Environmental Public Health Tracking: Climate Change
High temperatures can cause many health problems, such as heat rash, swelling, cramps, fainting, and heat stroke. We track heat-related illness to prepare for future heat events and warn the public when they are at risk. These data have implications for everyone at work and play in Wisconsin, especially when temperatures are warming up.
Explore definitions and explanations of terminology found on this webpage, like age-adjusted rate and confidence intervals.
Frequently asked questions
Heat-related illness occurs when the body’s temperature and control system becomes overloaded. Normally, the body cools itself by sweating, but this cooling mechanism can become ineffective if the body’s temperature rises too fast.
There are several forms of heat-related illness, including heat stroke, heat exhaustion, rhabdomyolysis, heat syncope, heat cramps, and heat rash.
Old age, youth ages 0-4, obesity, fever, dehydration, heart disease, mental illness, poor circulation, sunburn, prescription medication use, and alcohol use are factors that impact the body’s ability to regulate temperature.
Workers whose jobs require them to work outside in hot weather are also at risk of heat-related illness.
As a result of climate change, events like heat waves happen more often.
The frequency of heat waves may impact how often people suffer from heat-related illness.
- Tracking extreme heat events and heat-related injury gives public health professionals a better understanding of the health consequences of extreme heat across the country. We can monitor the impact of our warnings and preparedness efforts.
- Projecting extreme heat events can help certain areas prepare for these events in advance.
All of the heat and heat-related illness measures are from the CDC's (Centers for Disease Control and Prevention) National Environmental Public Health Tracking portal. See below for details about the original data sources.
- The original source of the emergency department (ED) and hospitalization data as displayed on the Wisconsin Tracking Portal is the Wisconsin Hospital Association Information Center, Inc.
- Heat-related mortality data are from the CDC’s National Center for Health Statistics.
- The North American Land Data Assimilation System from the National Aeronautics and Space Administration is the original source of the historical temperature data.
- Modeled temperature data obtained from 1/8 degree-CONUS Daily Downscaled Climate Projections by Katharine Hayhoe is the source for projected heat data.
- The vulnerability and preparedness data are from the Wisconsin Hospital Association Information Center, Inc. the U.S. Census Bureau, CDC’s Behavioral Risk Factor Surveillance System, American Hospital Association Annual Survey, and National Land Cover Database.
- Historical Temperature and Heat Index
- Number of extreme heat days
- Temperature and Heat Projections
- Projected number of future extreme heat days
- Projected number of future extreme heat nights
- Heat-related illness
- Heat-related emergency department visits
- Age-adjusted rate of emergency department visits per 100,000 population
- Crude rate of emergency department visits for heat stress per 100,000 population
- Number of emergency department visits for heat stress
- Heat-related hospitalizations
- Age-adjusted rate of hospitalizations for heat stress per 100,000 population
- Crude rate of hospitalizations for heat stress per 100,000 population
- Number of hospitalizations for heat stress
- Heat-related mortality
- Number of summertime (May-September) heat-related deaths, by year
- Heat-related emergency department visits
- Vulnerability and Preparedness: Heat
- Age-adjusted rate of hospitalization for heart attack per 10,000 population
- Median household income
- Number and percent of people living in poverty
- Number and percent of people without health insurance
- Heat index data takes both humidity and temperature into account.
- Hospital admission and emergency department visit data do not include people who experience symptoms but are not seen in the emergency room or admitted to the hospital.
- These data do not include inpatient admissions or emergency department visits at hospitals owned by the federal government, such as Veterans Administration hospitals.
- The death certificate dataset may be missing a small number of cases where the decedent is a Wisconsin resident but died in another state.
- Data users should keep in mind that many factors contribute to illness. These factors should be considered when interpreting the data. Factors include the following:
- Demographics (race, gender, age)
- Socioeconomic status (income level, education)
- Geography (rural, urban)
- Changes in the medical field (diagnosis patterns, reporting requirements)
- Individual behavior (diet, smoking)
- CDC – Extreme Heat and Your Health
- CDC – Heat Stress – Heat Related Illness
- CDC – Frequently Asked Questions (FAQ) About Extreme Heat
- Wisconsin Department of Health Services – Heat-Related Health and Safety Tips
- Wisconsin Department of Health Services – Climate and Health Program
- Wisconsin Initiative on Climate Change Impacts
- ReadyWisconsin Extreme Heat Preparedness
- NIOSH Protect Workers from Heat with Acclimatization
- NIOSH Work-Rest Schedules for Working in Hot Environments (PDF)
- OSHA Working in Indoor and Outdoor Heat
Click the link below to download the data you're looking for:
Heat and heat-related illness data details
Heat-related mortality
Number of heat-related deaths (counts)
These data include heat-related deaths collected from the Multiple Cause Mortality files from CDC’s National Center for Health Statistics. These data include deaths in which excessive heat exposure (ICD-10 code X30) or effects of heat and light (ICD-10 code T67) is reported as either the underlying or contributing cause of death. Deaths due to exposure to excessive heat of man-made origins (W92) are excluded. Only deaths that occurred in the summer months (May through September) are included in this measure. Data are suppressed when the number of deaths is less than 10 (including 0).
