatoform, disorder, stress,
heat waves of at least 3 days when
mental disorders for more persistent disordersdResults
admission in the whole group of Organic mental
Heat waves increased the risk for of 0, 3 & 7)
increase for all exposure metrics.
Total admissions RR ¼ 1.02 per 1 C
risk.
demonstrated greatest mortality
diagnosis of substance misuse
patients and those with a primary
temperature distribution. Younger
the 93rd percentile of the annual
1 C increase in temperature above
death of 4.9% (95% CI 2.0e7.8) per
FAIR
FAIR
(continued on next page)
of HW duration
Examine the impact
riods of HW
Examines specific pe-
temperature
Use of daily
Strengths:
exposure
reflect experienced
may not accurately
Use of external temps
numbers
limitations of
cause of death due to
Does not investigate an overall increase in the risk of
Patients with mental illness showed
associated with higher risk of death.
confounding Weaknesses:
and antipsychotics was also
misuse. Prescription of hypnotics
Adjusts for
mary care data with Southern and Eastern parts of diagnosis of alcohol or substance
Cross linking of priyounger patients, those living in
different diagnosis
Examines RR of This effect was more marked for
temperature thresholds.
above regionally defined
heat wave duration (lag
increase in temp AND
Examines relative risk
alcohol (RR ¼ 1.08) or substance misuse (RR ¼ 1.20) experienced
regional level
Use of daily data
significant heat-related mortality
Strengths: (RR ¼ 1.02), dementia (RR ¼ 1.03),
e4.138) Primary care patients with psychosis
disorders, schizophrenia,
ages and both sexes.
which comprised of all
over the study period
Relative risk per 1 C
distribution
percentile annual
increases above the 93rd
Relative risk per 1 C
disabilities/autism, mood
delirium, learning
disorders include dementia,
admissions. Specific
A total of 21,433 patients
groups: <65 and 65þ.
misuse
All mental health disorder
and stratified into two
including both sexes
during the study period
A total of 22,562 deaths
misuse and other substance
psychoses, dementia, alcohol
with primary diagnosis of
Registered deaths of people
days.
Vietnam, 2008e2012
mental hospital,
admissions to Hanoi
Time-series study of
England 1998e2007
population level in
Time series at
HW days (>35 C) and non-HW
comparisons made between
conditions identified and
external temps. HW
Daily TMax, TAve, TMin,
external
region in England and CET,
Daily TAve for every gov
Schizophrenia ¼ 2.079 (1.045
e7.155)
substance misuse ¼ 3.098 (1.342
Disorders due to psychoactive
e21.232)
Dementia (15e64) ¼ 5.058 (1.205
e4.922)
All MBDs (65e74) ¼ 2.395 (1.165
below:
days compared with non-HW days
(95% CI) of hospitalisation on HW
Mortality-specific conditions and IRR
Senility ¼ 2.366 (1.200e4.667)
e1.489)
Alzheimer's disease ¼ 1.154 (0.894
development ¼ 1.641 (1.086e2.480)
Disorders of psychological
e2.26)
Mental retardation ¼ 0.737 (0.268
factors ¼ 0.875 (0.678e1.130)
bances and physical
ated with physiological distur-
Behavioural syndromes associ-
Neurotic ¼ 1.097 (1.018e1.181)
e1.185)
Mood disorders ¼ 1.091 (1.004
p u b l i c h e a l t h x x x ( 2 0 1 8 ) 1 e2 1
13
Please cite this article in press as: Thompson R, et al., Associations between high ambient temperatures and heat waves with mental health outcomes: a systematic review, Public Health (2018), https://doi.org/10.1016/j.puhe.2018.06.008
[55]
[53]
Ref#
days. Data from weather
stations in the study regions
were spatially interpolated
before aggregating to larger
and humidity and visits
to mental health
emergency departments
in Quebec
Gosselin, Pierre (2012)
selected
psychosis admissions to
psychiatric hospitals
e one station per region
weather variables on
A. O'Callaghan, E. (2013)
Daily TMax and TMin, external
The effects of daily
McWilliams, S. Kinsella,
geographic areas.
external, and lag of 0 and 3
ambient temperature
Martin Ouarda, Taha B.
Daily TMean, Temp range,
Exposure
Relationship between
Title
Vida, Stephen Durocher,
of publication
Authors and year
Table 2 e (continued )
e2002
Republic of Ireland 1971
population level in the
Time series at the
Canada
regions of Quebec,
visits in three specific
emergency department
the population level of
Observational study at
Study design and setting
other psychoses
schizoaffective disorder and
diagnosis of schizophrenia,
hospitals recorded with a
All admissions to psychiatric
include admissions).
department visits (may not
All mental health emergency
sexes. Age not specified.
period including both
admissions during study
A total of 47,789
both sexes
from 15 y and above and
which covered all ages
over the study period,
patients are included
may not represent actual exposure
respectively. Admissions for mental disorders increased among men, residents in rural communities and the elderly population during heat waves. The groups of organic mental disorders, including symptomatic illnesses (IRR ¼ 3.62 per 7-day HW) and mental retardation (IRR ¼ 1.68 3day HW) had increased admissions during heat waves.
gender. See study for details Mental retardationdResults spectated by ICD-10 codes (F70-71; F72-73; F70-79), age group (0e17; 18e40 41þ), location (urban/rural) and gender. See study for details
provided
temps
e1.09). In most cases, the IRRs at 22.5 C and at 25 C were
however not statistically significant. The degree to which maximum temperature influenced admissions
b significant for three regions at lag 0, 2 & 7 (P < 0.05)
as exposure may not represent actual exposure
predictive of admissions at lag 0; however, the size of the time series suggests that the correlation is
undertaken further limiting power.
regions meant analysis could not be
daily time series in 5/12 study
Issues with autocorrelation from
whole region
perature across the
rately represent tem-
region may not accu-
weather station per
Use of only one
Use of external temp
highlight temperature as a factor
probably very weak.
Weaknesses:
ing weather variables
Adjust for confound-
set
Relatively large data
Use of daily data
Strengths:
actual exposure
may not represent
Use of external temp
diagnosis
Lack of detail on
Weaknesses:
Use of daily data
department data
Use of emergency
time-series regression did not
was difficult to ascertain because
temperature on admissions,
Detected a weak influence of air
below 65 years.
increases were seen only for those
geographic areas; in one area, the
for those aged 65 and older in two
both for persons under age 65 and
Significant increases were seen
homogeneous areas of Qu ebec.
in three meteorologically
mental and psychosocial problems
emergency department visits for
primarily monotonic increases in
with statistically significant and
Higher temperature was associated
(data not provided).
coefficient)
Regression analysis (b
focuses on higher
increasing temperature (IRR 1.00
thresholds
significantly higher than at 20 C
Excludes winter and
increased monotonically with
above particular
Large data set
For all geographical areas, all IRRs
Incidence rate ratios
Strengths:
Use of external temp
Weaknesses:
confounding
Adjusts for
disorders
ber of specific
Investigates a num-
(1.005e1.31) and 1.36 (1e1.90) for a 1, 3- and 7-day heat wave,
Estimation of risk
weaknesses
Strengths and
(urban/rural) and
estimated at 1.04 (0.95e1.13), 1.15
Results summary
41e60; 61þ), location
F0-9), age group (0e40;
Measure of effect
mental illness not specified.
A total of 347,552
Population
substance misuse and other
measured
Mental health outcome(s)
FAIR
FAIR
score
grade
Agreed
14 p u b l i c h e a l t h x x x ( 2 0 1 8 ) 1 e2 1
Please cite this article in press as: Thompson R, et al., Associations between high ambient temperatures and heat waves with mental health outcomes: a systematic review, Public Health (2018), https://doi.org/10.1016/j.puhe.2018.06.008
[47]
[45]
[44]
groups in terms of heart rate or blood
before and after tests
psychotic patients admitted to same ward during study period with either major depressive disorder, psychosis secondary to organic aetiology (e.g. substance abuse) or combination
calculated, weekly mental status using PANSS (Positive and Negative Syndrome Scale), comparisons using Student t-test for differences in duration of hospitalisation; Pearson test for correlation analysis.
consisted of all overtly
Daily body discomfort index
comparator group
ages from 18 to 80. A
of both sexes and all
study which comprised
exacerbation of the disorder.
during the study period in a
schizophrenia hospitalised
DSM-IV criteria for
longitudinal study
6th August 2003
Israel 1st January 2003 to
No significant difference between
and body weight measured
to an acute psychotic
mean value used.
12:00 in three locations, with
group during rest phase.
rate monitor, blood pressure
were included in the
signalling lower on comparator
both rectal and skin, heart
All subjects who met the
(P < 0.001). Rectal temp fell,
Temperature monitors for
severity of symptoms.
ward temp of 19 C would improve
Authors suggest that an optimum
leading to worsening of symptoms.
by ward temp; i.e. higher ward temps
patients' symptoms are modulated
Authors suggest that schizophrenic
finding.
patients. BDI did not affect this
24 C.
FAIR
FAIR
POOR
(continued on next page)
actually fall below
ward temp did not
mum temp of 19 C as
thors derive the opti-
Unclear how the au-
comparator group
Small sample size of
ple size of subjects
Relatively small sam-
Weaknesses:
for comparison
Use of control group
where appropriate
Use of daily data
actual exposure correlation amongst non-schizoid
representative of (r ¼ 0.52e0.64, P < 0.0002), with no
likely to be more
Use of indoor temp,
Strengths:
chological symptoms
exasperation of psy-
Does not examine
results
generalisability of
statistical power and
control size may limit
Small sample and
Weaknesses:
with temp in schizophrenic patients
PANSS score significantly correlated
between two diagnostic groups.
Admission duration no different
or physiological reasons.
if this effect is due to medication
medicated counterparts. Unclear
body than their healthy, non-
able to conduct heat from the
neuroleptic treatment are less
male patients on long-term
Results suggest schizophrenic
pressure.
comparator during exercise
mental health issues.
Use of control group
measurements
outcome
the patient group than in the
cise exposure and
(P < 0.001). Rate of rectal temp
Controlled and pre-
Strengths:
robust
for choice not very
ture data and reasons
Source of tempera-
results
generalisability of
power of results and
have an impact on
size included, may
Unclear population
exposure was derived
No clarity of how
Weaknesses:
and altitude
Adjusted for latitude
Use of daily data
Strengths:
group than in the comparator
significantly higher in the patient
eight matched men with no
Correlation coefficient
experiment time period
pressure change over
heart rate and blood
increase was significantly higher in
A total of 247 patients
cases
controls matched to
age of 31.5, eight
Rate of skin temp increase was
to higher outcome risk
Indicating that higher temps lead
months. Control group of
for at least 12 consecutive
maintained on medication
schizophrenia who had been
inpatientseA
severity of symptoms in
(2007)
psychiatric hospital in
Ecological study set in a
Israel.
climate chamber in
study performed in a
Matched caseecontrol
closed psychiatric ward due
temperature and the
Weizman, Abraham
temp measured every day at
Daily TMean, indoors e ward
sessions (indoors) for 50 mins
range: r2 ¼ 0.95; P < 0.01
Morbidity risk % and Temperature
outcomes and measures of temp
and positive correlation between
IPSS) show statistically significant
Data from both studies (DOSMD &
mean temp: r2 ¼ 0.75; P < 0.05
Skin temp, rectal temp,
Correlation coefficient
Mean overall outcome score and
Seven males, average
ages etc.
including total numbers,
of study subjects
Unclear as to the details
Mental Disorders
Seven men with
studies.
is taken from two other
assessment methods as data
unclear on diagnosis and
Schizophrenia, however
Outcome of Severe
(b)Determinants of
Study on Schizophrenia
(a)International Pilot
from two studies:
secondary use of data
Multi centre cohorts;
schizophrenia
between ward
antipsychotic drugs
Munitz, Hanan (2000)
Hanan Stryjer, Rafael
maintained with
Weizman, Abraham
A significant correlation
schizophrenia
Manaim, Hillel
Shiloh, Roni Munitz,
patients with chronic
Roni Epstein, Yoram
communication’, all external
of Schizophrenia
2 40 C in climate chamber
provided by ‘personal
Incidence and Outcome
Heat intolerance in
clear and in some cases
Temperature to the
Hermesh, Haggai Shiloh,
however locations are not
Environmental
Murray (1992)
Daily TMax, TMin, TMean,
The Relationship of
Sunjai Gupta & Robin M
p u b l i c h e a l t h x x x ( 2 0 1 8 ) 1 e2 1
15
Please cite this article in press as: Thompson R, et al., Associations between high ambient temperatures and heat waves with mental health outcomes: a systematic review, Public Health (2018), https://doi.org/10.1016/j.puhe.2018.06.008
HW days of the study period.
admissions due to
Indoor Air Temperature
and Agitation of Nursing
Home Residents With
Dementia
Federico Tartarini, Paul
Cooper, Richard
Fleming, and Marijka
Batterham (2017)
dementia in Madrid
identified, with a total of 216
factors and hospital
Diaz, J. (2017)
Ecological time series,
elevated temperature
factors
frequency scores significantly increased when indoor temp
coefficientdcorrelation values not provided
CMAI overall disruptiveness scores were found to increase significantly when average temps deviated from 22.5 C. Deviation from optimal
legal guardian. Sample consisted of both sexes with an age range of 61 e92.
both higher and lower temps.
level of disruptiveness increased at
temps showed both behaviour and
frequency scores
consent granted from
Short study period
Small sample size
and locations
patient movements
nightly assessment of
Recall bias for fort-
Weaknesses:
confounders
Adjusts for
temperature used
Precise metric of
exposure
rooms into overall significantly predicted higher CMAI
Assessment Scales and
spent in each room.
average temp in each room
spent in different
Incorporating time
temperature
Use of indoor
Strengths:
exposure
represent actual
as exposure may not
Use of external temp
exposed to temps above 26 C
deviated from 22.6 C (P < 0.01)
Agitative behaviours via CMAI total
34 C
lag ¼ 16% per 1 C increase above
AR of DDE admissions with a 1-day
Weaknesses:
day ¼ 1.19 per 1 C increase above 34 C
environmental
RR of DDE admissions with a lag of 1
Adjusts for other
The percent of time patients were
spent in different rooms.
environmental variables tested (O3 and PM).
Examines periods of
temperatures heat was greater than the other
at specific
admissions on day 2. The effect of
Estimates relative risk
set
Relatively large data
day 1 leading to increase in
had a slight lag time of 1 day, Tmax on
Impact of heat on DDE admissions
admissions start to rise.
Use of daily data
showed a clear point (34 C) at which
temps experienced Strengths:
Admissions plotted against TMax
Correlation
threshold
increase above
Relative risk with 1 C
of Psychogeriatric
higher than 4 on the
at least 3 months, scored
had been a resident for
A total 21 residents who
sex not specified
study period, ages and
admissions over the
Inventory (CMAI) and time
Cohen-Mansfield Agitation
behaviours. Assessed via
Dementia and agitated
dementia
of persons with diagnosed
A total of 1175
be representative of
as exposure may not
Use of external temp
short peaks
peratures misses
Use of monthly tem-
P < 0.001). All other climate variables did not show a significant correlation
Weaknesses:
group
Use of comparison
ple size
Relatively large sam-
Strengths:
weaknesses
Strengths and
max monthly temp (r ¼ 0.35,
and positively correlated with mean
Admission rates were significantly
periods (spring and summer).
patients were higher in warm
Admission rates of schizophrenic
Results summary
cognitive decline subject
setting in Australia
study, in a nursing home
prospective cohort
Longitudinal
Spain 2001e2009
admissions in Madrid,
hospital emergency
Daily emergency admissions
time period.
Correlation coefficient
Measure of effect
and the time the subject
TAve calculated from the
installed on internal walls.
indoordiButton sensors
Daily (dry bulb)TAve,
number of HW periods are
between environmental
Carmona, R. Ortiz, C.
Daily TMax, external. A
Short-term association
Linares, C. Culqui, D.
comparison group.
facilities during the same
with 3022 patients
over the study period,
criteria are admitted
meeting inclusion
patients admitted to same
of schizophrenia, compared
Aviv with acute exacerbation
seven psych hospitals in Tel
included in the
derived
Israel, 1981e1991
psychiatric hospital in
A total of 30,592 patients
Population
schizoaffective disorder
to psychiatric hospitals
Popper, Miriam
externaldunclear how the
Tel Aviv weather data were
All patients admitted to
measured
Mental health outcome(s)
with all patients diagnosed as
schizophrenia patients
Hermesh, Haggai
Ecological study set in a
Study design and setting
(2005)
admission rates of
Avraham Potchter, Oded
Monthly TMean, TMax, TMin,
Exposure
Weizman, Abraham
Effects of climate on
Title
Shiloh, Roni Shapira,
of publication
Authors and year
CDC, Centre for Disease Control; HW, heatwave; IRR, incidence rate ratio; RR, relative risk; Tmax, maximum temperature; Tmean, mean temperature; Tmin, minimum temperature.
[50]
[49]
[46]
Ref#
Table 2 e (continued )
FAIR
FAIR
FAIR
score
grade
Agreed
16 p u b l i c h e a l t h x x x ( 2 0 1 8 ) 1 e2 1
Please cite this article in press as: Thompson R, et al., Associations between high ambient temperatures and heat waves with mental health outcomes: a systematic review, Public Health (2018), https://doi.org/10.1016/j.puhe.2018.06.008
p u b l i c h e a l t h x x x ( 2 0 1 8 ) 1 e2 1
One study described the risk of both hospitalisation and mortality among individuals diagnosed with schizophrenia on heat wave days compared with non-heat wave days (defined at >3 consecutive days at or above 35 C).48 Risk of hospitalisation was not significantly elevated on heat wave days; the risk of mortality more than doubled (RR 2.08, 95% confidence interval 1.05e4.14, P < 0.05).
