Research
A Section 508–conformant HTML version of this article
is available at http://dx.doi.org/10.1289/ehp.1409119.
Multiple Trigger Points for Quantifying Heat-Health Impacts: New Evidence
from a Hot Climate
Diana B. Petitti,1,2 David M. Hondula,3,4 Shuo Yang,5 Sharon L. Harlan,5 and Gerardo Chowell 5,6
1Department of Biomedical Informatics, and 2Department of Family, Community and Preventive Medicine, College of Medicine-Phoenix,
University of Arizona, Phoenix, Arizona, USA; 3Center for Policy Informatics, Arizona State University, Phoenix, Arizona, USA; 4School
of Geographical Sciences and Urban Planning, Arizona State University, Tempe, Arizona, USA; 5School of Human Evolution & Social
Change, Arizona State University, Tempe, Arizona, USA; 6School of Public Health, Georgia State University, Atlanta, Georgia, USA
Background: Extreme heat is a public health challenge. The scarcity of directly comparable studies
on the association of heat with morbidity and mortality and the inconsistent identification of
threshold temperatures for severe impacts hampers the development of comprehensive strategies
aimed at reducing adverse heat-health events.
Objectives: This quantitative study was designed to link temperature with mortality and morbidity
events in Maricopa County, Arizona, USA, with a focus on the summer season.
Methods: Using Poisson regression models that controlled for temporal confounders, we assessed
daily temperature–health associations for a suite of mortality and morbidity events, diagnoses, and
temperature metrics. Minimum risk temperatures, increasing risk temperatures, and excess risk
temperatures were statistically identified to represent different “trigger points” at which heat-health
intervention measures might be activated.
Results: We found significant and consistent associations of high environmental temperature
with all-cause mortality, cardiovascular mortality, heat-related mortality, and mortality resulting
from conditions that are consequences of heat and dehydration. Hospitalizations and emergency
department visits due to heat-related conditions and conditions associated with consequences of
heat and dehydration were also strongly associated with high temperatures, and there were several
times more of those events than there were deaths. For each temperature metric, we observed large
contrasts in trigger points (up to 22°C) across multiple health events and diagnoses.
Conclusion: Consideration of multiple health events and diagnoses together with a comprehensive approach to identifying threshold temperatures revealed large differences in trigger points
for possible interventions related to heat. Providing an array of heat trigger points applicable for
different end-users may improve the public health response to a problem that is projected to worsen
in the coming decades.
Citation: Petitti DB, Hondula DM, Yang S, Harlan SL, Chowell G. 2016. Multiple trigger points
for quantifying heat-health impacts: new evidence from a hot climate. Environ Health Perspect
124:176–183; http://dx.doi.org/10.1289/ehp.1409119
Introduction
Many studies have retrospectively examined
high environmental temperature and
mortality. This research has largely focused
on estimating excess deaths from all-cause
mortality and on the statistical identification
of a single threshold temperature above which
deaths increase (e.g., Hajat and Kosatsky
2010; McMichael et al. 2008). Importantly,
the temperature thresholds identified in such
studies have been proposed as a basis for the
activation of heat-health warning systems
and other public health interventions (e.g.,
Henderson and Kosatsky 2012; Pascal et al.
2006). Other applications of retrospective
analyses include assessment of the potential
future health effects of local-, regional-, or
global-scale climate change (e.g., Huang
et al. 2011).
A related and rapidly accumulating body
of research assesses the relationship between
high temperature and health events other than
mortality: hospital admissions and emergency
department (ED) visits (Hess et al. 2014;
reviews by Kravchenko et al. 2013; Martiello
and Giacchi 2010; Ye et al. 2012), hospital
176
admissions among patients seen in the ED
(Pillai et al. 2014), ambulance/emergency
response calls (Alessandrini et al. 2011; Hartz
et al. 2013; Nitschke et al. 2011; Schaffer
et al. 2012; Williams et al. 2012a, 2012b),
teleradiology calls (Brunetti et al. 2014), and
outpatient visits (Pudpong and Hajat 2011).
However, only a few studies have considered
more than one measure of health effects
associated with heat, for a single geographic
region, at the same time (e.g., Kovats et al.
2004; Williams et al. 2012a, 2012b).
The fact that extreme heat persists as a
public health challenge (Berko et al. 2014)
despite compelling evidence for its adverse
effects on health calls for new approaches
toward preparedness and intervention strategies. Here, we propose that it is possible to
better understand and mitigate the current
and future risks posed by high temperatures
with adaptation strategies based on comprehensive and contextualized weather information spanning a range of health outcomes
associated directly and indirectly with heat.
Opportunities for improving public
health strategies aimed at mitigating the
volume
effects of heat on health may lie at the
intersection of many of the ideas and methodologies that have been brought forward
to date. For example, functional forms of
heat-health relationships are dependent on
the local setting (Anderson and Bell 2009;
Curriero et al. 2002). In addition, the relationship between temperature and mortality
and morbidity may have different functional
forms within a given location (Kovats et al.
2004). Intervention strategies aimed at
particular populations (e.g., outdoor workers
vs. elderly residents) would be most effective if they considered the diagnosis and
severity of health events that are most relevant
for that population. Furthermore, various
definitions of temperature thresholds are
employed in the literature, some of which are
brought forward with little more than generalities about the purpose of identifying such
metrics. The suite of different conceptualizations of “thresholds” for heat-related health
effects proposed thus far (e.g., Davis et al.
2003; Li et al. 2013; Pascal et al. 2006) offers
considerable variability in terms of utility for
heat-health adaptation strategies.
The aim of this study was to systematically identify the meteorological conditions
under which there might be reasons to enact
heat-health interventions based on empirical
relationships between hot weather and illness
or death. Our concern was that an opportunity to mitigate a large portion of adverse
health outcomes associated with heat may be
lost if the activation of preventive measures
for heat-related illness and death is keyed to
temperatures at which all-cause mortality
statistically exceeds a seasonal baseline. In hot
climates such as the one that characterizes
Address correspondence to D.B. Petitti, Department
of Biomedical Informatics, 1711 W. Lodge Dr.,
Phoenix, AZ 85041 USA. Telephone: (602) 7953804. E-mail: diana.petitti@yahoo.com
Supplemental Material is available online (http://
dx.doi.org/10.1289/ehp.1409119).