Heat-related ED visits
Age-adjusted rate of emergency department visits for heat stress per 100,000 population
These data include ED visits for heat stress and are collected from emergency room visit discharge records. ED visits resulting in subsequent hospitalization are also included. This measure includes ED visits with an ICD-9 code of 992.0-992.9 or an ICD-10 code of T67.0-T67.9, X30, or X32 in the principal or other diagnosis fields. ED visits due to exposure to excessive heat of man-made origin (ICD-9 code E900.1 or ICD-10 code W92) are excluded. Data for Wisconsin residents visiting an ED in Iowa and Minnesota are included. Counties with non-zero counts less than 6 and population less than 100,000 are suppressed for this measure.
An age-adjusted rate is statistically modified to eliminate the potential biasing effect of different age distributions across different populations. In other words, the age-adjusted rate accounts for the possibility that ED visits for heat stress may be more frequent among younger individuals and some counties have a higher population of younger individuals than others.
Direct age-adjustment is conducted using the 2000 U.S. standard population. Federally funded hospitals (for example, Veteran’s Administration (VA) hospitals, which are exempt from state reporting requirements) and are not included in these data. Data are based on the number of admissions, not the number of individuals admitted to the hospital.
Prior to 2018, the county variable in the emergency department data was assigned based on the zip code of the residence of the patient. In cases where a zip code spanned more than one county, the ED visit was placed into the county with the higher population. Beginning in 2018, geocoded data became available in Wisconsin ED visit data, which allows for a more accurate determination of county of residence. The new block group variable is the geocoded latitude/longitude of the patient’s residential address. The first five digits of the block group variable are used to determine the Wisconsin county. Due to this change, caution should be used when comparing county-level data prior to 2018 with data from 2018 onward.
Crude rate of ED visits for heat stress per 100,000 population
These data include emergency department visits for heat stress and are collected from emergency room visit discharge records. This measure includes emergency department visits with an ICD-9 code of 992.0-992.9, E900.9 or an ICD-10 code of T67.0-T67.9, X30, or X32 in the principal or other diagnosis fields. ED visits due to exposure to excessive heat of man-made origin (ICD-9 code E900.1 or ICD-10 code W92) are excluded. Data for Wisconsin residents hospitalized in Iowa and Minnesota are included. Counties with non-zero counts less than 6 and population less than 100,000 are suppressed for this measure. Federally funded hospitals (for example Veteran’s Administration [VA] hospitals, which are exempt from state reporting requirements) are not included in these data. Data are based on the number of admissions, not the number of individuals admitted to the hospital. The crude rate is the number of ED visits divided by the total number of people in that age category (for example, people aged 65+). This is expressed as a number per unit population such as “per 100,000 people.” The crude rate does not take into account the differences in age distributions across counties and are therefore subject to bias. Use age-adjusted rates for a better standardized measure.
Prior to 2018, the county variable in the emergency department data was assigned based on the zip code of the residence of the patient. In cases where a zip code spanned more than one county, the ED visit was placed into the county with the higher population. Beginning in 2018, geocoded data became available in Wisconsin ED visit data, which allows for a more accurate determination of county of residence. The new block group variable is the geocoded latitude/longitude of the patient’s residential address. The first five digits of the block group variable are used to determine the Wisconsin county. Due to this change, caution should be used when comparing county-level data prior to 2018 with data from 2018 onward.
Number of ED visits for heat stress
These data include emergency department visits for heat stress and are collected from emergency room visit discharge records. This measure includes emergency department visits with an ICD-9 code of 992.0-992.9, E900.0 or E900.9 or an ICD-10 code of T67.0-T67.9, X30, or X32 in the principal or other diagnosis fields. ED visits due to exposure to excessive heat of man-made origin (ICD-9 code for E900.1 or ICD-10 code W92) are excluded. Data for Wisconsin residents hospitalized in Iowa and Minnesota are included. Counties with non-zero counts less than six and population less than 100,000 are suppressed for this measure. Federally funded hospitals (for example VA hospitals, which are exempt from state reporting requirements) are not included in these data. Data are based on the number of admissions, not the number of individuals admitted to the hospital. Please note that counts are a statistically limited way to consider ED visits because they do not allow for accurate comparison between counties. Counties with higher populations, such as Milwaukee, will have more emergency visits simply because they have more people. An age-adjusted rate is a better measure for true comparison between counties.