Organic disorders including dementia Two studies found an increased risk of admission for organic mental disorders during heat waves, using geographically appropriate temperature thresholds (RR ¼ 1.213 during heat waves of >3 days;48 RR ¼ 3.62 during heat waves of >7 days21). Two studies described an increased risk of hospital admission for dementia above similar heat wave threshold temperatures.48,49 One study found that agitation and disruptiveness of nursing home residents with known dementia increased significantly with temperature.50 Two studies described the relationship between high temperatures and dementia mortality.48,51 One found that the risk of mortality increased significantly on heat wave days (RR ¼ 5.058),48 and the other found that the risk of mortality increased above 18 C (RR, 1.03 per 1 C increase).51
Alcohol and substance misuse One study found no significant increase in the risk of hospitalisation due to substance use during heat waves.48 The same study found that the risk of mortality for the same population increased significantly during heat waves in females of all ages.48 A second study found an increased risk of mortality with increasing temperature for those with a primary diagnosis of both alcohol and substance misuse51 (Table 2).
Other mental health outcomes and service use Three studies that described the relationships between temperature and multiple mental health outcomes have been summarised previously.21,48,51 These studies also provided single-study evidence on the increased risk to mood, neurotic disorder, psychological development48 and mental retardation on heat wave days21 and the increasing risk of mortality from psychosis with rising temperature.51 One study of those aged above 45 years in Australia found that an increase in temperature and humidity by one unit ( C or hPa, respectively) led to a significant increase in the proportion of the population reporting high or very high distress.52 One study described all-cause risk of emergency department visits to a psychiatric hospital at three different temperature points, finding that risk increases significantly as temperature increases.53 Another study also found a significant linear relationship between psychiatric emergencies and increasing temperatures.54 A third study detected a weak influence of air temperature on admissions which was not statistically significant.55 Two articles described the risk of hospital admission with a mental health disorder during heat waves compared with
17
non-heat wave periods,21,48 using different definitions of heat waves. One article defined a heat wave as 3 consecutive days when TMax reached or exceeded 35 C,48 whereas the other used three definitions; a single day, at least 3 consecutive days and at least 7 consecutive days when TMax reached or exceeded 35 C.21 Both the articles found that the risk of admission to a psychiatric hospital with a mental health disorder increased significantly on heat wave days. One study described a significantly increased risk of mortality among those diagnosed with mental health disorders when temperatures exceeded the 93rd percentile of the annual temperature distribution.51 A second study described the significantly increased risk of mortality during heat waves among those with previously diagnosed mental health disorders.48
Discussion This review has summarised the evidence of association between higher temperatures and mental health outcomes, describing the size and direction of association and correlation where available. To the best of our knowledge, this is the first attempt to summarise the available evidence on this area and has corroborated previous studies and commentaries.7 Higher temperatures increased the risk of all mental health outcomes for which evidence was found. Significant gaps exist in evidence for the impact of temperatures on many mental health conditions. Most articles used external temperature as the exposure measure, measured at monitoring stations that (depending on the country) may be many miles from the location of individuals in the study, who thus experience very different climatic conditions.56 External temperatures are thus a limited proxy for individual exposure. Furthermore, people spend the majority of their time indoors.57 Indoor temperatures are highly variable from location to location and are not recorded or available in most dwellings. Similar problems have been described in the design of studies of the impact of heat on physical health. In general, it is thought that such methodological challenges may dilute observed associations between high temperatures, morbidity and mortality with the true effect being greater than that described.58,59 Two articles used indoor (ward) temperatures as the measure of exposure for inpatients and suggested that high temperatures may worsen symptoms for those diagnosed with mental health disorders.47,50 These findings will be of relevance to service providers to ensure that the risk of harm to patients during periods of hot weather is minimised. Suicide prevention is a World Health Organisation global priority.60 The UK recently published its third progress report on suicide prevention.61 Although there are several studies on this topic identifying a positive association, they are small in scale and do not allow for (for example) the assessment of risk at population level such as quantifying the population attributable fraction of suicide due to high temperatures. The relationship between suicide frequency (especially violent suicide) and high temperatures was consistent.22e38,62,63 The impact was greatest among men in three of four studies that described risk by sex. This is in keeping
Please cite this article in press as: Thompson R, et al., Associations between high ambient temperatures and heat waves with mental health outcomes: a systematic review, Public Health (2018), https://doi.org/10.1016/j.puhe.2018.06.008
18
p u b l i c h e a l t h x x x ( 2 0 1 8 ) 1 e2 1
with the generally elevated risk of suicide in men compared with women.64 This finding may have implications for suicide prevention plans as temperature may be a risk factor to consider in these programmes. All studies on suicide risk are observational; it is more difficult to ascertain causal effects using such designs as they simply measure associations. Despite this, robust conclusions can be made about the sum total and implications of the evidence. One article37 explores this theme, postulating sociological, biological and psychological mechanisms of plausibility, including excess alcohol use, an interaction between high ambient temperature and serotonin or a previously identified link between such temperatures and disinhibition and aggression. Average temperatures are forecast to increase with increased frequency and intensity of high temperature episodes.3,10 In the absence of any population adaptation, annual heat-related deaths are predicted to rise by 257% by 2050 due to climate change, from an annual baseline of around 2000.65 None of the included studies looked specifically at climate change, whereas the findings of our review support the assertion that the risk of suicide and other mental health outcomes is likely to increase in line with climate projections. For commissioners and managers of mental health services, all studies that examined hospital admissions or emergency department visits found significant increases with rising temperatures, particularly during heat wave periods.21,39e43,46e49,52e55,66e68 During such periods, symptoms may be exacerbated in those already diagnosed with a mental illness, leading to increased service use. Three studies investigated the impact of high temperatures on patients prescribed medication to control mental disorders. Two found that patients on medication had reduced thermoregulation ability,45,66 whereas the third found that prescription of hypnotics and antipsychotics was associated with higher risk of death at high temperatures.51 The impacts of temperature on medication side-effects may be an important factor when reviewing treatment in hot weather and is consistent with the recommendations set out in Heatwave Plan for England.69 Public health authorities around the world already use allcause mortality temperature relationships70,71 to define trigger thresholds for action in adverse weather alert systems.69,72 Understanding the thresholds at which adverse mental health impacts occur must be considered in such systems to reduce harm to health. Twelve articles described temperature thresholds at which the risk of mental health outcome of interest increased.25,26,31,36,37,43,47e51,53 Although these temperature thresholds were clearly stated, they vary by country from 18 C in England and Wales37 to 35 C in Australia and Vietnam,21,48 with in-country variation between studies. The article identifying a threshold of 18 C37 used a 10-year time-series regression analysis to find that the risk of suicide was lowest at 18 C, with rises above and below that temperature (a ‘hockey stick’eshaped distribution). It is not infrequent for mean outdoor temperatures in England to be above 18 C; such values are being recorded on 222 days over the 11-year period of the study. As previously mentioned, this article suggested biologically plausible mechanisms of causation, such as increased alcohol use, which might in turn
result in an increased propensity for violence including suicide. The authors observe that it is unlikely that deaths from suicide are merely brought forward in time during hot weatherdreal additional suicides probably occur when temperatures are high and may be linked to indoor overheating.37 Considerable heterogeneity between study methods, settings and analyses limits the collective interpretation of the studies found. For example, the periods of both exposure and outcome varied across studies with daily, weekly and monthly temperature measures used as well as different periods of aggregating and analysing outcomes. Some studies allowed for lag periods, that is, the variable duration between exposure to high temperatures and development of the health outcome under consideration, whereas others looked solely at temperature on the day of admission or diagnosis. A common limitation for weekly and monthly periods of measure was that short-lived periods of high temperature may have been missed. This may lead to underestimation of the health effects seen in the included studies. Future studies should strive to use daily data so that this potential source of bias is minimised. The range of reported measures of effect precludes direct comparison. Although studies have reported RR estimates, the degree by which the risk increases above a threshold temperature differed among many studies. For example, some studies reported RR per 1 C increase above 18 C, whereas others report RR per half a standard deviation change in temperature above the mean.30 It would be helpful if future research in this area report measures of effect in the same units so that direct comparisons can be made and metaanalysis be performed. Only five studies reported no significant association between mental health outcomes and temperature.24,29,39,41,55 This could represent publication bias or be due to the relatively limited number of published studies on all outcomes bar suicide. Within these studies, there are a number of potential confounders that may also play a role in this, for example, temporal resolution of exposure metrics used or low sample size, etc. Alternatively, this may provide a true representation of the relationship between high temperatures and the outcome of interest. Only more research in this area will provide further insight. Cases presenting to secondary care or emergency departments are likely to represent the most severe end of the disease spectrum. It is reasonable to suggest that there is a large proportion of the population with mental health disorders who are impacted by high temperatures but not presenting at secondary care facilities; such cases may seek help informally or from primary care. Studies that examine just secondary care data are thus at risk of underestimating the impact of temperature on mental health. Study sizes also differed across studies and outcomes of interest. For example, studies that investigated suicide and temperature ranged from 582 records27 up to 501,950.33 While similar ranges of sample size existed across outcome groups, the general direction of association observed was consistent. This suggests that the associations described here are reflective of the true relationship under investigation. Half of the included studies examined suicide. The remainder examined other mental health illnesses; in some
Please cite this article in press as: Thompson R, et al., Associations between high ambient temperatures and heat waves with mental health outcomes: a systematic review, Public Health (2018), https://doi.org/10.1016/j.puhe.2018.06.008
p u b l i c h e a l t h x x x ( 2 0 1 8 ) 1 e2 1
cases, only one article examined a particular outcome, such as neurotic disorder. No evidence was found for several outcomes included in the search terms which may have been expected. Further research would benefit from looking at potential associations between some of these outcomes and high temperatures. Study quality was generally ‘fair’ (88.5%) with just one study rated ‘good’. Although each study has individual limitations, as a whole the generally, positive associations throughout the body of evidence support a true effect.
Strengths and limitations To the best of our knowledge, this is the first attempt to extensively examine the published literature on the impacts of high ambient temperatures on mental health outcomes without limit on location, population or methodology. Previous work has given an overview of the impacts of climate change as a whole, and thus, this key risk area has remained unexplored until now. Strengths of this review include adherence to PRISMA guidance and the use of published quality assessment tools to assess bias. This ensures that the method and interpretation of the review is robust. Nevertheless, certain limitations must be acknowledged. Excluding non-English articles may mean some relevant studies from other countries were missed. Articles that investigated the association between health outcomes and temperature change over the course of the day (cf. high temperatures per se) were outside the review scope; rapid changes even at cooler temperature may modulate symptoms of mental health disorders. The use of clinical diagnostic codes in our search strategy may have resulted in certain studies being overlooked. For example, hot weather may be linked to social disorder and the community mental health consequences of such disorder. Our review did not identify any such studies perhaps because such evidence does not exist or more likely due to the use of clinical definitions in the search. The review scope was to investigate the relationship between high temperatures and mental health outcomes. Other meteorological exposures including humidity, solar radiation and precipitation may affect mental health. The observational nature of most of the research found conducted in the real world using routinely collected data sets means that temperature was not the only factor to which populations were exposed. Humidity may have a particular effect; in one included study when humidity rose to the 99th percentile, the effect of heat on distress more than doubled.52 This relationship warrants further research. Furthermore, the heterogeneity of methods used in studies undertaken to date limits this review to a narrative synthesis without meta-analysis.
19
We found that the strongest evidence was for mortality and heat, specifically suicide. We can be confident in these findings; however, there is less confidence from the evidence on the relationship between heat and all-cause mortality in those with mental illness. The relationship between heat and psychological morbidity is less certain; the evidence of worsening symptoms among those with mental illnesses was mixed with a limited number of studies focussing on specific outcomes. We found that the evidence was relatively strong for increasing mental health service use during periods of high temperatures. Although the results of this review suggest a positive relationship between mental health outcomes and high temperatures, there remain a number of areas that require further attention. Specific areas that need addressing are the impact of high temperatures on common mental health disorders, ensuring that future researchers use consistent measures of exposure, outcome and effect at as fine a temporal resolution as possible; attempt to identify threshold temperatures at which adverse mental health outcomes increase and finally explore the link between indoor temperature and mental health outcomes. Of relevance to service providers and commissioners are the findings that high indoor temperatures lead to exacerbation of mental health illnesses, suggesting that they should ensure a cool room is available during periods of high temperature, in line with the recommendations in the Heatwave Plan for England. In addition, the finding that service use was found to increase significantly during periods of high temperature may have business continuity implications. Public health authorities may wish to update suicide prevention strategies to acknowledge the impact that high temperatures and wider environmental factors may have. Furthermore, the mental health impacts of high temperatures should be incorporated into plans for the public health response to high temperatures, and as more evidence is generated, psychological morbidity and mortality temperature thresholds should be incorporated into hot weatherewarning system thresholds for action. At a time of increasing global temperatures, the importance of understanding the impact on psychological morbidity and mortality, filling the gaps in our knowledge and using the available evidence to inform policy and practice has never been more urgent.
Author statements Ethical approval Not required as this was a review of the published literature.
Funding Conclusions This review is the first to summarise the impact of high temperatures on mental health. The results have highlighted gaps in the knowledge base and have identified areas relevant to providers and commissioners and mental health services and for public health action.
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.
Competing interests The authors have no competing interests to declare.
Please cite this article in press as: Thompson R, et al., Associations between high ambient temperatures and heat waves with mental health outcomes: a systematic review, Public Health (2018), https://doi.org/10.1016/j.puhe.2018.06.008
20
p u b l i c h e a l t h x x x ( 2 0 1 8 ) 1 e2 1
references
1. Administration TNOaA, editor. State of the climate: global climate report for annual 2016. The National Oceanic and Atmospheric Administration; 2017. 2. Cook J, Oreskes N, Doran P, Anderegg W, Verheggen B, W Maibach E, et al. Consensus on consensus: a synthesis of consensus estimates on human-caused global warming. Environ Res Lett 2016;11. 3. Climate change 2014: synthesis report. Contribution of Working Group II to the fifth assessment report of the Intergovernmental Panel on Climate Change. Geneva, Switzerland: Intergovernmental Panel on Climate Change; 2014. 4. Sendai framework for disaster risk reduction 2015e2030. 2015. 5. Hanigana IC, Butlera CD, Kokicc PN, Hutchinsond MF. Suicide and drought in new South Wales, Australia, 1970e2007. Proc Natl Acad Sci USA 2012;109. 6. Waite TD, Chaintarli K, Beck CR, Bone A, Richard A, Kovats S, et al. The English national cohort study of flooding and health: cross-sectional analysis of mental health outcomes at year one. BMC Public Health 2017;17. 7. Page LA, Howard LM. The impact of climate change on mental health (but will mental health be discussed at Copenhagen?). Psychol Med 2009;40:177e80. 8. Global, regional, and national incidence, prevalence, and years lived with disability for 301 acute and chronic diseases and injuries in 188 countries, 1990e2013: a systematic analysis for the Global Burden of Disease Study 2013. Lancet 2015;386:743e800. 9. Department for Environment FaRA, editor. A green future: our 25 year plan to improve the environment. HM Government; 2018. 10. Kirtman B, Power SB, Adedoyin JA, Boer GJ, Bojariu R, Camilloni I, et al. Near-term climate change: projections and predictability. In: Climate change 2013: the physical science basis. Contribution of Working Group I to the fifth assessment report of the Intergovernmental Panel on Climate Change. Cambridge, United Kingdom and New York, NY, USA: Intergovernmental Panel on Climate Change; 2013. 11. Sardon J. The 2003 heat wave. Eurosurveillance 2007;12. 12. Armstrong B. Models for the relationship between ambient temperature and daily mortality. Epidemiology 2006:624e31. 13. Hajat S, Kovats RS, Lachowycz K. Heat-related and coldrelated deaths in England and Wales: who is at risk? Occup Environ Med 2007:93e100. 14. Ye X, Wolff R, Yu W, Vaneckova P, Pan X, Tong S. Ambient temperature and morbidity: a review of epidemiological evidence. Environ Health Perspect 2012:19e28. 15. Smith KR, Woodward A, Campbell-Lendrum D, Chadee DD, Honda Y, Liu Q, et al. Human health: impacts, adaptation, and co-benefits. In: Climate change 2014: impacts, adaptation, and vulnerability. Part A: Global and sectoral aspects. Contribution of Working Group II to the fifth assessment report of the Intergovernmental Panel on Climate Change. Cambridge, United Kingdom and New York, NY, USA: Intergovernmental Panel on Climate Change; 2014. 16. Xu Z, FitzGerald G, Guo Y, Jalaludin B, Tong S. Impact of heatwave on mortality under different heatwave definitions: a systematic review and meta-analysis. Environ Int 2016;89e90:193e203. 17. UK climate change risk assessment 2017 synthesis report: priorities for the next five years. London: Adaptation Sub-Committee of the Committee on Climate Change; 2016. 18. Analytics C. Endnote X7. Thomson Reuters; 2013. 19. National Institute of Health study quality assessment tools. National Institute of Health; Available from: https://www. nhlbi.nih.gov/health-topics/study-quality-assessment-tools (last accessed 08 August 2017).