This research was supported by grants from the
National Science Foundation (GEO-0816168 and
BCS-1026865) and the Virginia G. Piper Charitable
Trust Health Policy Informatics Initiative at Arizona
State University.
The authors declare they have no actual or potential
competing financial interests.
Received: 24 August 2014; Accepted: 22 July
2015; Advance Publication: 28 July 2015; Final
Publication: 1 February 2016.
124 | number 2 | February 2016 • Environmental Health Perspectives
New evidence for quantifying heat-health impacts
Maricopa County, Arizona, health events
associated with heat exposure may begin
to occur well before a statistical threshold
temperature for all-cause mortality is crossed
(Harlan et al. 2014). Moreover, there are a
suite of health events and diagnoses associated with heat that may respond differently
to ambient conditions. Hence, our approach
moves beyond the use of a single threshold by
considering multiple different temperatures
(henceforth referred to as trigger points) to
characterize the complex relationship between
heat and health.
Materials and Methods
Study setting. The study setting, Maricopa
County, Arizona, USA, (2012 population, 3.9
million) comprises the city of Phoenix (2012
population 1.5 million), eight other contiguous cities with populations ranging from
100,000 to 400,000, 15 adjoining municipalities, and three Native American communities. In Phoenix, the daily mean temperature
in the summer (June–September), 33 o C
(91.4oF), is the highest of all major United
States metropolitan areas [National Oceanic
and Atmospheric Administration (NOAA)
2013]. In the Phoenix metropolitan area, 95%
of occupied housing units have central air
conditioning, which is > 50% greater than the
national average [American Housing Survey
(AHS) 2014].
Health data. The study considered 10
different health events: all-cause mortality;
cardiovascular (CVD) mortality, hospitalizations, and emergency department (ED)
visits; heat-related deaths, hospitalizations,
and ED visits; and mortality, hospitalizations, and ED visits for conditions that are
consequences of heat and dehydration. The
selected events represent different levels of
severity for personal suffering and loss (death,
hospitalization, emergency treatment) and
health problems that represent different types
of risk profiles: all-cause mortality (broadest
scope, most often studied), CVD (underlying
disease, greater physiological susceptibility,
large affected population), and direct heat
exposure (acute, specific, situational).
We obtained mortality data for 1 January
2000–31 December 2011 from the Arizona
Department of Health Services (ADHS).
Each record included date of death, underlying cause of death coded using the World
Health Organization’s (WHO’s) International
Classification of Diseases, 10th Revision
(ICD‑10), and text entered in the contributing
causes of death fields on the death certificate.
We also obtained data on hospitalizations
and ED visits at facilities located in Maricopa
County for 1 January 2008–31 December
2012 from ADHS. All Arizona hospitals
except Veteran’s Administration, military,
Indian Health Services, and behavioral health
Environmental Health Perspectives •
volume
hospitals were required by law to report
information to ADHS during this period.
Information obtained included admission
and discharge dates in addition to discharge
diagnoses and causes of injury coded using
the WHO’s International Classification of
Diseases, 9th Revision, Clinical Modification
(ICD‑9‑CM). During the study period,
codes were captured on ≤ 25 discharge diagnoses and ≤ 9 external causes of injury for
each individual for both hospitalizations
and ED visits.
In our analysis based on all-cause
mortality, we excluded most external causes
of death. Following the method reported by
Harlan et al. (2014), we excluded ICD-10
codes S00–99, T00–66, T68–98, U00–99,
X00–29, 32, 33–53, 55–84, Y00–98, and
Z00–99 but included T67.x, X30, X32,
and X54 because these are heat-related. The
conditions used to define mortality and
morbidity events in the CVD category and
their corresponding ICD-10 and ICD-9
codes are listed in Supplemental Material,
Table S1. We conducted two separate
analyses of CVD hospitalization and ED
visits, one using only the first discharge
diagnosis code to define a patient as having a
CVD event and one using all (≤ 25) discharge
diagnosis codes to define a patient as having
a CVD event. Only data for CVD as the first
discharge diagnosis are discussed because the
results were essentially the same when CVD
as any discharge diagnosis was examined (data
not shown).
The conditions used to define a mortality
or morbidity event as heat-related and the
corresponding ICD-10 and ICD-9-CM codes
are listed in Supplemental Material, Table S2.
In the heat-related mortality category, terms
associated with exposure to high environmental heat (e.g., “heat exhaustion”) entered
as free text in the underlying cause-of-death
fields of the death certificate Part 1 were
also used to define a death as heat-related
(see Supplemental Material, Table S2).
Hospitalizations and ED visits were classified
as directly heat-related if any discharge diagnosis code (≤ 25 possible for any individual
hospitalization or ED visit) or external cause
of injury code (≤ 9 possible) corresponded to
the predefined ICD codes for this category.
A category of conditions that are possible
consequences of heat and/or dehydration
was defined based on a model of the physiologic and pathophysiologic effects of heat.
The Supplemental Material presents a graphic
depiction of the model (see Supplemental
Material, Figure S1) along with a list of the
ICD-10 and ICD-9 codes for this category
(see Supplemental Material, Table S3).
Hospitalizations and ED visits were classified
as possible consequences of heat and/or dehydration if any of the ≤ 25 discharge diagnosis
124 | number 2 | February 2016
codes or ≤ 9 external cause of injury codes
corresponded to the predefined ICD-9 codes
for this category.
Individuals who were hospitalized more
than once or who had more than one ED
visit were counted multiple times. However,
individuals admitted to the hospital who were
also seen in the ED for that same episode of
illness were counted only once, as a hospitalization. Fatal outcomes during or after being
hospitalized or in the ED or after being seen
in the ED were counted in both the mortality
analysis and the analyses of hospitalization
and ED visits because the available data did
not permit deduplication across data sources.
Ethics review. The study was reviewed
and approved by both the Arizona State
University Institutional Review Board and
the ADHS Human Subjects Review Board.
