Climate Change Indicators in the United States Technical Documentation Cold Related Deaths

Cold-Related Deaths

Identification

1. Indicator Description

Unusually cold winter temperatures (e.g., multi-day cold waves) have become less frequent in the
United States in recent decades (see the High and Low Temperatures indicator), and studies project that
the frequency and intensity of extreme cold waves will decrease as a consequence of climate change
and warming temperatures (USGCRP, 2017). When people are exposed to extreme cold, they can suffer
from cold-related illnesses, such as frostbite and potentially deadly hypothermia. Thus, as extreme cold
events decrease, the risk of cold-related deaths and illness is also expected to decrease (Sarofim et al.,
2016). Tracking the rate of reported overall cold-related deaths over time and in conjunction with
tracking heat-related deaths provides a measure of how climate change and changing temperatures may
affect human well-being.

Components of this indicator include:

• The rate of U.S. annual deaths between 1979 and 2016 for which cold was classified on death
certificates as the underlying (direct) cause (Figure 1, orange line).

• The rate of U.S. deaths between 1999 and 2015, for which cold was classified as either the
underlying cause or a contributing factor (Figure 1, blue line).

2. Revision History

August 2017: Indicator published.

April 2021: Indicator updated with contributing cause of death data through 2015 and underlying

cause of death data through 2016.

Data Sources

3. Data Sources

This indicator is based on data from the U.S. Centers for Disease Control and Prevention's (CDC's)
National Vital Statistics System (NVSS), which compiles information from death certificates for nearly
every death in the United States. The NVSS is the most comprehensive source of mortality data for the
population of the United States. The CDC provided analysis of NVSS data.

4. Data Availability

Underlying Causes

The long-term time series (1979-2016) in Figure 1 is based on CDC's Compressed Mortality File, which
can be accessed through the CDC WONDER online database at: https://wonder.cdc.gov/mortSQL.html

Technical Documentation: Cold-Related Deaths

-------
(CDC, 2018a). CDC WONDER provides free public access to mortality statistics, allowing users to query
data for the nation as a whole or by state or region, demographic group (age, sex, race), or International
Classification of Diseases (ICD) code. Users can obtain the data for this indicator by accessing CDC
WONDER and querying the ICD codes listed in Section 5 for the entire U.S. population.

Underlying and Contributing Causes

The 1999-2015 time series in Figure 1 is based on an analysis conducted by the National Environmental
Public Health Tracking (EPHT) Program, which CDC coordinates. CDC staff from the National Center for
Environmental Health provided monthly national totals to EPA (CDC, 2018b). This analysis is similar to an
analysis that EPHT has published for heat-related deaths at: https://ephtracking.cdc.gov/DataExplorer.

Users can query underlying and contributing causes of death through CDC WONDER's Multiple Causes of
Death file (https://wonder.cdc.gov/mcd-icdlO.html). but note that EPHT performed additional steps for
Figure 1 that cannot be recreated through the publicly available data portal (see Section 6).

Death Certificates

Individual-level data (i.e., individual death certificates) are not publicly available due to confidentiality
issues.

Methodology

5. Data Collection

This indicator is based on causes of death as reported on death certificates. A death certificate typically
provides space to designate an immediate cause of death along with up to 20 contributing causes, one
of which will be identified as the underlying cause of death. The World Health Organization (WHO)
defines the underlying cause of death as "the disease or injury which initiated the train of events leading
directly to death, or the circumstances of the accident or violence which produced the fatal injury."

Causes of death are certified by a physician, medical examiner, or coroner, and are classified according
to a standard set of codes called the ICD. Deaths for 1979 through 1998 are classified using the Ninth
Revision of ICD (ICD-9). Deaths for 1999 and beyond are classified using the Tenth Revision (ICD-10).

Although causes of death rely to some degree on the judgment of the physician, medical examiner, or
coroner, the "measurements" for this indicator are expected to be generally reliable based on the
medical knowledge required of the "measurer" and the use of a standard classification scheme based on
widely accepted scientific definitions. When more than one cause or condition is entered, the underlying
cause is determined by the sequence of conditions on the certificate, provisions of the ICD, and
associated selection rules and modifications.

Mortality data are collected for the entire population and, therefore, are not subject to sampling design
error. For virtually every death that occurs in the United States, a physician, medical examiner, or
coroner certifies the causes of death on an official death certificate. State registries collect these death
certificates and report causes of death to the NVSS. NVSS's shared relationships, standards, and
procedures form the mechanism by which the CDC collects and disseminates the nation's official vital
statistics.

Technical Documentation: Cold-Related Deaths

-------
Standard forms for the collection of data and model procedures for the uniform registration of death
events have been developed and recommended for state use through cooperative activities of the
states and CDC's National Center for Health Statistics (NCHS). All states collect a minimum data set
specified by NCHS, including underlying causes of death and basic demographic data (e.g., age, race, and
ethnicity). CDC has published procedures for collecting vital statistics data (CDC, 1995).