20. Harden A, Thomas J. Methods for the thematic synthesis of qualitative research in systematic reviews. € v J, Giang KB, Kullgren G, Nilsson M. 21. Trang PM, Rocklo Heatwaves and hospital admissions for mental disorders in Northern Vietnam. PLoS One 2016;11, e0155609. 22. Ajdacic-Gross V, Lauber C, Sansossio R, Bopp M, Eich D, Gostynski M, et al. Seasonal associations between weather conditions and suicide - evidence against a classic hypothesis. Am J Epidemiol 2007;165:561e9. 23. Deisenhammer EA, Kemmler G, Parson P. Association of meteorological factors with suicide. Acta Psychiatr Scand 2003;108:455e9. 24. Dixon PG, McDonald AN, Scheitlin KN, Stapleton JE, Allen JS, Carter WM, et al. Effects of temperature variation on suicide in five U.S. counties, 1991-2001. Int J Biometeorol 2007;51:395e403. 25. Dixon PG, Sinyor M, Schaffer A, Levitt A, Haney CR, Ellis KN, et al. Association of weekly suicide rates with temperature anomalies in two different climate types. Int J Environ Res Public Health 2014;11:11627e44. 26. Fernandez-Arteaga V, Tovilla-Zarate CA, Fresan A, GonzalezCastro TB, Juarez-Rojop IE, Lopez-Narvaez L, et al. Association between completed suicide and environmental temperature in a Mexican population, using the Knowledge Discovery in Database approach. Comput Methods Programs Biomed 2016;135:219e24. 27. Gaxiola-Robles R, de la Rosa AdJC, Labrado-Martagon V, DiazCastro SC, Zenteno-Savin T. Environmental temperature increase and its possible association with suicide in the population of Baja California Sur (BCS) 1985-2008. Salud Ment 2013;36:421e7. 28. Grjibovski AM, Kozhakhmetova G, Kosbayeva A, Menne B. Associations between air temperature and daily suicide counts in Astana, Kazakhstan. Medicina (Kaunas) 2013;49:379e85. 29. Hiltunen L, Haukka J, Ruuhela R, Suominen K, Partonen T. Local daily temperatures, thermal seasons, and suicide rates in Finland from 1974 to 2010. Environ Health Prev Med 2014;19:286e94. 30. Kim Y, Kim H, Honda Y, Guo YL, Chen BY, Woo JM, et al. Suicide and ambient temperature in East Asian countries: a time-stratified case-crossover analysis. Environ Health Perspect 2016;124:75e80. 31. Kim Y, Kim H, Kim DS. Association between daily environmental temperature and suicide mortality in Korea (2001-2005). Psychiatry Res 2011;186:390e6. 32. Lee H-C, Lin H-C, Tsai S-Y, Li C-Y, Chen C-C, Huang C-C. Suicide rates and the association with climate: a populationbased study. J Affect Disord 2006;92:221e6. 33. Likhvar V, Honda Y, Ono M. Relation between temperature and suicide mortality in Japan in the presence of other confounding factors using time-series analysis with a semiparametric approach. Environ Health Prev Med 2011;16:36e43. 34. Lin H-C, Chen C-S, Xirasagar S, Lee H-C. Seasonality and climatic associations with violent and nonviolent suicide: a population-based study. Neuropsychobiology 2008;57:32e7. 35. Maes M, De Meyer F, Thompson P, Peeters D, Cosyns P. Synchronized annual rhythms in violent suicide rate, ambient temperature and the light-dark span. Acta Psychiatr Scand 1994;90:391e6. 36. Marion SA, Oluwafemi Agbayewa M, Wiggins S. The effect of season and weather on suicide rates in the elderly in British Columbia. Can J Public Health 1999;90:418e22. 37. Page LA, Hajat S, Kovats RS. Relationship between daily suicide counts and temperature in England and Wales. Br J Psychiatry 2007;191:106e12.
Please cite this article in press as: Thompson R, et al., Associations between high ambient temperatures and heat waves with mental health outcomes: a systematic review, Public Health (2018), https://doi.org/10.1016/j.puhe.2018.06.008
p u b l i c h e a l t h x x x ( 2 0 1 8 ) 1 e2 1
38. Qi X, Hu W, Mengersen K, Tong S. Socio-environmental drivers and suicide in Australia: Bayesian spatial analysis. BMC Public Health 2014;14:681. 39. Carney PA, Fitzgerald CT, Monaghan CE. Influence of climate on the prevalence of mania. Br J Psychiatry 1988;152:820e3. 40. Lee H-C, Tsai S-Y, Lin H-C. Seasonal variations in bipolar disorder admissions and the association with climate: a population-based study. J Affect Disord 2007;97:61e9. 41. McWilliams S, Kinsella A, O'Callaghan E. Daily weather variables and affective disorder admissions to psychiatric hospitals. Int J Biometeorol 2014;58:2045e57. 42. Shapira A, Shiloh R, Potchter O, Hermesh H, Popper M, Weizman A. Admission rates of bipolar depressed patients increase during spring/summer and correlate with maximal environmental temperature. Bipolar Disord 2004;6:90e3. 43. Sung TI, Chen MJ, Su HJ. A positive relationship between ambient temperature and bipolar disorder identified using a national cohort of psychiatric inpatients. Soc Psychiatry Psychiatr Epidemiol 2013;48:295e302. 44. Gupta S, Murray RM. The relationship of environmental temperature to the incidence and outcome of schizophrenia. Br J Psychiatry 1992;160:788e92. 45. Hermesh H, Shiloh R, Epstein Y, Manaim H, Weizman A, Munitz H. Heat intolerance in patients with chronic schizophrenia maintained with antipsychotic drugs. Am J Psychiatry 2000;157:1327e9. 46. Shiloh R, Shapira A, Potchter O, Hermesh H, Popper M, Weizman A. Effects of climate on admission rates of schizophrenia patients to psychiatric hospitals. Eur Psychiatry 2005;20:61e4. 47. Shiloh R, Munitz H, Stryjer R, Weizman A. A significant correlation between ward temperature and the severity of symptoms in schizophrenia inpatientsea longitudinal study. Eur Neuropsychopharmacol 2007;17:478e82. 48. Hansen A, Bi P, Nitschke M, Ryan P, Pisaniello D, Tucker G. The effect of heat waves on mental health in a temperate Australian city. Environ Health Perspect 2008;116:1369e75. 49. Linares C, Culqui D, Carmona R, Ortiz C, Diaz J. Short-term association between environmental factors and hospital admissions due to dementia in Madrid. Environ Res 2017;152:214e20. 50. Tartarini F, Cooper P, Fleming R, Batterham M. Indoor air temperature and agitation of nursing home residents with dementia. Am J Alzheimer's Dis Other Dementias 2017;32:272e81. 51. Page LA, Hajat S, Kovats R, Howard LM. Temperature-related deaths in people with psychosis, dementia and substance misuse. Br J Psychiatry 2012;200:485e90. 52. Ding N, Berry HL, Bennett CM. The importance of humidity in the relationship between heat and population mental health: evidence from Australia. PLoS One 2016;11, e0164190. 53. Vida S, Durocher M, Ouarda TB, Gosselin P. Relationship between ambient temperature and humidity and visits to mental health emergency departments in Quebec. Psychiatr Serv 2012;63:1150e3. 54. Garcia F, Boada S, Collsamata A, Joaquim I, Perez Y, Tricio O, et al. Meteorological factors and psychiatric emergencies. Actas Esp Psiquiatr 2009;37:34e41. 55. McWilliams S, Kinsella A, O'Callaghan E. The effects of daily weather variables on psychosis admissions to psychiatric hospitals. Int J Biometeorol 2013;57:497e508. 56. Nguyen JL, Schwartz J, Dockery DW. The relationship between indoor and outdoor temperature, apparent temperature, relative humidity, and absolute humidity. Indoor Air 2014;24:103e12.
21
57. Schweizer C, Edwards RD, Bayer-Oglesby L, Gauderman WJ, Ilacqua V, Juhani Jantunen M, et al. Indoor timeemicroenvironmenteactivity patterns in seven regions of Europe. J Expo Sci Environ Epidemiol 2006;17:170. 58. Lomas KJ, Porritt SM. Overheating in buildings: lessons from research. Build Res Inf 2017;45:1e18. 59. Anderson M, Carmichael C, Murray V, Dengel A, Swainson M. Defining indoor heat thresholds for health in the UK. Perspect Public Health 2013;133:158e64. 60. Preventing suicide: A global imperative. World Health Organization; 2014. 61. Health Do, editor. Preventing suicide in England: third progress report of the cross-government outcomes strategy to save lives. gov.uk: HM Government; 2017. 62. Linkowski P, Martin F, de Maertelaer V. Effect of some climatic factors on violent and non-violent suicides in Belgium. J Affect Disord 1992;25:161e6. 63. Ruuhela R, Hiltunen L, Venalainen A, Pirinen P, Partonen T. Climate impact on suicide rates in Finland from 1971 to 2003. Int J Biometeorol 2009;53:167e75. € ki M, Bell S, Gale CR, Shipley M, Whitley E, 64. Batty GD, Kivima et al. Psychosocial characteristics as potential predictors of suicide in adults: an overview of the evidence with new results from prospective cohort studies. Transl Psychiatry 2018;8:22. 65. Hajat S, Vardoulakis S, Heaviside C, Eggen B. Climate change effects on human health: projections of temperature-related mortality for the UK during the 2020s, 2050s and 2080s. J Epidemiol Commun Health 2014. 66. Boker SM, Leibenluft E, Deboeck PR, Virk G, Postolache TT. Mood oscillations and coupling between mood and weather in patients with Rapid Cycling Bipolar Disorder. Int J Child Health Hum Dev 2008;1:181e202. 67. Medici CR, Vestergaard CH, Hadzi-Pavlovic D, MunkJorgensen P, Parker G. Seasonal variations in hospital admissions for mania: examining for associations with weather variables over time. J Affect Disord 2016;205:81e6. 68. Banziger G, Owens K. Geophysical variables and behavior: II. Weather factors as predictors of local social indicators of maladaptation in two non-urban areas. Psychol Rep 1978;43:427e34. 69. England PH, editor. Heatwave plan for England. HM Government; 2015. https://www.gov.uk/government/ publications/heatwave-plan-for-england (last accessed 08 August 2017). 70. Armstrong BG, Chalabi Z, Fenn B, Hajat S, Kovats S, Milojevic A, et al. Association of mortality with high temperatures in a temperate climate: England and Wales. J Epidemiol Commun Health 2011;65:340e5. 71. Heatwaves and health: guidance on warning-system development. World Health Organization and World Meteorological Organization; 2015. p. 114. 72. England PH, editor. Cold weather plan for England. HM Government; 2015. https://www.gov.uk/government/ publications/cold-weather-plan-cwp-for-england (last accessed 08 August 2017).
Appendix A. Supplementary data Supplementary data related to this article can be found at https://doi.org/10.1016/j.puhe.2018.06.008.
Please cite this article in press as: Thompson R, et al., Associations between high ambient temperatures and heat waves with mental health outcomes: a systematic review, Public Health (2018), https://doi.org/10.1016/j.puhe.2018.06.008
Available online at www.sciencedirect.com
Public Health journal homepage: www.elsevier.com/puhe
Themed Paper e Original Research
Associations between high ambient temperatures and heat waves with mental health outcomes: a systematic review R. Thompson a,*, R. Hornigold a, L. Page b, T. Waite a a b
Public Health England, London, United Kingdom Sussex Partnership NHS Foundation Trust, Brighton, United Kingdom
article info
abstract
Article history:
Objectives: The objective of this study was to describe the mental health effects of high
Received 1 December 2017
ambient temperatures and heat waves, determine whether heat-related morbidity and
Received in revised form
mortality are increased among people with known mental disorders and identify knowl-
28 May 2018
edge gaps to inform targeting of future research.
Accepted 8 June 2018
Study design: Systematic literature review and narrative synthesis.
Available online xxx
Methods: A systematic review of published literature using MEDLINE, Embase and PsycINFO without geographical or temporal limits was conducted. We included studies that explicitly
Keywords:
linked high ambient temperatures and/or heat waves to mental health outcomes. Evidence
Heat
was graded using a National Institute of Health framework.
Hot weather
Results: A total of 35 studies were included in the review, among which 34 were observa-
High temperature
tional studies and one was a caseecontrol study. Six broad mental health outcome cate-
Mental health
gories were identified: suicide and heat; bipolar disorder, mania and depression and heat;
Public health
schizophrenia and heat; organic mental health outcomes and heat including dementia; alcohol and substance misuse and heat; and multiple mental health outcomes/mental health service usage and heat. The studies included were heterogeneous in terms of study design, population, setting, exposure measures, outcomes and location making metaanalysis unfeasible. Fifteen studies showed an increased suicide risk with heat (relative risk [RR] 1.014e1.37 per 1 C, P < 0.05; r ¼ 0.10e0.64, P < 0.05). Increased risks of mental health-related admissions and emergency department visits at higher temperatures were also found. No evidence of change in community care use was found. Evidence for the impact of heat on other mental health outcomes was mixed. Conclusions: High ambient temperatures have a range of mental health effects. The strongest evidence was found for increased suicide risk. Limited evidence was found for an increase in heat-related morbidity and mortality among people with known mental health problems. Knowledge gaps exist on the impact of high temperatures on many common mental health disorders. Mental health impacts should be incorporated into plans for the public health response to high temperatures, and as evidence evolves, psychological
* Corresponding author. E-mail address: [email protected] (R. Thompson). https://doi.org/10.1016/j.puhe.2018.06.008 0033-3506/Crown Copyright © 2018 Published by Elsevier Ltd on behalf of The Royal Society for Public Health. All rights reserved. Please cite this article in press as: Thompson R, et al., Associations between high ambient temperatures and heat waves with mental health outcomes: a systematic review, Public Health (2018), https://doi.org/10.1016/j.puhe.2018.06.008
2
p u b l i c h e a l t h x x x ( 2 0 1 8 ) 1 e2 1
morbidity and mortality temperature thresholds should be incorporated into hot weather ewarning systems. Crown Copyright © 2018 Published by Elsevier Ltd on behalf of The Royal Society for Public Health. All rights reserved.
Introduction Global average temperatures are increasing due to climate change. 2016 was the warmest year on record; the 10 warmest years on record occurred in the past 15 years1 with consensus that this warming is anthropogenic.2 Environmental hazards and disasters are associated with a range of adverse mental health impacts.3 Understanding this risk has been highlighted as a priority in a landmark UN agreement, the Sendai Framework for Disaster Risk Reduction.4 A number of direct impacts have been reported, including suicide epidemics in agriculture workers linked to prolonged drought,5 posttraumatic stress disorder (PTSD), depression after flooding in the UK6 and increased violent suicides at high ambient temperatures.7 Indirect impacts include mental health consequences of migration and displacement after disasters and conflict and reduced health infrastructure spending.7 Common mental disorders such as anxiety and depression are among the greatest health burdens globally.8 Severe mental health problems such as schizophrenia and bipolar disorder are less prevalent8 but still significant in terms of their impact on health. Treatment options for mental illhealth range from primary care interventions to admission and treatment in specialist facilities. The importance of environmental factors on mental health outcomes is beginning to receive attention.9 The fifth assessment report of the Intergovernmental Panel on Climate Change predicted rising temperatures and an increase in the frequency and intensity of heat waves by the mid-21st century.10 This suggests that extreme weather events may happen with greater frequency, such as the 2003 European heat wave that resulted in an estimated 35,000 excess deaths.11 The association between high ambient temperatures and physical illness is well documented;3,12e16 mental health has received less attention although the global evidence base is growing.7,15 The UK Climate Change Risk Assessment 201717 highlights risks to health, well-being and productivity from high temperatures as priority areas for action. It is apparent that a summary of the impacts of heat on mental health would be useful to guide planners and healthcare providers and also highlight evidence gaps. We therefore undertook a systematic review of such impacts to inform guidance on, plans for and response to extreme heat. The primary objectives of the review were to 1. determine the mental health effects of high ambient temperatures and heat waves on populations;
2. identify the geographic, sociodemographic and other contextual factors that contribute to mental disorder during hot weather; 3. determine whether heat-related mortality is increased among people with known mental health disorders; and 4. identify gaps in knowledge to support targeting of future research.
Methods A protocol following Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidance was registered with PROSPERO (2017 CRD42017076839). The search was limited to human studies without date limit. All English language studies were considered from all countries, languages and contexts.
Data sources Following advice from subject matter experts, MEDLINE, Embase and PsycINFO were searched on August 1, 2017. The search strategy is available as Supplementary Material (SM1). Additional articles were sought from reference lists of included studies and by consulting with topic experts to identify all relevant studies, regardless of publication status and hand searching for additional references.
Study selection Retrieved studies were loaded into the Endnote X7 reference management system,18 and duplicates were removed before title and abstract screening was undertaken in duplicate (R.T. and R.H.). Full-text screening was performed in duplicate by three reviewers of the review team (R.T., R.H. and T.W.). We included studies that explicitly linked high ambient temperatures and/or heat waves to mental health outcomes as follows: Population: All age groups, with or without pre-existing physical and mental health conditions, and psychiatric inpatients; Interventions: Exposure to high ambient temperature indoors or outdoors and heat waves/hot weather; Comparators: Moderate temperatures, between population groups at high temperatures; Outcomes: Morbidity: Alcohol dependence, Alzheimer’s disease, anxiety, behaviour disorder, bipolar disorder, dementia, depression, eating disorders, mania, learning disability/
Please cite this article in press as: Thompson R, et al., Associations between high ambient temperatures and heat waves with mental health outcomes: a systematic review, Public Health (2018), https://doi.org/10.1016/j.puhe.2018.06.008
p u b l i c h e a l t h x x x ( 2 0 1 8 ) 1 e2 1
autism, mood disorder, obsessiveecompulsive disorder (OCD), neuroses, panic, personality disorder, psychoses, PTSD, schizophrenia, self-harm, somatoform disorder, stress, substance misuse and suicide; Mortality: Rates of suicide and changes in mortality in those with pre-existing mental disorder. Studies that did not examine the impact of heat and/or high temperatures were excluded.