Meteorological data. We obtained
hourly air temperature and relative humidity
data from the National Weather Service
(NWS) monitoring station at Sky Harbor
International Airport in Phoenix for the
period 1 January 2000–31 December 2012.
From these data, we calculated six temperature metrics: daily minimum, mean, and
maximum air temperature (T min , T mean ,
and Tmax, respectively) and daily minimum,
mean, and maximum heat index (HI min ,
HI mean , HI max , respectively). We used
the lowest and highest daily values for the
minimum and maximum, respectively,
and the average of 24-hr temperatures as
the daily mean. The HI estimates thermal
stress resulting from ambient conditions by
combining temperature and humidity into
a single variable. Here, we used an NWS HI
algorithm that parameterizes the Steadman
apparent temperature model (NWS 2014;
Steadman 1979). Detail is provided in
the Supplemental Material, “Algorithm
for Calculation of Heat Index Based on
Steadman 1979; NWS 2014.”
Analysis. To minimize the effect of season
on health, we restricted the analysis to the
period 15 May–15 October of each year. In
this setting, we found same-day and 1-day lag
temperature and HI to be among the most
important discriminators between days with
high and low mortality, hospitalizations, or
ED visits. Thus, these variables were deemed
to have stronger associations with health
events than were other possible variables
(e.g., dew point temperature, departures from
climatological normals, variables with longer
lags or smoothers including conceptualizations of “heat waves”). A full examination of
this larger suite of potential explanatory variables is outside the scope of this analysis, but
the six variables we chose to examine are in
line with those found to be most relevant to
health (e.g., Anderson and Bell 2009; Hajat
et al. 2006).
177
Petitti et al.
178
health event was statistically significantly
greater than 1.0 based on the lower bound of
the 95% confidence interval for the relative
risk above 1.0. The reference level for estimation of relative risk is the expected rate of the
health event in a given month. Conceptually,
the ERT is the lowest temperature at which
mortality or morbidity rates are modeled to
be anomalously greater than the number of
events expected based on normal summer
weather and, for some of the health events
considered, other temporal factors that drive
seasonal v ariability in the time series of
event counts.
MRTs, IRTs, and ERTs could
be undefined.
The sensitivity of the results to the time
period of record was assessed by replicating
the abovementioned procedure for several
different combinations of study period start
and end years.
Table 1. For categories and types of events, total and average events per year for months in analysis.
Category/event type
All-cause mortality
Cardiovascular
Mortality
Hospitalizationa
ED visita
Heat-related
Mortality
Hospitalization
ED visit
Consequences of heat and dehydration
Mortality
Hospitalization
ED visit
Years in analysis
Total number of events
for months in analysis
112,853
Average events per year
for months in analysis
9,404
2000–2011
2008–2012
2008–2012
30,531
32,614
6,831
2,544
6,523
1,366
2000–2011
2008–2012
2008–2012
424
1,731
6,803
35
346
1,361
2000–2011
2008–2012
2008–2012
1,458
357,363
233,636
122
71,473
46,727
ICD-9 and ICD-10 codes used to define categories of conditions are given in the Supplemental Material, Tables S1–S3.
aFirst discharge diagnosis only.
All-cause Death
1.10
1.10
1.05
1.00
0.95
0.90
1.10
Relative Risk
Relative Risk
1.05
1.00
0.95
0.90
20
25
30
35
40
45
20
Tmax (°C)
25
30
35
1.05
1.00
0.95
0.90
10
35
40
15
45
50
HImax (°C)
20
25
30
35
30
35
Tmin (°C)
1.10
1.05
1.00
0.95
0.90
30
0.95
40
1.10
25
1.00
Tmean (°C)
1.10
20
1.05
0.90
15
50
Relative Risk
where M is a time series of mortality or CVD
morbidity, month is a factor term representing
month of year, year is a factor term representing calendar year, s is a fixed thin-plate
regression spline with k-1 degrees of freedom,
and env represents any of the six temperature
metrics considered.
Because the study was restricted to
warmer months (15 May–15 October), we
did not combine seasonal and long-term
trend effects into one single temporal variable
(e.g., Anderson and Bell 2009; Hondula
et al. 2013). Restricting the analysis to the
mid-May to mid-October window greatly
reduced concerns regarding confounding
effects from annual variability in all-cause
and CVD event rates, which are accounted
for by the month term in Equation 1. We
found that replacing month with a higherresolution time variable such as day of year
had no appreciable influence on the overall
results (data not shown). The models for heatrelated events did not include the term month
because any seasonality in these events was
believed to be directly related to temperature.
Based on the modeled relationships
between each of the six temperature metrics
and the 10 health events, we calculated three
separate trigger points to compare the relative
sensitivity to hot weather across metrics and
events. We defined trigger points as temperatures at which there is a prespecified increase
in the occurrence of the given health event.
The minimum risk temperature (MRT) is
conceptually similar to the temperature of
minimum mortality described by Curriero
et al. (2002), Keatinge et al. (2000), and
Kinney et al. (2008). For health events that
would not be expected in the absence of high
temperatures (heat-related mortality, hospitalizations, and ED visits and events associated
with mortality, hospitalization, and ED visits
that were categorized as consequences of heat
and dehydration), we defined the MRT as
the temperature at which the fewest events
were observed (which was typically the lowest
Relative Risk
[1]
Relative Risk
log(M) = month + year + s(env, k = 4),
temperature at which an event was observed).
For health events that may be influenced
by, but are not entirely dependent on, high
temperature (all-cause mortality and CVD
events), we defined the MRT as the lowest
temperature above which a consistent increase
in relative risk was observed (i.e., the slope of
the temperature–health event relationship is
always positive above the MRT).
The increasing risk temperature (IRT)
was defined as the lowest temperature
at which the relative risk of a given health
event was greater than the upper 95% confidence limit of the MRT. Thus, the IRT is
an indicator of the lowest temperature at
which there is a larger impact on the health
event than what is expected under optimal
weather conditions.
The excess risk temperature (ERT) was
defined as the lowest temperature above the
MRT at which the relative risk of a particular
Relative Risk
We estimated the relationship between
the temperature metrics and the health events
using a generalized additive model (GAM)
(Hastie and Tibshirani 1990). Separate
models were constructed for each of the six
temperature metrics and for each of the 10
different types of health events considered.