Prior to 2018, the county variable in the emergency department data was assigned based on the zip code of the residence of the patient. In cases where a zip code spanned more than one county, the ED visit was placed into the county with the higher population. Beginning in 2018, geocoded data became available in Wisconsin ED visit data, which allows for a more accurate determination of county of residence. The new block group variable is the geocoded latitude/longitude of the patient’s residential address. The first five digits of the block group variable are used to determine the Wisconsin county. Due to this change, caution should be used when comparing county-level data prior to 2018 with data from 2018 onward.
Heat-related hospitalizations
Age-adjusted rate of hospitalizations for heat stress per 100,000 population
These data include hospitalizations for heat stress and are collected from inpatient hospital discharge records. This measure includes hospitalizations with an ICD-9 code of 992.0-992.9, E900.0 or E900.9 or an ICD-10 code of T67.0-T67.9, X30, or X32 in the principal or other diagnosis fields. Hospitalizations due to exposure to excessive heat of man-made origin (ICD-9 code for E900.1 or ICD-10 code W92) are excluded. Data for Wisconsin residents hospitalized in Iowa and Minnesota are included. Counties with non-zero counts less than six and population less than 100,000 are suppressed for this measure. Federally, funded hospitals (for example VA hospitals, which are exempt from state reporting requirements) are not included in these data. Data are based on the number of admissions, not the number of individuals admitted to the hospital. An age-adjusted is statistically modified to eliminate the potential biasing effect of different age distributions across different populations. In other words, the age-adjusted rates accounts for the possibility that hospitalizations for heat stress may be more frequent among younger individuals and some counties have a higher population of younger individuals than others. Direct age-adjustment is conducted using the 2000 US standard population.
Prior to 2018, the county variable in the hospitalization data was assigned based on the zip code of the residence of the patient. In cases where a zip code spanned more than one county, the hospitalization was placed into the county with the higher population. Beginning in 2018, geocoded data became available in Wisconsin hospitalization data, which allows for a more accurate determination of county of residence. The new block group variable is the geocoded latitude/longitude of the patient’s residential address. The first five digits of the block group variable are used to determine the Wisconsin county. Due to this change, caution should be used when comparing county-level data prior to 2018 with data from 2018 onward.
Crude rate of hospitalizations for heat stress per 100,000 population
These data include hospitalizations for heat stress and are collected from inpatient hospital discharge records. This measure includes hospitalizations with an ICD-9 code of 992.0-992.9, E900.0 or E900.9 or an ICD-10 code of T67.0-T67.9, X30, or X32 in the principal or other diagnosis fields. Hospitalizations due to exposure to excessive heat of man-made origin (ICD-9 code for E900.1 or ICD-10 code W92) are excluded. Data for Wisconsin residents hospitalized in Iowa and Minnesota are included. Counties with non-zero counts less than six and population less than 100,000 are suppressed for this measure. Federally funded hospitals (for example VA hospitals, which are exempt from state reporting requirements) are not included in these data. Data are based on the number of admissions, not the number of individuals admitted to the hospital. The crude rate is the number of hospitalizations divided by the total number of people in the area of interest (for example, a county). This is expressed as a number per unit population, such as “per 100,000 people.” The crude rate does not take into account the differences in age distributions across counties and are therefore subject to bias. Use age-adjusted rates for a better standardized measure.
Prior to 2018, the county variable in the hospitalization data was assigned based on the zip code of the residence of the patient. In cases where a zip code spanned more than one county, the hospitalization was placed into the county with the higher population. Beginning in 2018, geocoded data became available in Wisconsin hospitalization data, which allows for a more accurate determination of county of residence. The new block group variable is the geocoded latitude/longitude of the patient’s residential address. The first five digits of the block group variable are used to determine the Wisconsin county. Due to this change, caution should be used when comparing county-level data prior to 2018 with data from 2018 onward.