This indicator excludes deaths to foreign residents and deaths to U.S. residents who died abroad.

General information regarding data collection procedures can be found in the Model State Vital
Statistics Act and Regulations (CDC, 1995). For additional documentation on the CDC WONDER database
(EPA's data source for part of this indicator) and its underlying sources, see:
https://wonder.cdc.gov/wonder/help/cmf.html.

CDC has posted a recommended standard certificate of death online at:

www.cdc.gov/nchs/data/dvs/DEATHll-03final-ACC.pdf. For a complete list and description of the ICD
codes used to classify causes of death, see: https://icd.who.int/browsel0/2019/en.

6. Indicator Derivation

This indicator reports annual rates of deaths per million population that have been classified with ICD
codes related to exposure to natural sources of cold. The NVSS collects data on virtually all deaths that
occur in the United States, meaning the data collection mechanism already covers the entire target
population. Thus, it was not necessary to extrapolate the results on a spatial or population basis. No
attempt has been made to reconstruct trends prior to the onset of comprehensive data collection, and
no attempt has been made to project data forward into the future.

Underlying Causes

The long-term trend line in Figure 1 reports the rate of deaths per year for which the underlying cause
had one of the following ICD codes:

• ICD-9 code E901: "excessive cold—hypothermia"—specifically subpart E901.0: "due to weather
conditions."

• ICD-10 code X31: "exposure to excessive natural cold—hypothermia."

This component of the indicator is reported for the entire year. EPA developed this analysis based on
the publicly available data compiled by CDC WONDER. EPA chose to use crude death rates rather than
death counts because rates account for changes in total population over time. Population figures are
obtained from CDC WONDER.

Underlying and Contributing Causes

The "underlying and contributing causes" trend line in Figure 1 reports the rate of deaths for which
either the underlying cause or the contributing causes had one or more of the following ICD codes:

• ICD-10 code X31: "exposure to excessive natural cold—hypothermia."

Technical Documentation: Cold-Related Deaths

-------
• ICD-10 code T68: "hypothermia associated with low environmental temperature." Note that
this code is used for contributing causes—never for the underlying cause.

• ICD-10 code T69: "other effects of reduced temperature." Note that this code is used for
contributing causes—never for the underlying cause.

To reduce the chances of including deaths that were incorrectly classified, EPHT did not count the
following deaths:

• Any deaths for which the ICD-10 code W93: "exposure to excessive cold of man-made origin/'
appears in any cause field. This step removes certain occupational-related deaths—for
example, being trapped in a freezer.

• Any deaths for which the ICD-10 code R68.0: "hypothermia, not associated with low
environmental temperature," appears in any cause field. This step excludes certain causes of
hypothermia that are not associated with exposure to cold environmental conditions.

• Any deaths for which the ICD-10 code P80: "hypothermia of newborn," appears in any cause
field. This step excludes deaths that may be due to a newborn child's inability to regulate body
temperature.

Foreign residents were excluded. EPHT obtained death counts directly from NVSS, rather than using the
processed data available through CDC WONDER. EPHT has not yet applied its methods to data prior to
1999. Crude death rates were calculated in the same manner as with the underlying causes time series.

7. Quality Assurance and Quality Control

Vital statistics regulations have been developed to serve as a detailed guide to state and local
registration officials who administer the NVSS. These regulations provide specific instructions to protect
the integrity and quality of the data collected. This quality assurance information can be found in CDC
(1995).

For the "underlying and contributing causes" component of this indicator, extra steps have been taken
to remove certain deaths that could potentially reflect a misclassification (see Section 6). These criteria
generally excluded only a small number of deaths.

Analysis

8. Comparability Over Time and Space

When plotting the long-term data in Figure 1, EPA inserted a break in the line between 1998 and 1999 to
reflect the transition from ICD-9 codes to ICD-10 codes. The change in codes makes it difficult to
accurately compare pre-1999 data with data from 1999 and later. Otherwise, all methods have been
applied consistently over time and space. ICD codes allow physicians and other medical professionals
across the country to use a standard scheme for classifying causes of deaths.

Technical Documentation: Cold-Related Deaths

-------
9. Data Limitations

Factors that may impact the confidence, application, or conclusions drawn from this indicator are as
follows:

1. Some deaths associated with cold exposure are not identified as such by the medical examiner
and might not be properly coded on the death certificate. In many cases, they might just
classify the cause of death as a cardiovascular or respiratory disease, not knowing for certain
whether cold was a contributing factor. These causes of death are much more common than
cold-related illnesses such as hypothermia. Furthermore, deaths can occur from exposure to
more moderately cold temperature (either as an underlying cause or as a contributing factor)
that is not classified as extreme and therefore is often not recorded as such. The stress of
several cold days can increase one's risk of dying from conditions such as myocardial infarction
or cardiac arrest (Medina-Ramon and Schwartz, 2007; Braga et al., 2002), and the effects of
cold on the body tend to persist for longer than the effects of heat (Gasparrini et al., 2015),
which could make it even more difficult to connect deaths to exposures. Still, this indicator
serves as a reportable national measure of overall deaths directly attributable to cold and can
be used as evidence for more focused investigations.