For the CVD category, we used a 1-day lag
between the air temperature or HI metric and
the events [following the method of Harlan
et al. (2014)]. For the other event types, we
examined same-day effects.
For all-cause mortality and CVD events
(mortality, hospitalization, and ED visits), the
GAM took the form:
1.05
1.00
0.95
0.90
15
20
25
30
35
HImean (°C)
40
10
15
20
25
HImin (°C)
Figure 1. The modeled relationship between the relative risk of all-cause mortality and six different
same-day temperature metrics during the warm season for Maricopa County, Arizona, 2000–2011. The
solid blue line shows the relative risk of mortality, and the shaded blue region shows the 95% confidence
interval. Specific points labeled on the curve identify the minimum risk temperature (MRT, black), the
increasing risk temperature (IRT, blue), and the excess risk temperature (ERT, red), representing different
conceptualizations of trigger points for intervention activities as discussed in “Methods.”
volume
124 | number 2 | February 2016 • Environmental Health Perspectives
New evidence for quantifying heat-health impacts
Environmental Health Perspectives •
volume
MRTs and IRTs were much lower for this
category of conditions than for all-cause
mortality, CVD mortality, and heat-related
conditions. For example, considering Tmax,
the MRT and IRT were 25°C and 31°C,
respectively, for mortality due to conditions
considered consequences of heat and dehydration, whereas the MRT and IRT were 35°C
and 39°C, respectively, for all-cause mortality.
We found strong and statistically significant associations between same-day temperature and the three directly heat-related health
events (Figures 4 and 5). The relationship
exhibited an exponential pattern across all
temperature metrics and types of events.
MRTs, IRTs, and ERTs were identified for
all six temperature metrics for all types of
heat-related events. Notably, for all of the
temperature metrics, both the MRT and
the IRT were consistently 2–7°C lower for
heat-related hospitalization and heat-related
ED visits than for heat-related mortality. For
example, considering Tmax, the corresponding
MRT was 26°C for mortality, but 22°C
for hospitalization and 22°C for ED visits;
similarly, the IRT was 33°C for mortality,
but 27°C for hospitalization and 29°C for
ED visits. For all of the temperature metrics,
however, the ERT was almost the same
(± 1–2°C) for each type of heat-related event.
For example, considering HImax, the ERT
was 39°C for heat-related death and 38°C for
both hospitalization and ED visits.
The conceptualization of trigger point
and choice of health event and diagnosis
led to large contrasts in the temperatures at
Cardiovascular Death
1.10
1.05
1.00
0.95
0.90
1.10
Relative Risk
Relative Risk
Relative Risk
1.10
1.05
1.00
0.95
0.90
20
25
30
35
40
45
20
30
35
1.05
1.00
0.95
0.90
10
35
40
15
45
25
30
35
30
35
1.10
1.05
1.00
0.95
50
20
Tmin (°C)
0.90
30
0.95
40
Relative Risk
Relative Risk
Relative Risk
25
1.10
25
1.00
Tmean (°C)
1.10
20
1.05
0.90
15
50
Tmax (°C)
1.05
1.00
0.95
0.90
15
20
HImax (°C)
25
30
35
10
40
15
HImean (°C)
20
25
HImin (°C)
Figure 2. The modeled relationship between the relative risk of cardiovascular mortality and six different
temperature metrics with a 1-day lag, as in Figure 1. Fewer than three points are indicated on the curve if
some of the trigger points could not be identified.
Death from Consequences of Heat and Dehydration
1.4
1.4
1.2
1.0
0.8
0.6
1.4
Relative Risk
Relative Risk
Relative Risk
1.2
1.0
0.8
0.6
20
25
30
35
40
45
20
Tmax (°C)
25
30
35
10
0.8
0.6
35
40
45
50
20
25
30
35
30
35
1.4
1.2
1.0
0.8
0.6
HImax (°C)
15
Tmin (°C)
Relative Risk
1.0
30
0.8
40
1.4
1.2
25
1.0
Tmean (°C)
1.4
20
1.2
0.6
15
50
Relative Risk
During the time period for which both
mortality and morbidity data were available, the number of morbidity events greatly
exceeded the number of mortality events
(Table 1). The average number of heat-related
deaths per year for the months in the analysis
from 2008 to 2011 (n = 35) was 10.1% of the
average number of heat-related hospitalizations (n = 346), which in turn was 25.4%
of the average number of heat-related ED
visits (n = 1,361). For reference, in Maricopa
County during the period 2008–2011,
approximately 460,000 hospitalizations and
1.1 million ED visits (not admitted to the
hospital) per year were recorded [Agency
for Healthcare Research and Quality
(AHRQ) 2014].
Across all temperature metrics, the relative
risk of all-cause mortality at the highest
recorded temperatures exceeded 1.05, with
95% confidence intervals that excluded 1.0
(Figure 1). All three trigger points (MRT,
IRT, and ERT) were identified for all six
temperature metrics. Regardless of the
temperature metric examined, the ERT
estimate for all-cause mortality was 2–3°C
higher than the IRT, and the IRT estimate
was 3–5°C higher than the MRT.
CVD mortality increased with temperature with a 1-day lag (Figure 2). Relative risks
exceeded 1.05 with 95% confidence intervals that excluded 1.0 for some temperature
metrics at the highest temperatures. CVD
trigger points were less consistent than those
for all-cause mortality: an ERT estimate could
not be identified for Tmax, HImax, and HImin,
and there was a large difference in IRT and
MRT using Tmax (22 and 36°C, respectively).
Where trigger points could be identified,
the ERT was 2–3°C higher than the IRT,
and, with the exception of Tmax, the IRT was
3–6°C higher than the MRT. The number of
CVD deaths (n = 30,531) was substantially
smaller than the number of deaths from all
causes (n = 112,853), and the lack of consistency may be a consequence of random error
due to the smaller sample size.
No clear pattern of increased risk with
higher temperature (1-day lag) emerged
for CVD hospitalization or ED visits with
CVD listed as the first discharge diagnosis
(see Supplemental Material, Figures S2 and
S3). Consequently, trigger points could not
be identified for these health events for any
temperature metric.