Search results and quality assessment Data were extracted independently by the reviewers (R.T., R.H. and T.W.) using a customised evidence table (Supplementary Material - SM2). Studies selected for inclusion were appraised independently by two reviewers to reduce bias (R.T., R.H. or T.W.) using a National Institute of Health (NIH) framework for Observational Cohort and Cross-Sectional Studies.19 Differences between reviewers were resolved through discussion or consultation with a subject matter expert (L.P.). Studies were grouped by disease area or principal setting and synthesised into a narrative analysis using Economic and Social Research Council guidance.20
Results A total of 9300 records were identified, of which 8716 remained after removal of duplicates. One additional article was supplied by topic experts which outlined hospital admissions during heat waves in Vietnam.21 Of 8728 total
3
records, 8640 records were excluded at title and abstract review as they were not within the scope of this review. A total of 88 full texts were reviewed, of which 35 were kept (Fig. 1). Thirty-four observational studies and one caseecontrol study were included. The association between heat and six broad mental health outcome categories was described. Seventeen studies described the association between heat and suicide;22e38 five described bipolar disorder, mania and depression;39e43 five described schizophrenia;44e48 five described organic mental health outcomes including dementia, Alzheimer's disease and senility;21,48e51 two described alcohol and substance misuse;48,51 and seven described multiple mental health outcomes and mental health service usage21,48,51e53 (Table 1). The studies varied in setting, population, country of origin and exposure metrics used. Included studies came from 19 different countries (Switzerland, Austria, Finland, Belgium, Republic of Ireland, England, Wales, Spain, Canada, United States of America, Mexico, Israel, Kazakhstan, Korea, Taiwan, Japan, Vietnam and Australia). Twenty-one studies used routinely collected health data; 15 studies used admission data to psychiatric hospitals or emergency departments. The remained used a variety of data sources including outpatient data and research and experimental data. Study quality, as assessed using the NIH framework, ranged from poor (3 studies) to good (1 study), with the majority rated ‘fair’ (31 studies) [Table 2]. The quality of many studies was affected by omission or lack of detail of the potential factors confounding the exposure/outcome relationship.
Fig. 1 e PRISMA flow diagram of search method. Please cite this article in press as: Thompson R, et al., Associations between high ambient temperatures and heat waves with mental health outcomes: a systematic review, Public Health (2018), https://doi.org/10.1016/j.puhe.2018.06.008
4
p u b l i c h e a l t h x x x ( 2 0 1 8 ) 1 e2 1
Table 1 e Number of studies and headline findings of review by outcome group. Outcome group Suicide
No. of studies
Main findings
17
49% of all included studies examined the association between suicide and temperature 15 of 17 studies found a positive and significant association between increasing temperatures and suicide frequency 4 studies found a positive and significant association between violent suicides and increasing temperature A positive and significant association between admissions due to bipolar disorder and increasing temperature No significant association was found between mania or depression and increasing temperature A positive and significant correlation between temperature and exacerbation of schizophrenic symptoms Risk of mortality more than doubled during episodes of heat wave Risk of admission for organic mental disorders increased significantly during periods of heat wave Agitation and disruptiveness of nursing home residents with known dementia increased significantly with temperature Risk of mortality due to alcohol and substance misuse increased significantly during episodes of heat wave A number of outcomes were covered by only one study with mixed findings Risk of admissions to psychiatric hospitals increased significantly during episodes of heat wave Mortality of those diagnosed with mental health illness increased significantly during periods of heat wave
Bipolar disorder, mania and depression Schizophrenia
5
Organic, dementia, Alzheimer's disease and senility
5
Alcohol and substance misuse Other mental health outcomes and service use
2
5
7
Several temperature metrics were used including maximum (TMax), minimum (TMin) and mean (TMean) temperature. Thirty-two studies used external temperature observations as the exposure. Two used internal temperatures, and one was conducted in a temperature-controlled chamber. Exposure period included daily, monthly and seasonal metrics. No meta-analysis was undertaken because of heterogeneity of study characteristics.
Suicide Seventeen articles described the relationship between suicide and temperature.22e38 Fifteen found a positive association between increasing temperature and suicide frequency22,23,25e28,30e38 with significant (P < 0.05) correlation estimates between r ¼ 0.10 and 0.6422,27,32 and relative risk (RR) estimates between 1.01 and 1.37 for increase per 1 C.23,28,30,31,36e38 One study estimated the absolute risk increase (suicide risk estimate 0.001e0.044 increase per 1 C).33 Two articles reported but did not quantify a positive relationship.25,26 Two articles found no significant relationship between suicide and temperature.24,29 Two articles found a positive association between violent suicides and temperature (r ¼ 0.20 or 0.32, P < 0.05),34,35 and one found that RR increased with temperature (RR 1.05 per 1 C, P > 0.05).37 One study reported but did not quantify a positive relationship.33 All four studies that examined nonviolent suicide rates found no association with temperature (P > 0.05). Four studies also examined the difference between male and female risk of suicide at higher temperatures. Three of these found an increased risk in male suicides compared with female suicides,22,29,31 whereas one study found that risk of suicide was comparable for males and females.37
Bipolar disorder, mania and depression Five articles described the relationship between bipolar disorder, mania or depression and temperature.39e43 Three described a positive relationship between admissions due to bipolar disorder and temperature40,42,43 (b coefficient 0.0022e0.0044, P < 0.001; r ¼ 0.27, P < 0.05),40,42 (RR 1.10 at >24 C, 1.51 at >30.7 C [reference 19.8 C]).43 Two articles described admissions due to mania; neither found a significant association with temperature;39,41 one also described no significant association between admission for depression and heat.41
Schizophrenia Five studies described the relationship between temperature and disease outcomes of individuals diagnosed with schizophrenia.44e48 One described the positive correlation between temperature and both the incidence and outcome score of schizophrenia (incidence r ¼ 0.95, P < 0.01; outcome score r ¼ 0.75, P < 0.05).44 Another study described experimentally induced temperature increase in medicated patients with schizophrenia compared with healthy controls. The authors found that the rate of increase of temperature in patients was significantly higher than among controls as environmental temperature was artificially increased (P < 0.001).45 Two studies described the correlation between temperature and exacerbation of schizophrenic symptoms.46,47 One found positive correlation between temperature and symptomology among inpatients (r ¼ 0.52e0.64, P < 0.0002);47 the other found a correlation between temperature and admission rates for acute exacerbation of schizophrenia (r ¼ 0.35, P < 0.001).46
Please cite this article in press as: Thompson R, et al., Associations between high ambient temperatures and heat waves with mental health outcomes: a systematic review, Public Health (2018), https://doi.org/10.1016/j.puhe.2018.06.008
[24]
[23]
[22]
Ref#
e Evidence against a
classic hypothesis
Eich, D. Gostynski, M.
Gutzwiller, F. Rossler,
variation on suicide in
five U.S. counties, 1991
e2001
A. N. McDonald &
K. N. Scheitlin &
J. E. Stapleton &
exposure
county used to represent
weather station in each
Monthly TAve, externaldone
(Orange NY, Pierce WA,
within five US counties
series, population level
Observational time
(1995e2000)
M. R. Holley & D. D.
KA and Ventura CA)
Effects of temperature
P. G. Dixon &
weather stations in Tyrol
Austria population data
&
with suicide
P. Parson (2003)
externaldderived from eight
Richland SC, Sedgwick
meteorological factors
G. Kemmler,
Time series of Tyrol,
between 1877 and 1968
Lugano) population
(Geneva, Zurich and
Time series, Switzerland
Study design and setting
J. S. Allen &W. M. Carter
Association of
E. A. Deisenhammer,
Daily TMean, TMax, TMin,
and southern locations
conditions and suicide
C. Sansossio, R. Bopp, M.
W. (2007)
externaldfrom three
locationsdeastern, western
between weather
Lauber,
Monthly Tmean,
Exposure
Seasonal associations
Title
Ajdacic-Gross, V.
of publication
Authors and year
Table 2 e Results summary table.
Total suicides
Statistics
National Centre for Health
Registered suicides via CDC
The positive association between
methods not provided
very large geograph-
record
correlation with suicide for mean, maximum and minimum temperatures. In a model that considered sociodemographic and geographical variables and
per 10 C TMean RR ¼ 1.13 (P ¼ 0012) Tmax RR ¼ 1.12 (P ¼ 0.011) Tmin RR ¼ 1.13 (P ¼ 0.017)
excluded if the county of
study. Records were
not specified within the
and females. Ages were
records of both males
consisted of 3355
Study population
temperature, and discriminant
R < 0.05
identifying months with
rates is not a useful tool for
departure from mean annual suicide
analysis shows that monthly
correlation between suicide and
Simple linear regression shows no
sentative of actual
etc.
FAIR
FAIR
FAIR
score
grade
Agreed
(continued on next page)
population density,
parable in terms of
Each county are com-
country
settings in one
Examines multiple
Strengths:
No lag analysis
exposures experienced
tions may not be repreof suicide risk by 12%.
different external loca-
derived from eight
Use of temperature
Weaknesses:
sures of exposure
Use of multiple mea-
sures of exposure
The use of daily mea-
temp by 10 C resulted in an increase
suicide. On average, an increase of
correlation.
Linier regression
There was a significant positive
Relative risk of suicide
general and direct nature.
fluctuations of suicide is not of a
temperature on monthly
implies that the effect of
particularly relevant in winter. This
Strengths:
changed over the some suicide methods, this issue is
on agreeable temperatures, and in
and frequency
Reporting method seems that outdoor suicides depend
associations with temperature. It
run over by a train, have the highest
ical area
measure to cover a jumping from high places and being
Two suicide methods, that is,
One temperature
experienced
actual exposures
representative of
perature may not be
mean external tem-
Use of unweighted
temperature
of abnormally high
temperatures.
to the lack of low
but is mainly due
due to warm temperatures
and suicide is not
between temperature
Essentially, the association
intermediate peak in summer.
supplemented by an
winter months,
misses shorter spells
Monthly mean temp
suicide emerges mainly during the
Weaknesses:
method of suicide
Investigates the
Large sample size
temperature and
was the only significant predictor of
Suicide cases who lived
Total suicides
and in women
Values for specific
meteorological factors, mean temp
specified in the study.
residence and place of death
temp
found in men
(P < 0.001)
excluded
were included but not
and correlation with
frequencies was
Female r ¼ 0.100
temperature
and suicide
Male r ¼ 0.197 (P < 0.001)
suicide and
Long data series of
Strengths:
weaknesses
Strengths and
between temperature
positive association
A preliminary
Results summary
suicide
coefficient, r for total
Cross correlation
Measure of effect
outside Tyrol were
184 females. All ages
day of death, place of
included; 518 males and
702 suicide cases were
specified
demographics not
female. Age and other
(128,322), both male and
All registered suicides
Population
whose records included the
between 1995 and 2000
All suicide deaths in Tyrol
Specific method of suicide
Total suicide frequency
Registered suicides
measured
Mental health outcome(s)
p u b l i c h e a l t h x x x ( 2 0 1 8 ) 1 e2 1
5
Please cite this article in press as: Thompson R, et al., Associations between high ambient temperatures and heat waves with mental health outcomes: a systematic review, Public Health (2018), https://doi.org/10.1016/j.puhe.2018.06.008
[27]
[26]
[25]
Ref# measured
Mental health outcome(s)
Population
and its possible
association with suicide
in the population of Baja
California Sur (BCS) 1985
e2008
Martagon, Vanessa
Diaz-Castro, Sara Cecilia
Zenteno-Savin, Tania
(2013)
approach
Jesus Celis Labrado-
Discovery in Database
Diaz, Y. (2016)
temperature increase
using the Knowledge
Narvaez, L. Hernandez-
de la Rosa, Alfredo de
Mexican population,
Rojop, I. E. Lopez-
Environmental
burns
temperature in a
Gaxiola-Robles, Ramon
from height, asphyxia and
environmental
from one or more locations
externaldunclear if data
seasonal average TMax,
Monthly averaged TMax and
and yearly
exposureddaily, monthly
death used as
station to the location of
1985e2008
California Sur, Mexico
population level, Baja
Ecological study at
Mexico 2005e2012
leveldTabasco State,
Total suicides
All registered suicides
Total suicides
firearms, poisoning, jumping
cutting blades, submersion,
hanging, using knifes or
specific methods of suicide:
suicidesdbroken down into
Castro, T. B. Juarez-
series, population
Fresan, A. Gonzalez-
from the closest weather
completed suicide and
Tovilla-Zarate, C. A.
Observational time
Association between
Fernandez-Arteaga, V.
from 14 years
both sexes and all ages
A total of 582 records,
more representative
higher during the hot weather than
R ¼ 0.64 (P < 0.01)
effect on suicide
temperature and its
miss short peaks in
Monthly temp may
exposure was derived
No clarity of how
ple size
Relatively small sam-
Weaknesses:
temperatures (P 0.01).
ambient defined with correlation R2 ¼ 0.68
commonly has high
Area of study
Strengths:
confounders
Unable to find info on
actual exposure
may not represent
perature as exposure
Use of external tem-
not considered
economic cofounders
Potential socio-
Weaknesses:
methods of suicide
Investigates different
rate and temp relationship well
during the cold. Hot season suicide
The suicide rate was significantly
is not stated.
hanging. Strength of this association
of exposure
station likely to be a 40 C (86e104 F) were related to males completing suicide by
the closest weather temperatures between 30 C and
Use of temp data from
Strengths:
objectives
Does not meet all
size or density
terms of population
not comparable in
The two settings are
actual exposure
may not represent
perature as exposure
Use of external tem-
Weaknesses:
Adjusts for sex
settings
Examines multiple
Strengths:
perature as exposure
Use of external tem-
ple size
Relatively small sam-
Weaknesses:
etc
Adjusts for education,
weaknesses
Strengths and
suicide: days with no rain and
environmental temperature and
An association was found between
Correlation coefficient
daysdmeasure unclear
males (n ¼ 1145) and females (n ¼ 212).
hot and very hot
age range of 10e76, both
A total of 1357 with an
External TMax & TMin, data
Number of suicides on
temperatures (32 C) rises approx. 2 the median temp (~22 C).
risk of suicide at warmer
Mississippi (1980e2006) data. 3355 total records.
All registered
females, and they were
(1986e2009) & Jackson,
and Hawkins Field Airport.
types
Suicides were more likely to occur
(22þ), males and
leveldToronto, Ontario
July, Aug and Sept.
higher than normal suicide totals for
RR but not stated in text.
during warmer conditions. Relative
derived from Toronto
in two different climate
Sheridan, S. C. (2014)
consisted of adults only
Total suicides
records were missing
TMin And monthly TAve
anomaly, all externally
temperature anomalies
Haney, C. R. Ellis, K. N.
series, population
‘Warm’ weeks were associated with
UNCLEARdchart with
warmer than the annual average
temperatures that are colder or
Results summary
only excluded if their
weekly TMax anomaly, weekly
suicide rates with
Schaffer, A. Levitt, A.
Study population
Registered suicide deaths
Measure of effect
Pearson International Airport
Weekly TAve anomaly,
Association of weekly
Dixon, P. G. Sinyor, M.
was different.
Observational time
Study design and setting
J. B. Roberts (2007)
Exposure
suicide and residence
Title
Inman &
of publication
Authors and year
Table 2 e (continued )
FAIR
FAIR
POOR
score
grade
Agreed
6 p u b l i c h e a l t h x x x ( 2 0 1 8 ) 1 e2 1
Please cite this article in press as: Thompson R, et al., Associations between high ambient temperatures and heat waves with mental health outcomes: a systematic review, Public Health (2018), https://doi.org/10.1016/j.puhe.2018.06.008
[30]
[29]
[28]
Temperature in East
Asian Countries: A
Time-Stratified Case-
Y. Guo, Y. L. Chen, B. Y.
Woo, J. M. Ebi, K. L.
(2016)
Crossover Analysis
Suicide and Ambient
Kim, Y. Kim, H. Honda,
number of stations per city
hourly data from unknown
externaldderived from
Daily average apparent temp,
specified geographical
within 25 km of
Japan and Taiwan
population level: Korea,
crossover study at
Multicountry case-
e2010
airport
stations at each regional
Health Prev Med
rates in Finland from
Partonen (2014)
study, population level
Ecological time-series
2005e2010
Total suicides
three East Asian countries
All registered suicides in
Total suicides
All recorded suicides
Total suicides
population level, Astana, Kazakhstan,
All registered suicides
Ecological time series,
regions of Finland 1974
seasons, and suicide
Kirsi Suominen, Timo
taken from local weather
Daily TMean, externalddata
from one location in Astana
TMin, apparent TMax, taken
Daily external TMean, TMax,
1974 to 2010. Environ
temperatures, thermal
Astana, Kazakhstan
(2013)
Haukka, Reija Ruuhela,
daily suicide counts in
Kosbayeva, A. Menne, B.
Local daily
air temperature and
Kozhakhmetova, G.
Laura Hiltunen, Jari
Associations between
Grjibovski, A. M.
included.
above 10 years were
and females, all ages
the study, both males
numbers not specified in
Total suicide record
within the study.
were not specified
and females. Age groups
study period, both males
recorded suicides over
A total of 10,802
in study
Other info not specified
A total of 685 records.
other weather factors.
TMean ¼ 1.37 investigated
temp to be
short-lived peaks in
associated with ambient temp in East Asia.
suggesting that suicide is positively Japan ¼ 4.5% per 4.2 C
differ significantly with sex or age, Korea ¼ 6.8% per 4.7 C
per 2.5 C
temperature and suicide did not
Taiwan ¼ 7.8% increase
in the study. Association between
suicide in all but two cities included
significantly with higher risks of
Higher temps were associated
Multiple variables tested.
inconsistent.
For women, the results were more
year analysis).