2. ICD-9 codes were used to specify underlying cause of death for the years 1979 to 1998.
Beginning in 1999, cause of death was specified with ICD-10 codes. The two revisions differ
substantially, so data from before 1999 cannot easily be compared with data from 1999 and
later.

3. The fact that a death is classified as "cold-related" does not mean that cold temperatures were
the only factor that caused the death. Pre-existing medical conditions, certain medications,
alcohol or drug use, and other factors can greatly increase an individual's vulnerability to cold.
Some deaths classified as "cold-related" reflect human behavior more than weather
conditions—for instance, deaths in which alcohol abuse increased a person's susceptibility to
cold weather, falling into cold water (which can cause hypothermia, even in the summer), or
getting lost on a wilderness excursion. This indicator includes these types of deaths.
Nonetheless, deaths classified as "cold-related" are much more common in colder months of
the year than in warmer months, as Figure TD-1 shows.

4. Cold weather is not the only factor that can affect trends in "cold-related" deaths. Other
factors include the vulnerability of the population and the extent to which people have adapted
and acclimated to cold temperatures.

Technical Documentation: Cold-Related Deaths

-------
Figure TD-1. Deaths Classified as "Cold-Related" in the United States by Month, 1999-2015

6,000

January February March April May June July August September October November December

Month

This graph shows all deaths for which cold was recorded as either the underlying cause or a contributing
cause of death, which corresponds with the blue line in Figure 1 of this indicator. Deaths have been
aggregated by month over the entire period of record.

Data source: CDC, 2018b

10. Sources of Uncertainty

Uncertainty estimates are not available for this indicator. Because statistics have been gathered from
virtually the entire target population (i.e., all deaths in a given year), these data are not subject to the
same kinds of errors and uncertainties that would be inherent in a probabilistic survey or other type of
representative sampling program.

Some uncertainty could be introduced as a result of the professional judgment required of the medical
professionals filling out the death certificates, which could potentially result in misclassification or
underreporting in some number of cases.

11. Sources of Variability

There is substantial year-to-year variability within the data. Further study is needed to determine the
extent to which year-to-year variations in cold-related deaths might reflect general weather conditions
(e.g., a particularly cold winter in a certain region), the status of winter illness in the population (e.g., a
particularly bad strain of influenza), or specific extreme cold events.

Technical Documentation: Cold-Related Deaths

-------
12. Statistical/Trend Analysis

This indicator does not report on the slope of the apparent trends in cold-related deaths, nor does it
calculate the statistical significance of these trends.

References

Braga, A.L., A. Zanobetti, and J. Schwartz. 2002. The effect of weather on respiratory and cardiovascular
deaths in 12 U.S. cities. Environ. Health Perspect. 110(9):859-863.

CDC (U.S. Centers for Disease Control and Prevention). 1995. Model State Vital Statistics Act and
Regulations (revised April 1995). DHHS publication no. (PHS) 95-1115.
www.cdc.gov/nchs/data/misc/mvsact92aacc.pdf.

CDC (U.S. Centers for Disease Control and Prevention). 2018a. Mortality data on CDC WONDER:
Underlying cause-of-death: All ages deaths from compressed mortality: Mortality for 1999-2016 with
ICD 10 codes, mortality for 1979-1998 with ICD 9 codes. Accessed March 2018.
https://wonder.cdc.gov/mortSQL.html.

CDC (U.S. Centers for Disease Control and Prevention). 2018b. Monthly national totals provided by
National Center for Environmental Health staff in March 2018, based on contributing causes of death
with exclusion criteria applied.

Gasparrini, A., et al. 2015. Mortality risk attributable to high and low ambient temperature: A
multicountry observational study. Lancet 386(9991):369-375.

Medina-Ramon, M., and J. Schwartz. 2007. Temperature, temperature extremes, and mortality: A study
of acclimatization and effect modification in 50 U.S. cities. Occup. Environ. Med. 64(12):827-833.

Sarofim, M.C., S. Saha, M.D. Hawkins, D.M. Mills, J. Hess, R. Horton, P. Kinney, J. Schwartz, and A. St.
Juliana. 2016. Chapter 2: Temperature-related death and illness. In: The impacts of climate change on
human health in the United States: A scientific assessment. U.S. Global Change Research Program.
https://health2016.globalchange.gov.

USGCRP (U.S. Global Change Research Program). 2017. Climate science special report: Fourth National
Climate Assessment, volume I. Wuebbles, D.J., D.W. Fahey, K.A. Hibbard, D.J. Dokken, B.C. Stewart, and
T.K. Maycock (eds.). https://science2017.globalchange.gov. doi:10.7930/J0J964J6

Technical Documentation: Cold-Related Deaths

7

-------