For the category of conditions called
“consequences of heat and dehydration,”
the relationship with temperature was
consistently positive for mortality, hospitalization, and ED visits (Figure 3; see also
Supplemental Material, Figures S4 and S5),
but the confidence intervals were wide. The
slope of the relationship was shallow. The
Relative Risk
Results
1.2
1.0
0.8
0.6
15
20
25
30
35
HImean (°C)
40
10
15
20
25
HImin (°C)
Figure 3. The modeled relationship between the relative risk of mortality from consequences of heat and
dehydration and six different temperature metrics with a 1-day lag, as in Figure 1. Fewer than three points
are indicated on the curve if some of the trigger points could not be identified.
124 | number 2 | February 2016
179
Petitti et al.
180
25
20
20
15
10
5
Relative Risk
25
20
Relative Risk
Mortality
25
0
15
10
5
0
20
25
30
35
40
45
50
5
20
25
30
35
40
10
4
3
3
2
1
Relative Risk
4
3
2
1
0
20
25
30
35
40
45
50
20
25
30
35
40
10
3
3
Relative Risk
3
Relative Risk
4
2
1
0
25
30
35
40
45
50
35
15
20
25
30
35
30
35
Tmin (°C)
4
0
30
1
4
1
25
2
Tmean (°C)
2
20
0
15
Tmax (°C)
20
15
Tmin (°C)
4
Relative Risk
Relative Risk
Hospitalization
10
Tmean (°C)
0
Relative Risk
15
0
15
Tmax (°C)
2
1
0
15
20
Tmax (°C)
25
30
35
40
10
15
Tmean (°C)
20
25
Tmin (°C)
Figure 4. The modeled relationship between the relative risk of heat-related mortality (top panels), heatrelated hospitalization (middle panels), and heat-related emergency department visits (lower panels),
and three same-day temperature metrics (Tmax, Tmean, Tmin) during the warm season for Maricopa County,
Arizona, 2000–2011 (2008–2012 for morbidity), as in Figure 1. For heat-related events, MRT is the temperature at which the fewest events were observed. Note that the vertical axis scale varies between panels.
25
25
20
20
20
15
10
5
0
Relative Risk
25
Relative Risk
Mortality
Relative Risk
Health Events from Direct Exposure to Environmental Heat
Discussion
15
10
5
0
20
25
30
35
40
45
50
5
20
25
30
35
40
10
4
3
3
2
1
Relative Risk
4
3
2
1
0
20
25
30
35
40
45
50
20
25
30
35
40
10
3
3
Relative Risk
3
Relative Risk
4
2
1
0
25
30
35
40
45
HImax (°C)
50
35
15
20
25
30
35
30
35
HImin (°C)
4
0
30
1
4
1
25
2
HImean (°C)
2
20
0
15
HImax (°C)
20
15
HImin (°C)
4
0
Relative Risk
10
HImean (°C)
Relative Risk
Relative Risk
Hospitalization
15
0
15
HImax (°C)
Emergency Department Visit
Most prior analyses of temperature/event
associations that aim to identify a threshold
temperature for heat-related events, including
our own work set in Maricopa County (Harlan
et al. 2014), define the threshold for action
as the temperature at which the frequency
of health events begins to rise rapidly (most
similar to the ERT in this analysis for all-cause
mortality, CVD mortality, and heat-related
events) although other definitions have been
used (e.g., Armstrong et al. 2011; Hajat
and Kosatsky 2010; Loughnan et al. 2010;
Zaninović and Matzarakis 2014). A statistically solid and reliable health outcomes–based
estimate of temperature trigger points has
the potential to guide the implementation of
interventions when they are most appropriate.
Issuing extreme heat warning products to the
general public by weather forecasting offices
is one such intervention (e.g., Pascal et al.
2006; Williams et al. 2012a), but triggering
criteria for warning systems are often based
on threshold conditions for a singular conceptualization of increases in all-cause mortality
(e.g., Hondula et al. 2014). An understanding
of the broader effects of heat on illness has the
potential to suggest enhancements to public
messaging efforts as well as interventions other
Relative Risk
Health Events from Direct Exposure to Environmental Heat
Emergency Department Visit
which estimated heat risk increased. Table 2
lists the MRT, IRT, and ERT for 8 of the
10 health events considered in order to
facilitate comparisons across categories, event
types, temperature metrics, and risk levels;
comparisons for Tmax for select events are
also illustrated in Figure 6. Cardiovascular
morbidity events are excluded from these
tables and figures because of the lack of a
consistent association with any temperature
metric. Spanning the entire range of risk
temperatures, health events, and categories
of mortality and morbidity, we observed that
trigger points varied by as much as 22°C,
holding the temperature metric constant.
For example, the ERT for all-cause mortality
(considering Tmax) was 42°C, but the MRT
for heat-related mortality was 26°C. When
examining contrasts across metrics within
each type of health event, the MRT, IRT,
and ERT were often within 2°C for the air
temperature and HI forms of the metric.
When the trigger points differed, in most
cases, the HI trigger point was 1–2°C lower
than the air temperature trigger point.
Sensitivity analyses revealed that the
overall scale and pattern of the differences
between trigger points based on different
conceptualizations of thresholds was consistent regardless of the specific time period
examined, although the specific values of the
MRT, IRT, and ERT were not identical for
all examined time periods (see Supplemental
Material, Tables S4 and S5).
2
1
0
15
20
25
30
HImean (°C)
35
40
10
15
20
25
HImin (°C)
Figure 5. The modeled relationship between the relative risk of heat-related mortality (top panels), heatrelated hospitalization (middle panels), and heat-related emergency department visits (lower panels), and
three same-day heat index metrics (HImax, HImean, HImin), as in Figure 1. MRT is the temperature at which
the fewest events were observed. Note that the vertical axis scale varies between panels.
volume
124 | number 2 | February 2016 • Environmental Health Perspectives
New evidence for quantifying heat-health impacts
Environmental Health Perspectives •
volume
Table 2. Excess, increasing, and minimum risk temperatures in degrees Celsius by category and event
type for each temperature metric.