FAIR
FAIR
FAIR
(continued on next page)
actual exposure
rately represent
region may not accu-
from one station per
Use of external temp
Weaknesses:
cal power of results
set adding to statisti-
Relatively large data
long period of time
Data availability over
sex
Stratified by age and
populations
countries and
Examined multiple
Use of daily data
Strengths:
actual exposure
rately represent
region may not accu-
from one station per
Use of external temp
analysis
included in the
treatment not
tors or access to
Socio-economic fac-
Weaknesses:
Adjusts for sex
cal power of results analysis and autumnewinter half-
set adding to statistitemperature was (both year-round
Relatively large data
long period of time
Data availability over
Use of daily data
Strengths:
and temperature
ship of female suicide
examine the relation-
statistical power to
providing enough
therefore not
tion by male suicides,
ple size and domina-
Relatively small sam-
actual exposure
may not represent
perature as exposure
the lower the daily mean
rate ratio for men's suicide rate was,
the Helsinki region: the lower that
temperature and suicide in men in
association in daily mean
mean temperature
increase of each city's
corresponding to SD/2
Increase of suicide
P > 0.05
Apparent TMax ¼ 1.19
No clear picture. There was an
with outcome after controlling for
other weather variables:
perature allows
Use of external tem-
temp only significant relationship
months, holidays and
Apparent TMean ¼ 2.11
analysis. Mean and max apparent
Adjusted for years,
The use of daily tem-
Strengths:
Weaknesses:
with suicide numbers in crude
increase in temperature
TMax ¼ 0.91
All measures of temp are associated
RR of suicide per 1 C
p u b l i c h e a l t h x x x ( 2 0 1 8 ) 1 e2 1
7
Please cite this article in press as: Thompson R, et al., Associations between high ambient temperatures and heat waves with mental health outcomes: a systematic review, Public Health (2018), https://doi.org/10.1016/j.puhe.2018.06.008
[34]
[33]
[32]
[31]
Ref#
association with
climate: A population-
based study
Herng-Ching Tsai,
Shang-Ying Li, Chung-Yi
Chen, Chu-Chieh
temperature and suicide
mortality in Japan in the
Yasushi Honda, Masaji
Ono (2011)
associations with
violent and non-violent
suicide: A population-
Chin-Shyan Xirasagar,
Sudha Lee, Hsin-Chien
(2008)
based study
Seasonality and climatic
Lin, Herng-Ching Chen,
approach
series at the population level in Taiwan
weather data from 19
weather stations
Observational time
Monthly TAve,
externaldderived from daily
aggregated to monthly values
violent, non-violent
All daily suicides classed as
Non-violent suicides
analysis with a
semiparametric
Violent suicides
deaths and homicides.
comparators of accidental
violent, non-violent and
All daily suicides classed as
Total suicides
suicidesdMonthly data
Total suicides
e1995
level in Japan, 1972
series at the population
Observational time
Taiwan 1997e2003
population level in
All registered
Total suicides
All registered suicides
measured
Mental health outcome(s)
using time-series
to each region)
monitoring station assigned
Daily TMax, external (one
areas.
excluding mountainous
monitoring stations,
externaldaveraged over 19
Time-series study at
e2005.
level in Korea 2001
series, at population
Observational time
Study design and setting
confounding factors
presence of other
Relation between
Victoria Likhvar,
(2006)
Huang, Chung-Chien
Suicide rates and the
Lee, Hsin-Chien Lin,
e2005)
Monthly TAve,
station sites
temperature and suicide
mortality in Korea (2001
averaged from 64 weather
daily environmental
S. (2011)
Daily TMean, external
Exposure
Association between
Title
Kim, Y. Kim, H. Kim, D.
of publication
Authors and year
Table 2 e (continued )
temperature. (not clear at what temp the RR increases from) Males >65 years and those with less education had higher risk than females <65 and with higher
Suicides from Islands off the south coast were excluded due to significantly warmer climate
correlated with suicide rate.
R ¼ 0.376 (P < 0.001)
violent suicides were not significantly associated with temp.
Violent R ¼ 0.2012 (P < 0.001)
ages and both sexes.
(P > 0.05)
Non-violent R ¼ 0.0110
violent suicides, whereas non-
was significantly associated with
time trends, ambient temperature
After adjusting for seasonality and
discernible pattern.
methods of death did not display any
stronger than for non-violent. Other
violent showed a clear association,
violent and non-violent methods,
temperature.
and non-violent
temperature and violent
Correlation of
non-violent not given
Values for violent and
0). When this is broken down into
suggesting a short-term effect (lag of
between 0.0010 and 0.0435
estimated risk of suicide mortality
Total suicide range
found between increasing temp and
term trends, an association was
1 C increase in temperature per city.
After allowing for seasons and long-
Risk estimate for every
Use of external temp
rate.
exposure
represent actual
as exposure may not
Use of external temp
records
illness in suicide
No data on mental
miss peaks
aggregated data may
Use of monthly
Weaknesses:
Adjusts for sex
Large sample size
Strengths:
actual exposure
rately represent
region may not accu-
from one station per
Use of external temp
Weaknesses:
Use of daily data
Large sample size
Strengths:
effect on suicide
temperature and its
miss short peaks in
Monthly temp may
actual exposure
rately represent
region may not accu-
from one station per
Weaknesses:
age
Adjusts for sex and
ple size
Relatively large sam-
Strengths:
Unclear at what temp
actual exposure
not accurately reflect
External temps may
significantly associated with suicide
and unemployment, temp was still
Controlling for seasonality, trends
was significantly and positively
coefficients
included in the study, all
non-violent) were
(11,633 violent and 6497
A total of 18,130 suicides
included.
and sexes were
the study period, all ages
deaths recorded over
A total of 501,950 suicide
both males and females
groups 15e64 and >64,
separated into two age
recorded suicides,
Cross correlation showed that temp
education attainment
every 1 C increase in daily mean
attainment all included.
Cross correlation
creases above 1.0
increased daily suicide risk with
and education
A total of 18,083
mortality risk in-
that Koreans may have a 1.4%
both males and females,
Weaknesses:
(95% CI ¼ 1.010e1.017), indicating
groups, <65 and 65þ,
increasing temp
Estimate % risk with
Use of daily data
Strengths:
weaknesses
Strengths and
among all populations was 1.014
temperature and daily suicide count
association between daily
1 C increase above mean temperature
The relative risks (RR) of the
Results summary
% increase in suicide per
Measure of effect
stratified into two age
the study period,
recorded suicides over
A total of 49,451
Population
FAIR
FAIR
FAIR
FAIR
FAIR
score
grade
Agreed
8 p u b l i c h e a l t h x x x ( 2 0 1 8 ) 1 e2 1
Please cite this article in press as: Thompson R, et al., Associations between high ambient temperatures and heat waves with mental health outcomes: a systematic review, Public Health (2018), https://doi.org/10.1016/j.puhe.2018.06.008
[38]
[37]
[36]
[35]
obtained from one location
temperature and the
valuesdexternal. Weather stations were assigned to
weather on suicide rates
in the elderly in British
Columbia
Agbayewa, M. Wiggins,
S. (1999)
(Central England Temp). Two HW episodes identified
daily suicide counts and
temperature in England
and Wales
Kovats, R. S. (2007)
each geographical location.
drivers and suicide in
Australia: Bayesian
K. Tong, S. (2014)
spatial analysis
Externaldmean values for
Socio-environmental
Qi, X. Hu, W. Mengersen,
Monthly TMean,
within the study period
TMax and TMin), external
Relationship between
Page, L. A. Hajat, S.
Daily TMean (calculated from
stations were selected
Australia, 1986e2005
the population level in
Time-series analysis at
e2003
England and Wales 1993
the population level in
Ecological time series at
Total suicide
territories
eight Australian states and
All recorded suicides from
and meteorological data and suicide. age was not included.
segments and the association
Analysis was split into 4/5-year
suicide and M3 sociodemographic
suicide; M2 meteorological data and
M1 sociodemographic data and sexes. Information on
which included both
during the study period
Study used three different models:
why however.
on suicide numbers. Unclear as to
HW event did not display an effect
mortality of the same event. 2003
sets
FAIR
FAIR
FAIR
(continued on next page)
Relatively large data
factors
sociodemographic
Investigates effect of
Strengths:
chiatric diagnosis
not specify any psy-
Record of death does
exposures the 10.8% increase in all-cause
reflect experienced 41.5% suicide deaths, compared with
may not accurately
Use of external temps 1995 HW resulted in an increase of
females (3.5%).
encompassing 0), males (3.8%) and
Weaknesses:
extreme heat not significant and 95% CI
tained an episode of (5.0%), non-violent suicide (2.1%d
Study period contotal suicide (3.8%), violent suicide
every 1 C increase above 18 C for
increasing temp Use of daily data
and risk of death with
method of suicide
Investigates the
Estimated % increase in suicide for
Relative risk
temp, whereas non-violent suicides
above 18 C
exposure Strengths:
Non-violent suicide
A total of 45,293 suicides
show significant correlation with
risk for every 1 C rise
did not. Male and female suicides
Total suicides and violent suicides
% increase in suicide
represent actual
as exposure may not season.
group, in which suicide was linked to
short-lived peaks Use of external temp
No significant difference in younger
values may miss elderly suicide rate.
Use of monthly
ple size
Relatively small sam-
Weaknesses:
sex
Adjusts for age and
and temp
lation group suicide
ship of specific popu-
Examines relation-
Strengths:
associated with a 30% increase in the
2.5 C warmer current month is
preceding 3 months together with a
The combination of a 2.5 C cooler
months (RR ¼ 1.12, 95% CI 1.01e1.23).
temperature for the preceding 3
and with lower mean daily
2.5 C change in mean temperature)
1.05e1.28 for each
the current month (RR ¼ 1.16, 95% CI
higher mean daily temperature for
Elderly suicides increased with
suicide count
peaks in temp and
may miss short-lived
show comparable relationships.
and both sexes
which included all ages
during the study period
A total of 53,623 suicides
sample size was 3900.
years. Comparator
raise in temperature
violent suicide occurrence.
the past few weeks are predictors of
with changes in temperature over
Suggests that ambient temp together
Violent suicide
Total suicide
methods of suicide
suicidesddata included
All daily registered
Total suicide
their proximity. In total, 32
suicides in the 10e64 yrs age
males and females >65
suicides included both
>65 yrs age group, with a comparator of all registered
Study population of 758
All registered suicides in
exposure Use of weekly temp
represent actual
ambient temp, past temp (3-wk prior) and RH.
as exposure may not
suicide could be explained by
Use of external temp
Weaknesses:
methods of suicide
of temp with different
Examines correlation
Strengths:
that 16% of the variance of violent
Multiple regression analysis showed
Relative risk for 2.5 C
relationship between non-violent
non-violent P > 0.05
Non-violent suicide
group. Monthly values.
e1991
British Columbia 1981
population level in
and temp (r ¼ 0.32, P < 0.0001); no
(r ¼ 0.32, P < 0.0001)
correlation between violent suicide
Regression analysis showed a sig
violent suicide and temp
Correlation coefficient
suicides or homicides.
of cases not stated.
specified. Total number
sexes and age was not
Study included both
Violent suicide
homicide deaths
with a comparator of
into violent and non-violent
All suicide deaths categorised
each study region based on
aggregated to monthly
The effect of season and
Marion, S. A. Oluwafemi Time-series study at
Belgium 1979e1987
(e.g. TMax) is not stated. Data
suicide rate, ambient
D. Cosyns, P. (1994)
Daily TMax and TMin,
population level in
however the exact metric
rhythms in violent
F. Thompson, P. Peeters,
lightedark span
Observational study at
Weekly average of temp,
Synchronised annual
Maes, Michael De Meyer,
p u b l i c h e a l t h x x x ( 2 0 1 8 ) 1 e2 1
9
Please cite this article in press as: Thompson R, et al., Associations between high ambient temperatures and heat waves with mental health outcomes: a systematic review, Public Health (2018), https://doi.org/10.1016/j.puhe.2018.06.008
[40]
[39]
Ref#
ple size
rately represent
significantly and positively associated with temperature (b ¼ 0.0022, P < 0.001), with no
admission was included
episodes.
association seen for maniac
region may not accu-
unspecified bipolar admissions were
effect on suicide
temperature and its
miss short peaks in
Monthly temp may
actual exposure
from one station per
Use of external temp
Weaknesses:
confounding
Some adjustment for
weeks, only the first
associated with temperature
significantly and positively
depressive bipolar admissions were
Relatively large sam-
admissions occurred within 8
not specified
while age and sex were
selected for the study,
controlling for seasonality,
(b ¼ 0.0044, P < 0.001), and mixed or
depressive episodes and
diagnosis of manic episodes,
coefficient)
confounding Strengths:
episodes. Where multiple
Taiwan, 1999e2003
admissions were
Regression model showed,
mixed or unspecified
mountainous areas.
association with
All admissions with principal
Regression analysis (b
based study
stations, excluding
admissions and the
Ching (2007)
population level in
A total of 15,060
No adjustment for
No detail on diagnosis
isability of results
power and general-
ple size may limit
Relatively small sam-
episodes
short-lived hot
climate: A population-
averaged over 19 monitoring
bipolar disorder
Shang-Ying Lin, Herng-
Bipolar disorder admissions.
did the previous month's temp
catchment area.
Time-series analysis at
admission rates in either period nor
addresses in the unit's
Monthly TMean, external
Monthly tempera-
significantly with monthly
participants had
Seasonal variations in
investigated Weaknesses:
variations did not correlate
and 27 males). All
e1984
Lee, Hsin-Chien Tsai,
temp to be the analyses. However, temperature
included (39 females
Galway, Ireland 1980
tures do not capture
short-lived peaks in significantly with day length in both
both sexes were
psychiatric unit
variation.
perature allows only. Monthly temp correlated
admissions. All ages and
of mania or hypomania to the
The use of daily tem-
Strengths:
medications
nosis or prescribed
mental health diag-
psychiatric unit in
study period (1980e1984) and 1984
P < 0.05
patients, with 104
of patients with a diagnosis
exposure No information on
admitted to a
Analysis carried out twicedover full
cide incidence.
factors in the spatial pattern of sui-
important roles than meteorological
demographic factors played more
The results indicate that socio-
suicide
sociation between temp and
but there was no substantive as-
tors was similar to that in Model 1,
Impact of sociodemographic fac-
M3:
2.34% (2001e2005)
average of 2.27% (1996e2000) and
increased suicide rate by an
A 1 C higher yearly mean temp
reflect actual
may not accurately
with suicide from 1996 to 2005 M2:
Use of external temp
ployment positively associated
miss peaks
Monthly data may
Proportion of Aboriginal and Torres Strait Islanders and unem-
Weaknesses:
confounders
Adjusts for
weaknesses
Strengths and
M1:
in two models.
between temp and suicide assessed
Results summary
Correlation coefficients
Measure of effect
Sample size was 66
Population
Monthly rates of admission
measured
Mental health outcome(s)
county or city of Galway
series of patients
Observational time
Study design and setting
from one location in the
externaldUnclear if data are
the Prevalence of Mania
Fitzgerald, C. E.
Monaghan (1988)
Daily TMean,
Exposure
Influence of Climate on
Title
P. A. Carney, C. T.
of publication
Authors and year
Table 2 e (continued )
FAIR
POOR
score
grade
Agreed
10 p u b l i c h e a l t h x x x ( 2 0 1 8 ) 1 e2 1
Please cite this article in press as: Thompson R, et al., Associations between high ambient temperatures and heat waves with mental health outcomes: a systematic review, Public Health (2018), https://doi.org/10.1016/j.puhe.2018.06.008
[52]
[43]
[42]
[41]
which RR increases
(increasing). Neither temp nor humidity had a sig effect on predicting treatment for depression
depression or anxiety; history of depression; history of anxiety
One unit increase in temperature
and anxiety.
actual exposure modifying the effect of temp currently treated for
population
representative of relationship, with humidity only female. psychological distress; being
likely to be more suggesting that high temp drives this
representative of the
the last four weeks, presence
significant when combined
observations
study. Large cohort
over, both male and
generated from in situ
sets), all aged 45 and of high or very high levels of
study was
gridded daily temps
ple size
GOOD
FAIR
FAIR
FAIR
(continued on next page)
perature as exposure
Use of external tem-
Weaknesses:
confounding
Adjusts for
Spatial resolution however together heat alone is
from Australia
psychological distress over
participant: general
distress when considered separately,
Mental Health: Evidence
cohort study 45 and up
subgroup of a large
(53,144 complete data
external. Derived from
Relatively large sam-
Strengths:
actual exposure
may not represent
Use of external temp
provided in the data
No medical history
60,404 participants
interview/assessment),
predictors of high and very high
outcome study consisted of
Heat and Population
health were taken for each
Relationship between
randomly selected
weeks (at the time of
Humidity in the
Heat and humidity both significant
Increase % chance of Subgroup of cohort
during extreme heat conditions.
bipolar disorder episodes, especially
exacerbating the recurrence of
Weaknesses:
with temperatures above 24.0 C and heat plays an important role in
mate temperatures at bipolar disorder admissions increase
Bennett, C. M. (2016)
Five measures of mental
missing data Establishes approxi-
vulnerability to admission Results indicate that the risks of
limits effect of Females indicated greater
bipolar diagnosis.
enced exposure and
accuracy of experitemperature exceeded 30.7 C.
comparison group of non-
The Importance of Participants were a
1.23e1.84) when the daily mean
depressive), with a
potentially increases
exposure assessment (55th percentile), and 1.51 (95% CI:
unspecified maniac-
polation methods for
Use of spatial inter-
Use of daily data
sets
Relatively large data
when daily mean temp was >24 C
disorder admissions significantly
admissions with RRs of bipolar
ambient temperature and bipolar
was found between increasing
unspecified bipolar,
temperature points
admission at specific
ple size Strengths:
increased to 1.10 (95% CI 1.01e1.16)
comparators.
and 58,921 non-bipolar
with 5054 bipolar cases
ages and both sexes,
A positive correlation with 1-day lag
Relatively small sam-
peaks
miss short-lived
bipolar, mixed bipolar,
maniac bipolar, depressed
recurrent maniac disorders,
subcategories (single and
disorders, including
Population contained all
Ding, N. Berry, H. L.