Number of hospitalizations for heat stress
These data include hospitalizations for heat stress and are collected from inpatient hospital discharge records. This measure includes hospitalizations with an ICD-9 code of 992.0-992.9, E900.0 or E900.9 or an ICD-10 code of T67.0-T67.9, X30, or X32 in the principal or other diagnosis fields. Hospitalizations due to exposure to excessive heat of man-made origin (ICD-9 code for E900.1 or ICD-10 code W92) are excluded. Counties with non-zero counts less than six and population less than 100,000 are suppressed for this measure. Federally funded hospitals (for example VA hospitals, which are exempt from state reporting requirements) are not included in these data. Data are based on the number of admissions, not the number of individuals admitted to the hospital. Please note that counts are a statistically limited way to consider hospitalizations because they do not allow for accurate comparison between counties. Counties with higher populations, such as Milwaukee, will have more hospitalizations simply because they have more people. An age-adjusted rate is a better measure for true comparison between counties.
Prior to 2018, the county variable in the hospitalization data was assigned based on the zip code of the residence of the patient. In cases where a zip code spanned more than one county, the hospitalization was placed into the county with the higher population. Beginning in 2018, geocoded data became available in Wisconsin hospitalization data, which allows for a more accurate determination of county of residence. The new block group variable is the geocoded latitude/longitude of the patient’s residential address. The first five digits of the block group variable are used to determine the Wisconsin county. Due to this change, caution should be used when comparing county-level data prior to 2018 with data from 2018 onward.
Historical temperature and heat index
Number of extreme heat days
The purpose of this dataset is to keep record of past extreme days.
The CONUS Downscaled Climate Projections calculated temperature and precipitation projections using outputs from 16 global change models at 1/8th degree resolution to generate a single comprehensive dataset. Grid-level data were converted into county-level estimates by processing modeled data. The measure involves calculation of a daily maximum heat index, which takes into account relative humidity and temperature. The measure uses a threshold of 90 degrees Fahrenheit to constitute an extreme heat day. Data were collected only for summer months (May through September). Measures for daily maximum temperature and heat index are available in relative thresholds (90th, 95th, 98th, 99th percentiles) and absolute thresholds (90 ̊F, 95 ̊F, 100 ̊F, 105 ̊F).
Temperature and heat projections
Projected difference in extreme heat days as compared to the historical period
These estimates represent the difference in extreme heat days between the time period selected (2016-2045, 2036-2065, 2070-2099) and the referent period (1976-2005). These measures are derived using Localized Constructed Analogs (LOCA): a statistical downscaling technique that uses historical data derived from CMIP5 simulations to predict future climate scenarios. The county-level temperature estimates, which are available at a 1/16th-degree resolution, are obtained by using LOCA to downscale data from 32 global climate models from 1950-2100 with a historical period from 1950-2005. The resulting models are used to estimate two future scenarios (RCP 4.5 and RCP 8.5) over the period 2006-2100. The process of converting grid-level data to county-level estimates uses a population-weighted centroid approach which may lead to the misclassification of temperature for some areas.
The modeling tools are used to make a suite of future climate changes that illustrate the possibilities that may lie ahead. The scenarios include time series of emissions and concentrations of the full suite of greenhouse gases (GHGs), aerosols, and chemically-active gases, as well as land use. The representative concentration pathway (RCP) is used to represent two of many possible scenarios – in this case, the data on the portal is available for RCP 8.5 to represent high emissions and RCP 4.5 to represent low emissions.
Projected difference was determined by either absolute (i.e., 90 ̊F, 95 ̊F, 100 ̊F, 105 ̊F) or relative (i.e., 99th) thresholds. These thresholds are displayed for both RCP scenarios.
Projected difference in extreme heat nights as compared to the historical period
These estimates represent the difference in extreme heat nights between the time period selected (2016-2045, 2036-2065, 2070-2099) and the referent period (1976-2005). These measures are derived using Localized Constructed Analogs (LOCA): a statistical downscaling technique that uses historical data derived from CMIP5 simulations to predict future climate scenarios. The county-level temperature estimates, which are available at a 1/16th-degree resolution, are obtained by using LOCA to downscale data from 32 global climate models from 1950-2100 with a historical period from 1950-2005. The resulting models are used to estimate two future scenarios (RCP 4.5 and RCP 8.5) over the period 2006-2100. The process of converting grid-level data to county-level estimates uses a population-weighted centroid approach which may lead to the misclassification of temperature for some areas.
The modeling tools are used to make a suite of future climate changes that illustrates the possibilities that may lie ahead. The scenarios include time series of emissions and concentrations of the full suite of greenhouse gases (GHGs), aerosols, and chemically-active gases, as well as land use. The representative concentration pathway (RCP) is used to represent two of many possible scenarios – in this case, the data on the portal is available for RCP 8.5 to represent high emissions and RCP 4.5 to represent low emissions.
Projected difference was determined by either absolute (i.e., 75 ̊F, 80 ̊F, 85 ̊F, 90 ̊F) or relative (i.e., 99th) thresholds. These thresholds are displayed for both RCP scenarios.