Category/event type
Excess risk temperature
Mortality
All-cause
Cardiovascular
Heat-related
Consequences of heat and dehydration
Hospitalization
Heat-related
Consequences of heat and dehydration
ED visits
Heat-related
Consequences of heat and dehydration
Increasing risk temperature
Mortality
All-cause
Cardiovascular
Heat-related
Consequences of heat and dehydration
Hospitalization
Heat-related
Consequences of heat and dehydration
ED visits
Heat-related
Consequences of heat and dehydration
Minimum risk temperature
Mortality
All-cause
Cardiovascular
Heat-related
Consequences of heat and dehydration
Hospitalization
Heat-related
Consequences of heat and dehydration
ED visits
Heat-related
Consequences of heat and dehydration
Tmax
HImax
Tmean
HImean
Tmin
HImin
42
—
41
42
40
—
39
40
36
37
35
36
35
37
33
35
31
32
28
31
31
—
28
—
40
42
38
—
34
36
32
35
27
32
27
31
39
40
38
38
34
34
32
33
27
28
27
28
39
36
33
31
38
40
32
25
34
34
26
21
33
34
27
20
29
29
19
16
29
30
18
19
27
27
25
26
22
25
20
23
15
21
15
16
29
31
27
23
22
20
20
18
14
14
13
11
35
22
26
25
33
36
25
21
31
31
21
16
30
31
19
16
26
24
15
12
25
24
13
15
22
22
21
20
16
17
15
17
11
14
11
9
22
29
20
20
17
17
15
16
11
11
9
9
Average daily
maximum
temperatures in
Maricopa County, AZ
45
Daily Maximum Temperature (Tmax) (°C)
than warnings that might mitigate the adverse
effects of heat.
Here, we have interrogated temperature
threshold estimates based on three different
criteria (MRT, IRT, and ERT). We found
large differences across these measures and
across different health events and diagnoses.
The strongest and most consistent associations for high environmental temperature in
our setting were with directly heat-related
health events. Trigger points for these events
were consistently lower than those derived
from all-cause mortality. In a hot location
like Maricopa County, using a single high
threshold temperature (e.g., ERT for all-cause
mortality) vastly discounts the number of days
on which heat is associated with an increased
risk of heat-related mortality and morbidity.
This progression of increasing thresholds for
more severe outcomes and the overall finding
that heat-related mortality is merely the top
of the heat severity pyramid was also reported
in Adelaide, South Australia (Williams et al.
2012a). The highest trigger points (ERTs)
that we calculated for several health events
were near climatological averages for summer
daily temperatures (Figure 6). This finding
demonstrates a need to reconsider the heatrisk communication paradigm in hot climates.
We suggest that one improvement would
be for researchers to offer intended endusers an array of trigger points that could be
applied for their specific purposes instead of a
single, all-purpose threshold temperature. In
Maricopa County, we are using the results of
this study to begin conversations with a range
of end-users about actions they could take
when dangerous heat occurs. The ultimate
utility of the trigger points will be determined after engaging in dialogue with service
providers. Potential applications for these
trigger points include identifying days and
times to increase enforcement of workplace
safety guidelines, running seasonal public
awareness campaigns, suspending utility
shutoffs, rescheduling or cancelling outdoor
school events including athletic practices and
competitions, and opening or expanding
access to homeless shelters and cooling centers.
The trigger point framework may also offer
additional opportunities to consider multiple
health outcomes, risk levels, and exposure
variables in studies that project future heat
impacts associated with climate change.
The HI, which is widely used by the NWS
and heat-health researchers in the United
States (e.g., Anderson et al. 2013), provided
information about sensitivity to heat that was
not substantively different from information
derived from air temperature in Maricopa
County. In our study setting, and perhaps in
others characterized by low relative humidity,
actions to mitigate the effects of heat on health
events may not need to use metrics that are
July and August 15 (41°C)
June 15 (40°C)
40
September 15 (38°C)
35
May 15 (35°C)
October 15 (32°C)
30
Excess Risk
Temperature
Increasing Risk
Temperature
25
Minimum Risk
Temperature
20
All-cause
mortality
Heat-related
mortality
Heat-related
hospitalizations
Heat-related
ED visits
Figure 6. Minimum, increasing, and excess risk temperatures (MRT, IRT, ERT) based on daily maximum
temperature (T max) for four health events examined in this study. Values on the right-hand side of
the figure denote climatological averages at regularly spaced intervals during the warm season in
Maricopa County.
124 | number 2 | February 2016
181
Petitti et al.
more complex than air temperature and are,
therefore, more difficult to communicate to
the public. Identification of the optimal
variable(s) to use when triggering protective actions related to extreme heat depends
on rigorous statistical analysis of predictive
capacity (e.g., Barnett et al. 2010; Zhang et al.
2012), local context, and public understanding
of and receptivity to such information.
Exploration of these important dimensions of
heat intervention design falls outside the scope
of this analysis but is the subject of ongoing
efforts by the authors and local public agencies.
Notably, our study did not find an
association between high temperatures and
CVD hospitalization and/or ED visits (see
Supplemental Material, Figures S2 and S3).
In a recent systematic review of studies of
heat and cardiovascular morbidity, Turner
et al. (2012) concluded that the effects of
temperature on cardiorespiratory morbidity
were smaller and more variable than those
on mortality. Administrative data have a
limited ability to shed light on the effects of
temperature on CVD morbidity. As others
have noted (Basu et al. 2012), more studies
that assess specific symptoms in relation to
individual heat exposure are needed.
Our study has several important limitations. We used administrative data to assess
hospitalization and ED visits, as has been done
in previous studies (e.g., Williams et al. 2012b),
although the data sets were created to support
insurance billing and not for use in this type of
research. Our methodology of using ICD-10
codes to identify heat-related mortality from
ADHS records underestimates the number of
heat deaths. In particular, Maricopa County’s
procedures to identify heat-related deaths
have been improving over time, and their heat
mortality surveillance program detected 312
heat-related deaths during the period 2008–
2011 [Maricopa County Department of Public
Health (MCDPH) 2014] compared with
the 153 heat-related deaths that we identified
using procedures more consistent with those
employed by ADHS.