Average TMax from previous 4
Psychiatric Inpatient
from the 1996e2007
from Taiwan.
disorder identified using Medical Claim data set
geographically.
temperature and bipolar
series analysis of data
psychiatric inpatients
interpolate estimated values
between ambient
H. J. (2013)
Retrospective time-
a national cohort of
stations used to spatially
A positive relationship
Sung, T. I. Chen, M. J. Su,
Daily TMean, externald372
stations
one or multiple weather Relative risk of
significantly correlated with temp.
above.
measures were derived from
Admissions of bipolar
depressed patients were not
sexes of 18 yrs and
Unclear if exposure
absolute minimal and
seasonal data may
Admission rates of unipolar
temperature
Use of monthly and
temperature (r ¼ 0.27, P ¼ 0.002)
which consisted of both
environmental
included in the study
maximal monthly temps
correlate with maximal
(2004)
maximal monthly environmental
Weizman, Abraham
of temp Weaknesses:
correlated significantly with mean
unipolar disorder)
maximum monthly temp,
minimum and mean e1991
disorder and 1036
spring/summer and
Popper, Miriam
hospitaldmonthly
Used multiple metrics
bipolar depressed patients
admitted to a psychiatric
patients increase during
Oded Hermesh, Haggai
admissions (4117 bipolar
setting in Tel Aviv, 1981
bipolar depressed
Shiloh, Roni Potchter,
Mean monthly temp, mean
Admission rates of
Shapira, Avraham unipolar depression
Strengths:
Mean monthly admission rates for
whole region
perature across the
rately represent tem-
region may not accu-
weather station per
Use of only one
exposure
represent actual
as exposure may not
Use of external temp
Weaknesses:
set
psychiatric hospital
Correlation
Association not statistically
specified.
bipolar)
A total of 6189
Tmin) in a small number of regions.
both sexes. Age not
depression (unipolar or significant for any measure.
mania and depression AND Tmax and
P > 0.05
study period including
diagnosis of mania or
All patients with bipolar or
e2002
Relatively large data
Use of daily data
admission frequency and temp (both
coefficient)
Republic of Ireland 1971
Strengths:
Only weak association seen between
Regression analysis (b
admissions during the
hospitals recorded with a
population level in the
A total of 34,465
All admissions to psychiatric
Time series at the
Time-series analysis in a
A number of temp metrics:
region selected
admissions to
psychiatric hospitals
externaldone station per
and affective disorder
A. O'Callaghan, E. (2014)
Daily TMax and TMin,
Daily weather variables
McWilliams, S. Kinsella,
p u b l i c h e a l t h x x x ( 2 0 1 8 ) 1 e2 1
11
Please cite this article in press as: Thompson R, et al., Associations between high ambient temperatures and heat waves with mental health outcomes: a systematic review, Public Health (2018), https://doi.org/10.1016/j.puhe.2018.06.008
[48]
[54]
Ref#
and psychiatric
emergencies
Collsamata, A. Joaquim,
I. Perez, Y. Tricio, O.
The effect of heat waves
on mental health in a
temperate Australian
city
Hansen, A. Bi, P.
Nitschke, M. Ryan, P.
Pisaniello, D. Tucker, G.
(2008)
Ibiza, C. (2009)
Meteorological factors
Title
Garcia, F. i Boada, S.
of publication
Authors and year
Table 2 e (continued )
significance. The main conclusion of the study is that there is a direct relationship
alcohol) and other nonspecific mental health illnesses.
four continuous days exceeding 40 C. The highest recorded maximum temperature was 44.3 C. There appeared to be a threshold temperature of 26.7 C (P ¼ 0.001),
with principal discharge diagnosis or cause of death was attributed to mental, behavioural and cognitive disorders
e1.102)
Schizophrenia ¼ 1.034 (0.969
misuse ¼ 1.005 (0.913e1.105)
due to psychoactive substance
Mental and behavioural disorders
Dementia ¼ 1.174 (1.017e1.355)
Organic ¼ 1.213 (1.091e1.349)
MBD ¼ 1.073 (1.017e1.132)
below:
days compared with non-HW days
IRR (95% CI) of hospitalisation on HW
Morbidity-specific conditions and
markedly.
disorder (MBD) admissions increased
above which mental and behavioural
January 2006 with temperatures over
admissions and mortalities
and 30th June 2006
including an extreme heat event in
ranging in duration from 3 to 8 days,
31 and 29 heat waves, respectively,
periods
deaths
admissions and 2599
psychiatric hospital
Australia. Both counts of
mortality data for south
Adelaide Metro area and
between 1st July 1993
metropolitan area
series, Adelaide
During the study periods, there were
in regards to humidity
relationship could also be observed
hospital emergency department. A
number of patients attending the
temperature increase and the
Study also incorporates HW
representative, external.
weather station deemed to be
A total of 171,614
Incidence rate ratios
comparisons did not reach statistical
stress, substance misuse (not
All hospital admissions in
(non-linear form). The remaining
psychoses, schizophrenia,
Observational time
Relatively short study
was also statistically significant
personality disorders,
Daily TMax from one central
investigated Weaknesses:
humidity and number of daily visits
mood disorders, neuroses,
between environmental
temp to be
(TMIN). The relationship between
anxiety, eating disorders,
sent actual exposure
not accurately repre-
from one station may
Use of external temp
statistical power.
nosis may limit
comes in some diag-
Small number of out-
Weaknesses:
Use of daily data
diagnosis
ship of multiple
Examines relation-
Strengths:
clear
Objective of study not
exposure was derived
No clarity of how
period
short-lived peaks in
ages
(TMAX) and minimum temperature
perature allows
The use of daily tem-
Spain. Clinical diagnosis of
endpoints of medium temperature
population Strengths:
2004e2005
in text, only significance
male and female and all
significant for the meteorological
The results were statistically
late to younger
sults may not trans-
stations
no coefficients provided
were included, both
questionnaire
effect of heat more than doubled to
Study population re-
self-reporting via
percentile within the study, the 0.5% (P < 0.001)
tures usually occur Potential bias from
When humidity rose to the 99th
before high tempera-
very high distress, respectively.
collected in spring,
(P < 0.001) increased risk of high or
Majority of data
weaknesses
Strengths and
with a 0.2% (P < 0.001) and 0.1%
and vapour pressure was associated
Results summary
(TMED), maximum temperature
Correlation e however
Measure of effect
A total of 3048 patients
Population
Reus (Psychiatric hospital)
the Hospital Pere Mata de
emergency department of
All attendees of the
measured
Mental health outcome(s)
department in Spain,
hospital emergency
series in a psychiatric
Retrospective time
Study design and setting
derived from a number of
from one weather station or
TMin, unclear if data is taken
Daily external TMean, TMax,
Exposure
FAIR
FAIR
score
grade
Agreed
12 p u b l i c h e a l t h x x x ( 2 0 1 8 ) 1 e2 1
Please cite this article in press as: Thompson R, et al., Associations between high ambient temperatures and heat waves with mental health outcomes: a systematic review, Public Health (2018), https://doi.org/10.1016/j.puhe.2018.06.008
[21]
[51]
Admissions for Mental
Disorders in Northern
Vietnam
Giang, Gunnar Kullgren,
Maria Nilsson (2016)
and substance misuse
(2012)
Joacim Rockl€ ov, Kim Bao
psychosis, dementia
Howard, Louise M.
Heat waves and Hospital
deaths in people with
Shakoor Kovats, R.
Phan Minh Trang,
Temperature-related
Page, Lisa A. Hajat,
at both national and
nationally available mortality data
England and for those with a primary
compared with non-heat wave periods. Relative risks were
spectated by ICD-10 codes (F0-3; F4-6; F7-9;
somatoform, disorder, stress,
heat waves of at least 3 days when
mental disorders for more persistent disordersdResults
admission in the whole group of Organic mental
Heat waves increased the risk for of 0, 3 & 7)
increase for all exposure metrics.
Total admissions RR ¼ 1.02 per 1 C
risk.
demonstrated greatest mortality
diagnosis of substance misuse
patients and those with a primary
temperature distribution. Younger
the 93rd percentile of the annual
1 C increase in temperature above
death of 4.9% (95% CI 2.0e7.8) per
FAIR
FAIR
(continued on next page)
of HW duration
Examine the impact
riods of HW
Examines specific pe-
temperature
Use of daily
Strengths:
exposure
reflect experienced
may not accurately
Use of external temps
numbers
limitations of
cause of death due to
Does not investigate an overall increase in the risk of
Patients with mental illness showed
associated with higher risk of death.
confounding Weaknesses:
and antipsychotics was also
misuse. Prescription of hypnotics
Adjusts for
mary care data with Southern and Eastern parts of diagnosis of alcohol or substance
Cross linking of priyounger patients, those living in
different diagnosis
Examines RR of This effect was more marked for
temperature thresholds.
above regionally defined
heat wave duration (lag
increase in temp AND
Examines relative risk
alcohol (RR ¼ 1.08) or substance misuse (RR ¼ 1.20) experienced
regional level
Use of daily data
significant heat-related mortality
Strengths: (RR ¼ 1.02), dementia (RR ¼ 1.03),
e4.138) Primary care patients with psychosis
disorders, schizophrenia,
ages and both sexes.
which comprised of all
over the study period
Relative risk per 1 C
distribution
percentile annual
increases above the 93rd
Relative risk per 1 C
disabilities/autism, mood
delirium, learning
disorders include dementia,
admissions. Specific
A total of 21,433 patients
groups: <65 and 65þ.
misuse
All mental health disorder
and stratified into two
including both sexes
during the study period
A total of 22,562 deaths
misuse and other substance
psychoses, dementia, alcohol
with primary diagnosis of
Registered deaths of people
days.
Vietnam, 2008e2012
mental hospital,
admissions to Hanoi
Time-series study of
England 1998e2007
population level in
Time series at
HW days (>35 C) and non-HW
comparisons made between
conditions identified and
external temps. HW
Daily TMax, TAve, TMin,
external
region in England and CET,
Daily TAve for every gov
Schizophrenia ¼ 2.079 (1.045
e7.155)
substance misuse ¼ 3.098 (1.342
Disorders due to psychoactive
e21.232)
Dementia (15e64) ¼ 5.058 (1.205
e4.922)
All MBDs (65e74) ¼ 2.395 (1.165
below:
days compared with non-HW days
(95% CI) of hospitalisation on HW
Mortality-specific conditions and IRR
Senility ¼ 2.366 (1.200e4.667)
e1.489)
Alzheimer's disease ¼ 1.154 (0.894
development ¼ 1.641 (1.086e2.480)
Disorders of psychological
e2.26)
Mental retardation ¼ 0.737 (0.268
factors ¼ 0.875 (0.678e1.130)
bances and physical
ated with physiological distur-
Behavioural syndromes associ-
Neurotic ¼ 1.097 (1.018e1.181)
e1.185)
Mood disorders ¼ 1.091 (1.004
p u b l i c h e a l t h x x x ( 2 0 1 8 ) 1 e2 1
13
Please cite this article in press as: Thompson R, et al., Associations between high ambient temperatures and heat waves with mental health outcomes: a systematic review, Public Health (2018), https://doi.org/10.1016/j.puhe.2018.06.008
[55]
[53]
Ref#
days. Data from weather
stations in the study regions
were spatially interpolated
before aggregating to larger
and humidity and visits
to mental health
emergency departments
in Quebec
Gosselin, Pierre (2012)
selected
psychosis admissions to
psychiatric hospitals
e one station per region
weather variables on
A. O'Callaghan, E. (2013)
Daily TMax and TMin, external
The effects of daily
McWilliams, S. Kinsella,
geographic areas.
external, and lag of 0 and 3
ambient temperature
Martin Ouarda, Taha B.
Daily TMean, Temp range,
Exposure
Relationship between
Title
Vida, Stephen Durocher,
of publication
Authors and year
Table 2 e (continued )
e2002
Republic of Ireland 1971
population level in the
Time series at the
Canada
regions of Quebec,
visits in three specific
emergency department
the population level of
Observational study at
Study design and setting
other psychoses
schizoaffective disorder and
diagnosis of schizophrenia,
hospitals recorded with a
All admissions to psychiatric
include admissions).
department visits (may not
All mental health emergency
sexes. Age not specified.
period including both
admissions during study
A total of 47,789
both sexes
from 15 y and above and
which covered all ages
over the study period,
patients are included
may not represent actual exposure
respectively. Admissions for mental disorders increased among men, residents in rural communities and the elderly population during heat waves. The groups of organic mental disorders, including symptomatic illnesses (IRR ¼ 3.62 per 7-day HW) and mental retardation (IRR ¼ 1.68 3day HW) had increased admissions during heat waves.
gender. See study for details Mental retardationdResults spectated by ICD-10 codes (F70-71; F72-73; F70-79), age group (0e17; 18e40 41þ), location (urban/rural) and gender. See study for details
provided
temps
e1.09). In most cases, the IRRs at 22.5 C and at 25 C were
however not statistically significant. The degree to which maximum temperature influenced admissions
b significant for three regions at lag 0, 2 & 7 (P < 0.05)
as exposure may not represent actual exposure
predictive of admissions at lag 0; however, the size of the time series suggests that the correlation is
undertaken further limiting power.
regions meant analysis could not be
daily time series in 5/12 study
Issues with autocorrelation from
whole region
perature across the
rately represent tem-
region may not accu-
weather station per
Use of only one
Use of external temp
highlight temperature as a factor
probably very weak.
Weaknesses:
ing weather variables
Adjust for confound-
set
Relatively large data
Use of daily data
Strengths:
actual exposure
may not represent
Use of external temp
diagnosis
Lack of detail on
Weaknesses:
Use of daily data
department data
Use of emergency
time-series regression did not
was difficult to ascertain because
temperature on admissions,
Detected a weak influence of air
below 65 years.
increases were seen only for those
geographic areas; in one area, the
for those aged 65 and older in two
both for persons under age 65 and
Significant increases were seen
homogeneous areas of Qu ebec.
in three meteorologically
mental and psychosocial problems
emergency department visits for
primarily monotonic increases in
with statistically significant and
Higher temperature was associated
(data not provided).
coefficient)
Regression analysis (b
focuses on higher
increasing temperature (IRR 1.00
thresholds
significantly higher than at 20 C
Excludes winter and
increased monotonically with
above particular
Large data set
For all geographical areas, all IRRs
Incidence rate ratios
Strengths:
Use of external temp
Weaknesses:
confounding
Adjusts for
disorders
ber of specific
Investigates a num-
(1.005e1.31) and 1.36 (1e1.90) for a 1, 3- and 7-day heat wave,
Estimation of risk
weaknesses
Strengths and
(urban/rural) and
estimated at 1.04 (0.95e1.13), 1.15
Results summary
41e60; 61þ), location
F0-9), age group (0e40;
Measure of effect
mental illness not specified.
A total of 347,552
Population
substance misuse and other
measured
Mental health outcome(s)
FAIR
FAIR
score
grade
Agreed
14 p u b l i c h e a l t h x x x ( 2 0 1 8 ) 1 e2 1
Please cite this article in press as: Thompson R, et al., Associations between high ambient temperatures and heat waves with mental health outcomes: a systematic review, Public Health (2018), https://doi.org/10.1016/j.puhe.2018.06.008
[47]
[45]
[44]
groups in terms of heart rate or blood
before and after tests
psychotic patients admitted to same ward during study period with either major depressive disorder, psychosis secondary to organic aetiology (e.g. substance abuse) or combination
calculated, weekly mental status using PANSS (Positive and Negative Syndrome Scale), comparisons using Student t-test for differences in duration of hospitalisation; Pearson test for correlation analysis.
consisted of all overtly
Daily body discomfort index
comparator group
ages from 18 to 80. A
of both sexes and all
study which comprised
exacerbation of the disorder.
during the study period in a
schizophrenia hospitalised
DSM-IV criteria for
longitudinal study
6th August 2003
Israel 1st January 2003 to
No significant difference between
and body weight measured
to an acute psychotic
mean value used.
12:00 in three locations, with
group during rest phase.
rate monitor, blood pressure
were included in the
signalling lower on comparator
both rectal and skin, heart
All subjects who met the
(P < 0.001). Rectal temp fell,
Temperature monitors for
severity of symptoms.
ward temp of 19 C would improve
Authors suggest that an optimum
leading to worsening of symptoms.
by ward temp; i.e. higher ward temps
patients' symptoms are modulated
Authors suggest that schizophrenic
finding.
patients. BDI did not affect this
24 C.
FAIR
FAIR
POOR
(continued on next page)
actually fall below
ward temp did not
mum temp of 19 C as
thors derive the opti-
Unclear how the au-
comparator group
Small sample size of
ple size of subjects
Relatively small sam-
Weaknesses:
for comparison
Use of control group
where appropriate
Use of daily data
actual exposure correlation amongst non-schizoid
representative of (r ¼ 0.52e0.64, P < 0.0002), with no
likely to be more
Use of indoor temp,
Strengths:
chological symptoms
exasperation of psy-
Does not examine
results
generalisability of
statistical power and
control size may limit
Small sample and
Weaknesses:
with temp in schizophrenic patients
PANSS score significantly correlated
between two diagnostic groups.
Admission duration no different
or physiological reasons.
if this effect is due to medication
medicated counterparts. Unclear
body than their healthy, non-
able to conduct heat from the
neuroleptic treatment are less
male patients on long-term
Results suggest schizophrenic
pressure.
comparator during exercise
mental health issues.