Vulnerability and preparedness: heat
Age-adjusted estimates of the percent of adults >= 20 years diagnosed with diabetes
These data are provided by the CDC’s National Diabetes Surveillance System. Prevalence rates by county for adults 20 years and older were estimated using data from CDC’s Behavioral Risk Factor Surveillance Systems and age-adjusted using data from U.S. Census Bureau’s population estimates.
Age-adjusted rate of hospitalizations for heart attack among persons 35 and over per 10,000 population
These data include hospitalizations for heart attack and are collected from inpatient hospital discharge records. This measure includes hospitalizations with an ICD-9 code of 410.0-410.92, prior to October 2015 or an IDC-10 code of 121.0-122.9 from October 2015 and on. Starting in 2015, transfers between hospitals were excluded. Data for Wisconsin residents hospitalized in Iowa and Minnesota are included. Counties with non-zero counts less than six and population less than 100,000 are suppressed for this measure. Federally funded hospitals (for example Veteran’s Administration [VA] hospitals, which are exempt from state reporting requirements) are not included in these data. Data are based on the number of admissions, not the number of individuals admitted to the hospital. An age-adjusted rate is a rate that is statistically modified to eliminate the potential biasing effect of different age distributions across different populations. In other words, the age-adjusted rate accounts for the possibility that heart attacks are more frequent among older individuals and some counties have a higher count of older individuals than other. Direct age-adjustment is conducted using the 2000 U.S. standard population.
Median household income
Data are provided by the U.S. Census Bureau, Small Area Income and Poverty Estimates.
Number of hospital beds
Data are obtained from the 2016 American Hospital Association Annual Survey. Facility-level insights are not available for hospitals that did not complete the survey.
Number of hospital beds per 10,000 population
Data are obtained from the 2016 American Hospital Association Annual Survey. Facility-level insights are not available for hospitals that did not complete the survey. The rate of hospital beds per 10,000 was calculated using the total population per county as provided by the U.S. Census Bureau.
Number of hospitals
Data are obtained from the 2016 American Hospital Association Annual Survey. Facility-level insights are not available for hospitals that did not complete the survey.
Number of hospitals per 100,000 population
Data are obtained from the 2016 American Hospital Association Annual Survey. Facility-level insights are not available for hospitals that did not complete the survey. The rate of hospital beds per 10,000 people was calculated using the total population per county as provided by the U.S. Census Bureau.
Number of people living in poverty
Data are provided by the U.S. Census Bureau, Small Area Income and Poverty Estimates. Data are also available for populations aged 0-17 in addition to the rest of the population.
Number of people without health insurance
Data are provided by the U.S. Census Bureau, Small Area Health Insurance Estimates. Data are also available for populations aged 0-18.
Percent of land covered by forest
Data are collected from National Land Cover Database provided by the U.S. Department of the Interior, U.S. Geological Survey. Percentages for each county were estimated as a proportion of gridded data available from the Multi-Resolution Land Characteristics Consortium.
Forest classification is defined in three categories; deciduous forest, evergreen forest, and mixed forest. A deciduous forest is defined as having 75% of trees shedding foliage in the fall. An evergreen forest has 75% of trees maintain their foliage for the full year. In a mixed forest, neither deciduous or evergreen trees make up 75% of tree species.
Percent of land used for development
Data are collected from National Land Cover Database provided by the U.S. Department of the Interior, Multi-Resolution Land Characteristics Consortium. Percentages for each county were estimated as a proportion of gridded data available from the Multi-Resolution Land Characteristics Consortium. Developed Land Use is the combined data from Developed-Open Space, Developed Low-Intensity, Developed-Medium Intensity, and Developed-High Intensity in gridded data from National Land Cover Database.
Developed open spaces are defined as areas with some constructed materials but mostly vegetation (e.g., golf courses).
Percent of population aged 65 years and over living alone in a non-family household
Data are provided by the U.S. Census Bureau, American Factfinder, American Community Survey five-year estimates. Data are displayed by aggregating census tracts to a minimum of 5,000 people or 200,000 people.
Percent of population living in poverty
Data are provided by the U.S. Census Bureau, Small Area Income and Poverty Estimates. Data are available for those populations aged 0-17.
Percent of population of a race other than white
Data are collected from Census 2000 Summary File 3 created by the U.S. Census Bureau.
Percent of population without health insurance
Data are provided by the U.S. Census Bureau, Small Area Health Insurance Estimates. This data does not include those over 65 years covered by Medicare. Data are available for populations aged 0-18 in addition to the rest of the population.