It is also worth noting that our study
focused on a single setting; thus, our findings
may not be generalizable to other settings.
There are many human adaptations to high
temperatures, and Maricopa County may
be particularly heat-adapted (Hartz et al.
2013). Because the presence of dangerously
hot weather in the summer is predictable in
this setting, some residents travel to cooler
places and may be able to avoid activities
that involve heat exposure. During the study
period, heat warnings, networks for water
distribution, and cooling facilities were available to the public. These efforts may have
mitigated the effects of heat on illness and
death. There are potential modifiers of the
temperature-health relationship that we did
182
not examine, including air pollution, time
of season, cumulative days of high temperatures, and displacement. The applications of
this framework should be updated continually. Trigger points should be monitored and
evaluated for changes because of temporal
variability in weather and climate [indicated
by the reevaluation of climate “normals,”
Arguez et al. (2012)] and because the
behavior of people, the physical environment
(e.g., building materials), the availability of
technology (e.g., air conditioning), and public
health systems adapt to higher temperatures
in ways that may affect the human health
response to heat (Guo et al. 2012). Finally,
the meteorological data were obtained from a
single station, whereas the health events were
experienced across a larger geographic area.
Conclusion
In summary, this study found strong and
consistent associations of environmental
temperature with all-cause mortality, CVD
mortality, heat-related mortality, hospitalization, and ED visits and with a category of
conditions considered possible consequences
of heat or dehydration based on pathophysio
logic reasoning. Consideration of different
health events and various conceptualizations
of threshold temperatures revealed large
contrasts in the trigger points at which activation of different heat intervention efforts
might be appropriate. Plans to mitigate the
effects of high environmental heat on human
health that incorporate different levels of
sensitivity for determining the most effective
adaptation strategies and when to deploy them
might have important benefits in terms of
illnesses and deaths avoided.
References
Agency for Healthcare Research and Quality (AHRQ).
Healthcare Cost and Utilization Project (HCUP).
Overview of the State Inpatient Databases
(SID). Available http://www.hcup-us.ahrq.gov/
sidoverview.jsp [accessed 24 December 2015].
AHS (American Housing Survey). 2014. American Housing
Survey Data Homepage. Available: http://factfinder2.
census.gov/faces/nav/jsf/pages/searchresults.
xhtml?refresh=t [accessed 21 July 2014].
Alessandrini E, Zauli Sajani S, Scotto F, Miglio R,
Marchesi S, Lauriola P. 2011. Emergency ambulance dispatches and apparent temperature: a
time series analysis in Emilia-Romagna, Italy.
Environ Res 111:1192–1200.
Anderson BG, Bell ML. 2009. Weather-related mortality:
how heat, cold, and heat waves affect mortality in
the United States. Epidemiology 20:205–213.
Anderson BG, Bell ML, Peng RD. 2013. Methods to
calculate the heat index as an exposure metric
in environmental health research. Environ Health
Perspect 121:1111–1119; doi:10.1289/ehp.1206273.
Arguez A, Durre I, Applequist S, Vose RS, Squires MF,
Yin X, et al. 2012. NOAA’s 1981–2010 U.S. climate
normals: an overview. Bull Amer Meteor Soc
93:1687–1697.
Armstrong BG, Chalabi Z, Fenn B, Hajat S, Kovats S,
volume
Milojevic A, et al. 2011. Association of mortality
with high temperatures in a temperate climate:
England and Wales. J Epidemiol Community
Health 65:340–345.
Barnett AG, Tong S, Clements ACA. 2010. What
measure of temperature is the best predictor of
mortality? Environ Res 110:604–611.
Basu R, Pearson D, Malig B, Broadwin R, Green R.
2012. The effect of high ambient temperature on
emergency room visits. Epidemiology 23:813–820.
Berko J, Ingram DD, Saha S, Parker JD. 2014. Deaths
attributed to heat, cold, and other weather events
in the United States, 2006–2010. Natl Health Stat
Report 76:1–15.
Brunetti ND, Amoruso D, De Gennaro L,
Dellegrottaglie G, Di Giuseppe G, Antonelli G, et al.
2014. Hot spot: impact of July 2011 heat wave in
southern Italy (Apulia) on cardiovascular disease
assessed by emergency medical service and telemedicine support. Telemed J E Health 20:272–281.
Curriero FC, Heiner KS, Samet JM, Zeger SL, Strug L,
Patz JA. 2002. Temperature and mortality in 11
cities of the eastern United States. Am J Epidemiol
155:80–87.
Davis RE, Knappenberger PC, Michaels PJ,
Novicoff WM. 2003. Changing heat-related
mortality in the United States. Environ Health
Perspect 111:1712–1718; doi:10.1289/ehp.6336.
Guo Y, Barnett AG, Tong S. 2012. High temperaturesrelated elderly mortality varied greatly from year
to year: important information for heat-warning
systems. Sci Rep 2:830. doi:10.1038/srep00830.
Hajat S, Armstrong B, Baccini M, Biggeri A, Bisanti L,
Russo A, et al. 2006. Impact of high temperatures
on mortality: is there an added heat wave effect?
Epidemiology 17:632–638.
Hajat S, Kosatsky T. 2010. Heat-related mortality: a
review and exploration of heterogeneity. J Epidemiol
Community Health 64:753–760.
Harlan SL, Chowell G, Yang S, Petitti DB, Morales
Butler EJ, Ruddell BL, et al. 2014. Heat-related
deaths in hot cities: estimates of human tolerance
to high temperature thresholds. Int J Environ Res
Public Health 11:3304–3326.
Hartz DA, Brazel AJ, Golden JS. 2013. A comparative
climate analysis of heat-related emergency 911
dispatches: Chicago, Illinois and Phoenix, Arizona
USA 2003 to 2006. Int J Biometeorol 57:669–678.
Hastie TJ, Tibshirani RJ. 1990. Generalized Additive
Models. Boca Raton, FL:Chapman and Hall.
Henderson SB, Kosatsky T. 2012. A data-driven approach
to setting trigger temperatures for heat health emergencies. Can J Public Health 103:227–230.