Use of control group
measurements
outcome
the patient group than in the
cise exposure and
(P < 0.001). Rate of rectal temp
Controlled and pre-
Strengths:
robust
for choice not very
ture data and reasons
Source of tempera-
results
generalisability of
power of results and
have an impact on
size included, may
Unclear population
exposure was derived
No clarity of how
Weaknesses:
and altitude
Adjusted for latitude
Use of daily data
Strengths:
group than in the comparator
significantly higher in the patient
eight matched men with no
Correlation coefficient
experiment time period
pressure change over
heart rate and blood
increase was significantly higher in
A total of 247 patients
cases
controls matched to
age of 31.5, eight
Rate of skin temp increase was
to higher outcome risk
Indicating that higher temps lead
months. Control group of
for at least 12 consecutive
maintained on medication
schizophrenia who had been
inpatientseA
severity of symptoms in
(2007)
psychiatric hospital in
Ecological study set in a
Israel.
climate chamber in
study performed in a
Matched caseecontrol
closed psychiatric ward due
temperature and the
Weizman, Abraham
temp measured every day at
Daily TMean, indoors e ward
sessions (indoors) for 50 mins
range: r2 ¼ 0.95; P < 0.01
Morbidity risk % and Temperature
outcomes and measures of temp
and positive correlation between
IPSS) show statistically significant
Data from both studies (DOSMD &
mean temp: r2 ¼ 0.75; P < 0.05
Skin temp, rectal temp,
Correlation coefficient
Mean overall outcome score and
Seven males, average
ages etc.
including total numbers,
of study subjects
Unclear as to the details
Mental Disorders
Seven men with
studies.
is taken from two other
assessment methods as data
unclear on diagnosis and
Schizophrenia, however
Outcome of Severe
(b)Determinants of
Study on Schizophrenia
(a)International Pilot
from two studies:
secondary use of data
Multi centre cohorts;
schizophrenia
between ward
antipsychotic drugs
Munitz, Hanan (2000)
Hanan Stryjer, Rafael
maintained with
Weizman, Abraham
A significant correlation
schizophrenia
Manaim, Hillel
Shiloh, Roni Munitz,
patients with chronic
Roni Epstein, Yoram
communication’, all external
of Schizophrenia
2 40 C in climate chamber
provided by ‘personal
Incidence and Outcome
Heat intolerance in
clear and in some cases
Temperature to the
Hermesh, Haggai Shiloh,
however locations are not
Environmental
Murray (1992)
Daily TMax, TMin, TMean,
The Relationship of
Sunjai Gupta & Robin M
p u b l i c h e a l t h x x x ( 2 0 1 8 ) 1 e2 1
15
Please cite this article in press as: Thompson R, et al., Associations between high ambient temperatures and heat waves with mental health outcomes: a systematic review, Public Health (2018), https://doi.org/10.1016/j.puhe.2018.06.008
HW days of the study period.
admissions due to
Indoor Air Temperature
and Agitation of Nursing
Home Residents With
Dementia
Federico Tartarini, Paul
Cooper, Richard
Fleming, and Marijka
Batterham (2017)
dementia in Madrid
identified, with a total of 216
factors and hospital
Diaz, J. (2017)
Ecological time series,
elevated temperature
factors
frequency scores significantly increased when indoor temp
coefficientdcorrelation values not provided
CMAI overall disruptiveness scores were found to increase significantly when average temps deviated from 22.5 C. Deviation from optimal
legal guardian. Sample consisted of both sexes with an age range of 61 e92.
both higher and lower temps.
level of disruptiveness increased at
temps showed both behaviour and
frequency scores
consent granted from
Short study period
Small sample size
and locations
patient movements
nightly assessment of
Recall bias for fort-
Weaknesses:
confounders
Adjusts for
temperature used
Precise metric of
exposure
rooms into overall significantly predicted higher CMAI
Assessment Scales and
spent in each room.
average temp in each room
spent in different
Incorporating time
temperature
Use of indoor
Strengths:
exposure
represent actual
as exposure may not
Use of external temp
exposed to temps above 26 C
deviated from 22.6 C (P < 0.01)
Agitative behaviours via CMAI total
34 C
lag ¼ 16% per 1 C increase above
AR of DDE admissions with a 1-day
Weaknesses:
day ¼ 1.19 per 1 C increase above 34 C
environmental
RR of DDE admissions with a lag of 1
Adjusts for other
The percent of time patients were
spent in different rooms.
environmental variables tested (O3 and PM).
Examines periods of
temperatures heat was greater than the other
at specific
admissions on day 2. The effect of
Estimates relative risk
set
Relatively large data
day 1 leading to increase in
had a slight lag time of 1 day, Tmax on
Impact of heat on DDE admissions
admissions start to rise.
Use of daily data
showed a clear point (34 C) at which
temps experienced Strengths:
Admissions plotted against TMax
Correlation
threshold
increase above
Relative risk with 1 C
of Psychogeriatric
higher than 4 on the
at least 3 months, scored
had been a resident for
A total 21 residents who
sex not specified
study period, ages and
admissions over the
Inventory (CMAI) and time
Cohen-Mansfield Agitation
behaviours. Assessed via
Dementia and agitated
dementia
of persons with diagnosed
A total of 1175
be representative of
as exposure may not
Use of external temp
short peaks
peratures misses
Use of monthly tem-
P < 0.001). All other climate variables did not show a significant correlation
Weaknesses:
group
Use of comparison
ple size
Relatively large sam-
Strengths:
weaknesses
Strengths and
max monthly temp (r ¼ 0.35,
and positively correlated with mean
Admission rates were significantly
periods (spring and summer).
patients were higher in warm
Admission rates of schizophrenic
Results summary
cognitive decline subject
setting in Australia
study, in a nursing home
prospective cohort
Longitudinal
Spain 2001e2009
admissions in Madrid,
hospital emergency
Daily emergency admissions
time period.
Correlation coefficient
Measure of effect
and the time the subject
TAve calculated from the
installed on internal walls.
indoordiButton sensors
Daily (dry bulb)TAve,
number of HW periods are
between environmental
Carmona, R. Ortiz, C.
Daily TMax, external. A
Short-term association
Linares, C. Culqui, D.
comparison group.
facilities during the same
with 3022 patients
over the study period,
criteria are admitted
meeting inclusion
patients admitted to same
of schizophrenia, compared
Aviv with acute exacerbation
seven psych hospitals in Tel
included in the
derived
Israel, 1981e1991
psychiatric hospital in
A total of 30,592 patients
Population
schizoaffective disorder
to psychiatric hospitals
Popper, Miriam
externaldunclear how the
Tel Aviv weather data were
All patients admitted to
measured
Mental health outcome(s)
with all patients diagnosed as
schizophrenia patients
Hermesh, Haggai
Ecological study set in a
Study design and setting
(2005)
admission rates of
Avraham Potchter, Oded
Monthly TMean, TMax, TMin,
Exposure
Weizman, Abraham
Effects of climate on
Title
Shiloh, Roni Shapira,
of publication
Authors and year
CDC, Centre for Disease Control; HW, heatwave; IRR, incidence rate ratio; RR, relative risk; Tmax, maximum temperature; Tmean, mean temperature; Tmin, minimum temperature.
[50]
[49]
[46]
Ref#
Table 2 e (continued )
FAIR
FAIR
FAIR
score
grade
Agreed
16 p u b l i c h e a l t h x x x ( 2 0 1 8 ) 1 e2 1
Please cite this article in press as: Thompson R, et al., Associations between high ambient temperatures and heat waves with mental health outcomes: a systematic review, Public Health (2018), https://doi.org/10.1016/j.puhe.2018.06.008
p u b l i c h e a l t h x x x ( 2 0 1 8 ) 1 e2 1
One study described the risk of both hospitalisation and mortality among individuals diagnosed with schizophrenia on heat wave days compared with non-heat wave days (defined at >3 consecutive days at or above 35 C).48 Risk of hospitalisation was not significantly elevated on heat wave days; the risk of mortality more than doubled (RR 2.08, 95% confidence interval 1.05e4.14, P < 0.05).
Organic disorders including dementia Two studies found an increased risk of admission for organic mental disorders during heat waves, using geographically appropriate temperature thresholds (RR ¼ 1.213 during heat waves of >3 days;48 RR ¼ 3.62 during heat waves of >7 days21). Two studies described an increased risk of hospital admission for dementia above similar heat wave threshold temperatures.48,49 One study found that agitation and disruptiveness of nursing home residents with known dementia increased significantly with temperature.50 Two studies described the relationship between high temperatures and dementia mortality.48,51 One found that the risk of mortality increased significantly on heat wave days (RR ¼ 5.058),48 and the other found that the risk of mortality increased above 18 C (RR, 1.03 per 1 C increase).51
Alcohol and substance misuse One study found no significant increase in the risk of hospitalisation due to substance use during heat waves.48 The same study found that the risk of mortality for the same population increased significantly during heat waves in females of all ages.48 A second study found an increased risk of mortality with increasing temperature for those with a primary diagnosis of both alcohol and substance misuse51 (Table 2).
Other mental health outcomes and service use Three studies that described the relationships between temperature and multiple mental health outcomes have been summarised previously.21,48,51 These studies also provided single-study evidence on the increased risk to mood, neurotic disorder, psychological development48 and mental retardation on heat wave days21 and the increasing risk of mortality from psychosis with rising temperature.51 One study of those aged above 45 years in Australia found that an increase in temperature and humidity by one unit ( C or hPa, respectively) led to a significant increase in the proportion of the population reporting high or very high distress.52 One study described all-cause risk of emergency department visits to a psychiatric hospital at three different temperature points, finding that risk increases significantly as temperature increases.53 Another study also found a significant linear relationship between psychiatric emergencies and increasing temperatures.54 A third study detected a weak influence of air temperature on admissions which was not statistically significant.55 Two articles described the risk of hospital admission with a mental health disorder during heat waves compared with
17
non-heat wave periods,21,48 using different definitions of heat waves. One article defined a heat wave as 3 consecutive days when TMax reached or exceeded 35 C,48 whereas the other used three definitions; a single day, at least 3 consecutive days and at least 7 consecutive days when TMax reached or exceeded 35 C.21 Both the articles found that the risk of admission to a psychiatric hospital with a mental health disorder increased significantly on heat wave days. One study described a significantly increased risk of mortality among those diagnosed with mental health disorders when temperatures exceeded the 93rd percentile of the annual temperature distribution.51 A second study described the significantly increased risk of mortality during heat waves among those with previously diagnosed mental health disorders.48
Discussion This review has summarised the evidence of association between higher temperatures and mental health outcomes, describing the size and direction of association and correlation where available. To the best of our knowledge, this is the first attempt to summarise the available evidence on this area and has corroborated previous studies and commentaries.7 Higher temperatures increased the risk of all mental health outcomes for which evidence was found. Significant gaps exist in evidence for the impact of temperatures on many mental health conditions. Most articles used external temperature as the exposure measure, measured at monitoring stations that (depending on the country) may be many miles from the location of individuals in the study, who thus experience very different climatic conditions.56 External temperatures are thus a limited proxy for individual exposure. Furthermore, people spend the majority of their time indoors.57 Indoor temperatures are highly variable from location to location and are not recorded or available in most dwellings. Similar problems have been described in the design of studies of the impact of heat on physical health. In general, it is thought that such methodological challenges may dilute observed associations between high temperatures, morbidity and mortality with the true effect being greater than that described.58,59 Two articles used indoor (ward) temperatures as the measure of exposure for inpatients and suggested that high temperatures may worsen symptoms for those diagnosed with mental health disorders.47,50 These findings will be of relevance to service providers to ensure that the risk of harm to patients during periods of hot weather is minimised. Suicide prevention is a World Health Organisation global priority.60 The UK recently published its third progress report on suicide prevention.61 Although there are several studies on this topic identifying a positive association, they are small in scale and do not allow for (for example) the assessment of risk at population level such as quantifying the population attributable fraction of suicide due to high temperatures. The relationship between suicide frequency (especially violent suicide) and high temperatures was consistent.22e38,62,63 The impact was greatest among men in three of four studies that described risk by sex. This is in keeping
Please cite this article in press as: Thompson R, et al., Associations between high ambient temperatures and heat waves with mental health outcomes: a systematic review, Public Health (2018), https://doi.org/10.1016/j.puhe.2018.06.008
18
p u b l i c h e a l t h x x x ( 2 0 1 8 ) 1 e2 1
with the generally elevated risk of suicide in men compared with women.64 This finding may have implications for suicide prevention plans as temperature may be a risk factor to consider in these programmes. All studies on suicide risk are observational; it is more difficult to ascertain causal effects using such designs as they simply measure associations. Despite this, robust conclusions can be made about the sum total and implications of the evidence. One article37 explores this theme, postulating sociological, biological and psychological mechanisms of plausibility, including excess alcohol use, an interaction between high ambient temperature and serotonin or a previously identified link between such temperatures and disinhibition and aggression. Average temperatures are forecast to increase with increased frequency and intensity of high temperature episodes.3,10 In the absence of any population adaptation, annual heat-related deaths are predicted to rise by 257% by 2050 due to climate change, from an annual baseline of around 2000.65 None of the included studies looked specifically at climate change, whereas the findings of our review support the assertion that the risk of suicide and other mental health outcomes is likely to increase in line with climate projections. For commissioners and managers of mental health services, all studies that examined hospital admissions or emergency department visits found significant increases with rising temperatures, particularly during heat wave periods.21,39e43,46e49,52e55,66e68 During such periods, symptoms may be exacerbated in those already diagnosed with a mental illness, leading to increased service use. Three studies investigated the impact of high temperatures on patients prescribed medication to control mental disorders. Two found that patients on medication had reduced thermoregulation ability,45,66 whereas the third found that prescription of hypnotics and antipsychotics was associated with higher risk of death at high temperatures.51 The impacts of temperature on medication side-effects may be an important factor when reviewing treatment in hot weather and is consistent with the recommendations set out in Heatwave Plan for England.69 Public health authorities around the world already use allcause mortality temperature relationships70,71 to define trigger thresholds for action in adverse weather alert systems.69,72 Understanding the thresholds at which adverse mental health impacts occur must be considered in such systems to reduce harm to health. Twelve articles described temperature thresholds at which the risk of mental health outcome of interest increased.25,26,31,36,37,43,47e51,53 Although these temperature thresholds were clearly stated, they vary by country from 18 C in England and Wales37 to 35 C in Australia and Vietnam,21,48 with in-country variation between studies. The article identifying a threshold of 18 C37 used a 10-year time-series regression analysis to find that the risk of suicide was lowest at 18 C, with rises above and below that temperature (a ‘hockey stick’eshaped distribution). It is not infrequent for mean outdoor temperatures in England to be above 18 C; such values are being recorded on 222 days over the 11-year period of the study. As previously mentioned, this article suggested biologically plausible mechanisms of causation, such as increased alcohol use, which might in turn
result in an increased propensity for violence including suicide. The authors observe that it is unlikely that deaths from suicide are merely brought forward in time during hot weatherdreal additional suicides probably occur when temperatures are high and may be linked to indoor overheating.37 Considerable heterogeneity between study methods, settings and analyses limits the collective interpretation of the studies found. For example, the periods of both exposure and outcome varied across studies with daily, weekly and monthly temperature measures used as well as different periods of aggregating and analysing outcomes. Some studies allowed for lag periods, that is, the variable duration between exposure to high temperatures and development of the health outcome under consideration, whereas others looked solely at temperature on the day of admission or diagnosis. A common limitation for weekly and monthly periods of measure was that short-lived periods of high temperature may have been missed. This may lead to underestimation of the health effects seen in the included studies. Future studies should strive to use daily data so that this potential source of bias is minimised. The range of reported measures of effect precludes direct comparison. Although studies have reported RR estimates, the degree by which the risk increases above a threshold temperature differed among many studies. For example, some studies reported RR per 1 C increase above 18 C, whereas others report RR per half a standard deviation change in temperature above the mean.30 It would be helpful if future research in this area report measures of effect in the same units so that direct comparisons can be made and metaanalysis be performed. Only five studies reported no significant association between mental health outcomes and temperature.24,29,39,41,55 This could represent publication bias or be due to the relatively limited number of published studies on all outcomes bar suicide. Within these studies, there are a number of potential confounders that may also play a role in this, for example, temporal resolution of exposure metrics used or low sample size, etc. Alternatively, this may provide a true representation of the relationship between high temperatures and the outcome of interest. Only more research in this area will provide further insight. Cases presenting to secondary care or emergency departments are likely to represent the most severe end of the disease spectrum. It is reasonable to suggest that there is a large proportion of the population with mental health disorders who are impacted by high temperatures but not presenting at secondary care facilities; such cases may seek help informally or from primary care. Studies that examine just secondary care data are thus at risk of underestimating the impact of temperature on mental health. Study sizes also differed across studies and outcomes of interest. For example, studies that investigated suicide and temperature ranged from 582 records27 up to 501,950.33 While similar ranges of sample size existed across outcome groups, the general direction of association observed was consistent. This suggests that the associations described here are reflective of the true relationship under investigation. Half of the included studies examined suicide. The remainder examined other mental health illnesses; in some
Please cite this article in press as: Thompson R, et al., Associations between high ambient temperatures and heat waves with mental health outcomes: a systematic review, Public Health (2018), https://doi.org/10.1016/j.puhe.2018.06.008
p u b l i c h e a l t h x x x ( 2 0 1 8 ) 1 e2 1
cases, only one article examined a particular outcome, such as neurotic disorder. No evidence was found for several outcomes included in the search terms which may have been expected. Further research would benefit from looking at potential associations between some of these outcomes and high temperatures. Study quality was generally ‘fair’ (88.5%) with just one study rated ‘good’. Although each study has individual limitations, as a whole the generally, positive associations throughout the body of evidence support a true effect.