Hess JJ, Saha S, Luber G. 2014. Summertime acute
heat illness in U.S. emergency departments
from 2006 through 2010: analysis of a nationally
representative sample. Environ Health Perspect
122(11):1209–1215; doi:10.1289/ehp.1306796.
Hondula DM, Davis RE, Rocklöv J, Saha MV. 2013. A
time series approach for evaluating intra-city heatrelated mortality. J Epidemiol Community Health
67:707–712.
Hondula DM, Vanos JK, Gosling SN. 2014. The SSC:
a decade of climate–health research and future
directions. Int J Biometeorol 58:109–120.
Huang C, Barnett AG, Wang X, Vaneckova P,
FitzGerald G, Tong S. 2011. Projecting future heatrelated mortality under climate change scenarios:
a systematic review. Environ Health Perspect
119:1681–1690; doi:10.1289/ehp.1103456.
Keatinge WR, Donaldson GC, Cordioli E, Martinelli M,
Kunst AE, Mackenbach JP, et al. 2000. Heat
related mortality in warm and cold regions of
Europe: observational study. BMJ 321:670–673.
124 | number 2 | February 2016 • Environmental Health Perspectives
New evidence for quantifying heat-health impacts
Kinney PL, O’Neill MS, Bell ML, Schwartz J. 2008.
Approaches for estimating effects of climate
change on heat-related deaths: challenges and
opportunities. Environ Sci Policy 11:87–96.
Kovats RS, Hajat S, Wilkinson P. 2004. Contrasting
patterns of mortality and hospital admissions
during hot weather and heat waves in Greater
London, UK. Occup Environ Med 61:893–898.
Kravchenko J, Abernethy AP, Fawzy M, Lyerly HK.
2013. Minimization of heatwave morbidity and
mortality. Am J Prev Med 44:274–282.
Li T, Horton RM, Kinney P. 2013. Projections of
seasonal patterns in temperature-related deaths
for Manhattan. Nat Clim Chang 3:717–721.
Loughnan M, Nicholls N, Tapper N. 2010. Mortality–
temperature thresholds for ten major population
centres in rural Victoria, Australia. Health Place
16:1287–1290.
Martiello MA, Giacchi MV. 2010. High temperatures
and health outcomes: a review of the literature.
Scand J Public Health 38:826–837.
MCDPH (Maricopa County Department of Public
Health). 2014. Heat-Associated Deaths in Maricopa
County, AZ. Final Report for 2012. Available: http://
www.maricopa.gov/publichealth/Services/EPI/pdf/
heat/2012annualreport.pdf [accessed 1 July 2014].
McMichael AJ, Wilkinson P, Kovats RS, Pattenden S,
Hajat S, Armstrong B. 2008. International study
of temperature, heat and urban mortality: the
‘ISOTHURM’ project. Int J Epidemiol 37:1121–1131.
Nitschke M, Tucker GR, Hansen AL, Williams S,
Environmental Health Perspectives •
volume
Zhang Y, Bi P. 2011. Impact of two recent extreme
heat episodes on morbidity and mortality in
Adelaide, South Australia: a case-series analysis.
Environ Health 10:42; doi:10.1186/1476-069X-10–42.
NOAA (National Oceanic and Atmospheric
Administration). 2013. Comparative Climatic Data
for the United States through 2012. Available:
http://www1.ncdc.noaa.gov/pub/data/ccd-data/
CCD-2014.pdf [accessed 25 July 2014].
NWS (National Weather Service). 2014. Weather
Prediction Center. The Heat Index Equation.
Available: http://www.wpc.ncep.noaa.gov/html/
heatindex_equation.shtml [accessed 15 May 2014].
Pascal M, Laaidi K, Ledrans M, Baffert E, CaserioSchönemann C, Le Tertre A, et al. 2006. France’s
heat health watch warning system. Int J
Biometeorol 50:144–153.
Pillai SK, Noe RS, Murphy MW, Vaidyanathan A,
Young R, Kieszak S, et al. 2014. Heat illness:
predictors of hospital admissions among emergency department visits—Georgia, 2002–2008.
J Community Health 39:90–98.
Pudpong N, Hajat S. 2011. High temperature effects on
out-patient visits and hospital admissions in Chiang
Mai, Thailand. Sci Total Environ 409:5260–5267.
Schaffer A, Muscatello D, Broome R, Corbett S,
Smith W. 2012. Emergency department visits,
ambulance calls, and mortality associated with
an exceptional heat wave in Sydney, Australia,
2011: a time-series analysis. Environ Health 11:3;
doi:10.1186/1476-069X-11–3.
124 | number 2 | February 2016
Steadman RG. 1979. The assessment of sultriness.
Part I: a temperature-humidity index based on
human physiology and clothing science. J Appl
Meteor 18:861–873.
Turner LR, Barnett AG, Connell D, Tong S. 2012.
Ambient temperature and cardiorespiratory
morbidity: a systematic review and meta-analysis.
Epidemiology 23:594–606.
Williams S, Nitschke M, Sullivan T, Tucker GR,
Weinstein P, Pisaniello DL, et al. 2012a. Heat and
health in Adelaide, South Australia: assessment of
heat thresholds and temperature relationships. Sci
Total Environ 414:126–133.
Williams S, Nitschke M, Weinstein P, Pisaniello DL,
Parton KA, Bi P. 2012b. The impact of summer
temperatures and heat waves on mortality and
morbidity in Perth, Australia 1994-2008. Environ Int
40:33–38.
Ye X, Wolff R, Yu W, Vaneckova P, Pan X, Tong S. 2012.
Ambient temperature and morbidity: a review
of epidemiological evidence. Environ Health
Perspect 120:19–28; doi:10.1289/ehp.1003198.
Zaninović K, Matzarakis A. 2014. Impact of heat
waves on mortality in Croatia. Int J Biometeorol
58:1135–1145; doi:10.1007/s00484-013-0706–3.
Zhang K, Rood RB, Michailidis G, Oswald EM,
Schwartz JD, Zanobetti A, et al. 2012. Comparing
exposure metrics for classifying ‘dangerous heat’
in heat wave and health warning systems. Environ
Int 46:23–29.
183