Strengths and limitations To the best of our knowledge, this is the first attempt to extensively examine the published literature on the impacts of high ambient temperatures on mental health outcomes without limit on location, population or methodology. Previous work has given an overview of the impacts of climate change as a whole, and thus, this key risk area has remained unexplored until now. Strengths of this review include adherence to PRISMA guidance and the use of published quality assessment tools to assess bias. This ensures that the method and interpretation of the review is robust. Nevertheless, certain limitations must be acknowledged. Excluding non-English articles may mean some relevant studies from other countries were missed. Articles that investigated the association between health outcomes and temperature change over the course of the day (cf. high temperatures per se) were outside the review scope; rapid changes even at cooler temperature may modulate symptoms of mental health disorders. The use of clinical diagnostic codes in our search strategy may have resulted in certain studies being overlooked. For example, hot weather may be linked to social disorder and the community mental health consequences of such disorder. Our review did not identify any such studies perhaps because such evidence does not exist or more likely due to the use of clinical definitions in the search. The review scope was to investigate the relationship between high temperatures and mental health outcomes. Other meteorological exposures including humidity, solar radiation and precipitation may affect mental health. The observational nature of most of the research found conducted in the real world using routinely collected data sets means that temperature was not the only factor to which populations were exposed. Humidity may have a particular effect; in one included study when humidity rose to the 99th percentile, the effect of heat on distress more than doubled.52 This relationship warrants further research. Furthermore, the heterogeneity of methods used in studies undertaken to date limits this review to a narrative synthesis without meta-analysis.
19
We found that the strongest evidence was for mortality and heat, specifically suicide. We can be confident in these findings; however, there is less confidence from the evidence on the relationship between heat and all-cause mortality in those with mental illness. The relationship between heat and psychological morbidity is less certain; the evidence of worsening symptoms among those with mental illnesses was mixed with a limited number of studies focussing on specific outcomes. We found that the evidence was relatively strong for increasing mental health service use during periods of high temperatures. Although the results of this review suggest a positive relationship between mental health outcomes and high temperatures, there remain a number of areas that require further attention. Specific areas that need addressing are the impact of high temperatures on common mental health disorders, ensuring that future researchers use consistent measures of exposure, outcome and effect at as fine a temporal resolution as possible; attempt to identify threshold temperatures at which adverse mental health outcomes increase and finally explore the link between indoor temperature and mental health outcomes. Of relevance to service providers and commissioners are the findings that high indoor temperatures lead to exacerbation of mental health illnesses, suggesting that they should ensure a cool room is available during periods of high temperature, in line with the recommendations in the Heatwave Plan for England. In addition, the finding that service use was found to increase significantly during periods of high temperature may have business continuity implications. Public health authorities may wish to update suicide prevention strategies to acknowledge the impact that high temperatures and wider environmental factors may have. Furthermore, the mental health impacts of high temperatures should be incorporated into plans for the public health response to high temperatures, and as more evidence is generated, psychological morbidity and mortality temperature thresholds should be incorporated into hot weatherewarning system thresholds for action. At a time of increasing global temperatures, the importance of understanding the impact on psychological morbidity and mortality, filling the gaps in our knowledge and using the available evidence to inform policy and practice has never been more urgent.
Author statements Ethical approval Not required as this was a review of the published literature.
Funding Conclusions This review is the first to summarise the impact of high temperatures on mental health. The results have highlighted gaps in the knowledge base and have identified areas relevant to providers and commissioners and mental health services and for public health action.
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.
Competing interests The authors have no competing interests to declare.
Please cite this article in press as: Thompson R, et al., Associations between high ambient temperatures and heat waves with mental health outcomes: a systematic review, Public Health (2018), https://doi.org/10.1016/j.puhe.2018.06.008
20
p u b l i c h e a l t h x x x ( 2 0 1 8 ) 1 e2 1
references
1. Administration TNOaA, editor. State of the climate: global climate report for annual 2016. The National Oceanic and Atmospheric Administration; 2017. 2. Cook J, Oreskes N, Doran P, Anderegg W, Verheggen B, W Maibach E, et al. Consensus on consensus: a synthesis of consensus estimates on human-caused global warming. Environ Res Lett 2016;11. 3. Climate change 2014: synthesis report. Contribution of Working Group II to the fifth assessment report of the Intergovernmental Panel on Climate Change. Geneva, Switzerland: Intergovernmental Panel on Climate Change; 2014. 4. Sendai framework for disaster risk reduction 2015e2030. 2015. 5. Hanigana IC, Butlera CD, Kokicc PN, Hutchinsond MF. Suicide and drought in new South Wales, Australia, 1970e2007. Proc Natl Acad Sci USA 2012;109. 6. Waite TD, Chaintarli K, Beck CR, Bone A, Richard A, Kovats S, et al. The English national cohort study of flooding and health: cross-sectional analysis of mental health outcomes at year one. BMC Public Health 2017;17. 7. Page LA, Howard LM. The impact of climate change on mental health (but will mental health be discussed at Copenhagen?). Psychol Med 2009;40:177e80. 8. Global, regional, and national incidence, prevalence, and years lived with disability for 301 acute and chronic diseases and injuries in 188 countries, 1990e2013: a systematic analysis for the Global Burden of Disease Study 2013. Lancet 2015;386:743e800. 9. Department for Environment FaRA, editor. A green future: our 25 year plan to improve the environment. HM Government; 2018. 10. Kirtman B, Power SB, Adedoyin JA, Boer GJ, Bojariu R, Camilloni I, et al. Near-term climate change: projections and predictability. In: Climate change 2013: the physical science basis. Contribution of Working Group I to the fifth assessment report of the Intergovernmental Panel on Climate Change. Cambridge, United Kingdom and New York, NY, USA: Intergovernmental Panel on Climate Change; 2013. 11. Sardon J. The 2003 heat wave. Eurosurveillance 2007;12. 12. Armstrong B. Models for the relationship between ambient temperature and daily mortality. Epidemiology 2006:624e31. 13. Hajat S, Kovats RS, Lachowycz K. Heat-related and coldrelated deaths in England and Wales: who is at risk? Occup Environ Med 2007:93e100. 14. Ye X, Wolff R, Yu W, Vaneckova P, Pan X, Tong S. Ambient temperature and morbidity: a review of epidemiological evidence. Environ Health Perspect 2012:19e28. 15. Smith KR, Woodward A, Campbell-Lendrum D, Chadee DD, Honda Y, Liu Q, et al. Human health: impacts, adaptation, and co-benefits. In: Climate change 2014: impacts, adaptation, and vulnerability. Part A: Global and sectoral aspects. Contribution of Working Group II to the fifth assessment report of the Intergovernmental Panel on Climate Change. Cambridge, United Kingdom and New York, NY, USA: Intergovernmental Panel on Climate Change; 2014. 16. Xu Z, FitzGerald G, Guo Y, Jalaludin B, Tong S. Impact of heatwave on mortality under different heatwave definitions: a systematic review and meta-analysis. Environ Int 2016;89e90:193e203. 17. UK climate change risk assessment 2017 synthesis report: priorities for the next five years. London: Adaptation Sub-Committee of the Committee on Climate Change; 2016. 18. Analytics C. Endnote X7. Thomson Reuters; 2013. 19. National Institute of Health study quality assessment tools. National Institute of Health; Available from: https://www. nhlbi.nih.gov/health-topics/study-quality-assessment-tools (last accessed 08 August 2017).
20. Harden A, Thomas J. Methods for the thematic synthesis of qualitative research in systematic reviews. € v J, Giang KB, Kullgren G, Nilsson M. 21. Trang PM, Rocklo Heatwaves and hospital admissions for mental disorders in Northern Vietnam. PLoS One 2016;11, e0155609. 22. Ajdacic-Gross V, Lauber C, Sansossio R, Bopp M, Eich D, Gostynski M, et al. Seasonal associations between weather conditions and suicide - evidence against a classic hypothesis. Am J Epidemiol 2007;165:561e9. 23. Deisenhammer EA, Kemmler G, Parson P. Association of meteorological factors with suicide. Acta Psychiatr Scand 2003;108:455e9. 24. Dixon PG, McDonald AN, Scheitlin KN, Stapleton JE, Allen JS, Carter WM, et al. Effects of temperature variation on suicide in five U.S. counties, 1991-2001. Int J Biometeorol 2007;51:395e403. 25. Dixon PG, Sinyor M, Schaffer A, Levitt A, Haney CR, Ellis KN, et al. Association of weekly suicide rates with temperature anomalies in two different climate types. Int J Environ Res Public Health 2014;11:11627e44. 26. Fernandez-Arteaga V, Tovilla-Zarate CA, Fresan A, GonzalezCastro TB, Juarez-Rojop IE, Lopez-Narvaez L, et al. Association between completed suicide and environmental temperature in a Mexican population, using the Knowledge Discovery in Database approach. Comput Methods Programs Biomed 2016;135:219e24. 27. Gaxiola-Robles R, de la Rosa AdJC, Labrado-Martagon V, DiazCastro SC, Zenteno-Savin T. Environmental temperature increase and its possible association with suicide in the population of Baja California Sur (BCS) 1985-2008. Salud Ment 2013;36:421e7. 28. Grjibovski AM, Kozhakhmetova G, Kosbayeva A, Menne B. Associations between air temperature and daily suicide counts in Astana, Kazakhstan. Medicina (Kaunas) 2013;49:379e85. 29. Hiltunen L, Haukka J, Ruuhela R, Suominen K, Partonen T. Local daily temperatures, thermal seasons, and suicide rates in Finland from 1974 to 2010. Environ Health Prev Med 2014;19:286e94. 30. Kim Y, Kim H, Honda Y, Guo YL, Chen BY, Woo JM, et al. Suicide and ambient temperature in East Asian countries: a time-stratified case-crossover analysis. Environ Health Perspect 2016;124:75e80. 31. Kim Y, Kim H, Kim DS. Association between daily environmental temperature and suicide mortality in Korea (2001-2005). Psychiatry Res 2011;186:390e6. 32. Lee H-C, Lin H-C, Tsai S-Y, Li C-Y, Chen C-C, Huang C-C. Suicide rates and the association with climate: a populationbased study. J Affect Disord 2006;92:221e6. 33. Likhvar V, Honda Y, Ono M. Relation between temperature and suicide mortality in Japan in the presence of other confounding factors using time-series analysis with a semiparametric approach. Environ Health Prev Med 2011;16:36e43. 34. Lin H-C, Chen C-S, Xirasagar S, Lee H-C. Seasonality and climatic associations with violent and nonviolent suicide: a population-based study. Neuropsychobiology 2008;57:32e7. 35. Maes M, De Meyer F, Thompson P, Peeters D, Cosyns P. Synchronized annual rhythms in violent suicide rate, ambient temperature and the light-dark span. Acta Psychiatr Scand 1994;90:391e6. 36. Marion SA, Oluwafemi Agbayewa M, Wiggins S. The effect of season and weather on suicide rates in the elderly in British Columbia. Can J Public Health 1999;90:418e22. 37. Page LA, Hajat S, Kovats RS. Relationship between daily suicide counts and temperature in England and Wales. Br J Psychiatry 2007;191:106e12.
Please cite this article in press as: Thompson R, et al., Associations between high ambient temperatures and heat waves with mental health outcomes: a systematic review, Public Health (2018), https://doi.org/10.1016/j.puhe.2018.06.008
p u b l i c h e a l t h x x x ( 2 0 1 8 ) 1 e2 1
38. Qi X, Hu W, Mengersen K, Tong S. Socio-environmental drivers and suicide in Australia: Bayesian spatial analysis. BMC Public Health 2014;14:681. 39. Carney PA, Fitzgerald CT, Monaghan CE. Influence of climate on the prevalence of mania. Br J Psychiatry 1988;152:820e3. 40. Lee H-C, Tsai S-Y, Lin H-C. Seasonal variations in bipolar disorder admissions and the association with climate: a population-based study. J Affect Disord 2007;97:61e9. 41. McWilliams S, Kinsella A, O'Callaghan E. Daily weather variables and affective disorder admissions to psychiatric hospitals. Int J Biometeorol 2014;58:2045e57. 42. Shapira A, Shiloh R, Potchter O, Hermesh H, Popper M, Weizman A. Admission rates of bipolar depressed patients increase during spring/summer and correlate with maximal environmental temperature. Bipolar Disord 2004;6:90e3. 43. Sung TI, Chen MJ, Su HJ. A positive relationship between ambient temperature and bipolar disorder identified using a national cohort of psychiatric inpatients. Soc Psychiatry Psychiatr Epidemiol 2013;48:295e302. 44. Gupta S, Murray RM. The relationship of environmental temperature to the incidence and outcome of schizophrenia. Br J Psychiatry 1992;160:788e92. 45. Hermesh H, Shiloh R, Epstein Y, Manaim H, Weizman A, Munitz H. Heat intolerance in patients with chronic schizophrenia maintained with antipsychotic drugs. Am J Psychiatry 2000;157:1327e9. 46. Shiloh R, Shapira A, Potchter O, Hermesh H, Popper M, Weizman A. Effects of climate on admission rates of schizophrenia patients to psychiatric hospitals. Eur Psychiatry 2005;20:61e4. 47. Shiloh R, Munitz H, Stryjer R, Weizman A. A significant correlation between ward temperature and the severity of symptoms in schizophrenia inpatientsea longitudinal study. Eur Neuropsychopharmacol 2007;17:478e82. 48. Hansen A, Bi P, Nitschke M, Ryan P, Pisaniello D, Tucker G. The effect of heat waves on mental health in a temperate Australian city. Environ Health Perspect 2008;116:1369e75. 49. Linares C, Culqui D, Carmona R, Ortiz C, Diaz J. Short-term association between environmental factors and hospital admissions due to dementia in Madrid. Environ Res 2017;152:214e20. 50. Tartarini F, Cooper P, Fleming R, Batterham M. Indoor air temperature and agitation of nursing home residents with dementia. Am J Alzheimer's Dis Other Dementias 2017;32:272e81. 51. Page LA, Hajat S, Kovats R, Howard LM. Temperature-related deaths in people with psychosis, dementia and substance misuse. Br J Psychiatry 2012;200:485e90. 52. Ding N, Berry HL, Bennett CM. The importance of humidity in the relationship between heat and population mental health: evidence from Australia. PLoS One 2016;11, e0164190. 53. Vida S, Durocher M, Ouarda TB, Gosselin P. Relationship between ambient temperature and humidity and visits to mental health emergency departments in Quebec. Psychiatr Serv 2012;63:1150e3. 54. Garcia F, Boada S, Collsamata A, Joaquim I, Perez Y, Tricio O, et al. Meteorological factors and psychiatric emergencies. Actas Esp Psiquiatr 2009;37:34e41. 55. McWilliams S, Kinsella A, O'Callaghan E. The effects of daily weather variables on psychosis admissions to psychiatric hospitals. Int J Biometeorol 2013;57:497e508. 56. Nguyen JL, Schwartz J, Dockery DW. The relationship between indoor and outdoor temperature, apparent temperature, relative humidity, and absolute humidity. Indoor Air 2014;24:103e12.
21
57. Schweizer C, Edwards RD, Bayer-Oglesby L, Gauderman WJ, Ilacqua V, Juhani Jantunen M, et al. Indoor timeemicroenvironmenteactivity patterns in seven regions of Europe. J Expo Sci Environ Epidemiol 2006;17:170. 58. Lomas KJ, Porritt SM. Overheating in buildings: lessons from research. Build Res Inf 2017;45:1e18. 59. Anderson M, Carmichael C, Murray V, Dengel A, Swainson M. Defining indoor heat thresholds for health in the UK. Perspect Public Health 2013;133:158e64. 60. Preventing suicide: A global imperative. World Health Organization; 2014. 61. Health Do, editor. Preventing suicide in England: third progress report of the cross-government outcomes strategy to save lives. gov.uk: HM Government; 2017. 62. Linkowski P, Martin F, de Maertelaer V. Effect of some climatic factors on violent and non-violent suicides in Belgium. J Affect Disord 1992;25:161e6. 63. Ruuhela R, Hiltunen L, Venalainen A, Pirinen P, Partonen T. Climate impact on suicide rates in Finland from 1971 to 2003. Int J Biometeorol 2009;53:167e75. € ki M, Bell S, Gale CR, Shipley M, Whitley E, 64. Batty GD, Kivima et al. Psychosocial characteristics as potential predictors of suicide in adults: an overview of the evidence with new results from prospective cohort studies. Transl Psychiatry 2018;8:22. 65. Hajat S, Vardoulakis S, Heaviside C, Eggen B. Climate change effects on human health: projections of temperature-related mortality for the UK during the 2020s, 2050s and 2080s. J Epidemiol Commun Health 2014. 66. Boker SM, Leibenluft E, Deboeck PR, Virk G, Postolache TT. Mood oscillations and coupling between mood and weather in patients with Rapid Cycling Bipolar Disorder. Int J Child Health Hum Dev 2008;1:181e202. 67. Medici CR, Vestergaard CH, Hadzi-Pavlovic D, MunkJorgensen P, Parker G. Seasonal variations in hospital admissions for mania: examining for associations with weather variables over time. J Affect Disord 2016;205:81e6. 68. Banziger G, Owens K. Geophysical variables and behavior: II. Weather factors as predictors of local social indicators of maladaptation in two non-urban areas. Psychol Rep 1978;43:427e34. 69. England PH, editor. Heatwave plan for England. HM Government; 2015. https://www.gov.uk/government/ publications/heatwave-plan-for-england (last accessed 08 August 2017). 70. Armstrong BG, Chalabi Z, Fenn B, Hajat S, Kovats S, Milojevic A, et al. Association of mortality with high temperatures in a temperate climate: England and Wales. J Epidemiol Commun Health 2011;65:340e5. 71. Heatwaves and health: guidance on warning-system development. World Health Organization and World Meteorological Organization; 2015. p. 114. 72. England PH, editor. Cold weather plan for England. HM Government; 2015. https://www.gov.uk/government/ publications/cold-weather-plan-cwp-for-england (last accessed 08 August 2017).
Appendix A. Supplementary data Supplementary data related to this article can be found at https://doi.org/10.1016/j.puhe.2018.06.008.
Please cite this article in press as: Thompson R, et al., Associations between high ambient temperatures and heat waves with mental health outcomes: a systematic review, Public Health (2018), https://doi.org/10.1016/j.puhe.2018.06.008