An Epidemiologic Study of Temperature and Mortality in California: Implications for Climate Change
Rupa Basu, PhD, MPH, Research Scientist Bart D. Ostro, PhD, Chief Air Pollution Epidemiology Section/OEHHA California EPA
Public Health Significance
• 400 annual heat-related deaths in US (CDC 2002) • Increase with global warming
– In years with severe heat waves ~1,700 deaths – 2003 European heat wave: >6,000 excess deaths in Italy and 9 French cities – 140 deaths reported during July 2006 CA heat wave
• Heat-related deaths underreported
Mechanisms for Thermoregulation
COLD
.
HOT
1) Shift in blood circulation 2) Stress on heart 3) Sweating
Source: Moffett et al. 1993
Populations at Risk
• • • • • • Elderly People with pre-existing diseases People taking certain medications Infants Low socioeconomic status Socially isolated populations
Background
• Few epidemiologic studies of temperature quantifying mortality risk • Estimates not always comparable • Previous studies did not always control for confounding by pollutants and other factors
– Ostro et al. 2006: PM2.5 and mortality, adjusting for temperature and humidity
Study Objectives
• To assess the impact of apparent temperature on mortality in 9 CA counties
– May 1-September 30, 1999-2003
• To determine how this association differs by cause-specific outcomes, race, age, education level, gender
Data
• Mean daily apparent temperature (EPA AIRS database) – Incorporates temperature and relative humidity • Daily mortality (CA Department of Health Services) – All-cause – All-cause by gender, age, race, education – Cause-specific • Air pollutants (CA Air Resources Board) – PM2.5, O3, CO, NO2
Mean Daily Apparent Temperature (°F) for Nine California Counties, May-September 1999-2003
Mean Apparent Temp deg F
65 - 68 69 - 70
Sacramento 71 15600 Contra Costa 67 11100 Santa Clara 65 15300 Fresno 75 9400
71 - 73 74 - 75 76 - 78
Kern 78 8100
Riverside 75 20200
County Mean Apparent Temp Total Deaths
Los Angeles 69 104700 Orange 72 28400 San Diego 71 35300
Color symbols: ColorBrewer.org
Map by Rachel Broadwin Aug 2006
Data Analysis
• Time-series and case-crossover methods
–Basu et al. 2005
• Separate analyses by county • County estimates combined in meta-analysis • Parallel study by Harvard group of 9 non-CA counties
Time-series Study Design
• Often used for air pollution studies • Examine association between daily apparent temperature and daily mortality counts • Adjust for all other factors that change over time
Case-crossover Study Design
• Compare temperature on day of death (case) to temperature on different days for same person when death did not occur (control) • Choose control periods within the same month as the cases
–Addresses concerns about effects of seasonality and other time-varying factors
Preliminary Results
Apparent Temperature per 10oF and Mortality
6
Case-crossover Time-series
Percent Change (95% CI)
4
2
0
-2
ALL-CAUSES
CARDIOVASCULAR RESPIRATORY
Apparent Temperature per 10oF and Disease-specific Mortality
20
Percent Change (95% CI)
10
5.4 2.5
0
2.7
2.7
1.2
-10 CHF IHD MI DIABETES CEREBRO
Apparent Temperature per 10oF and All-cause Mortality by Race
8
6
Percent Change (95% CI)
4.9
4
2
2.5 1.8
0
WHITE
BLACK
HISPANIC
Apparent Temperature per 10oF and All-cause Mortality by Age Group
12 10
Percent Change (95% CI)
8
6
4
4.2 2.2 2.6 1.7
2
0
-2
<=5 yrs
>=65 yrs
>=75 yrs
>=85 yrs
Apparent Temperature per 10oF and Allcause Mortality Adjusted by Pollutant
4
Percent Change (95% CI)
3
2.6
2
2.8
2.9 2.3 2.0
1
0
Apptemp only
Ozone
PM2.5
CO
NO2
Temperature per 10oF and All-cause Mortality
10
8
Percent Change (95% CI)
Case-crossover Time-series
6
5.3 4.7
4
3.7 2.6
2
0
NON-CA
CA
Future Projections
• Projected temperature increases for one year (2034) from Cayan et al. 2006 • Examined medium sensitive model (GFDL/A2) • Used CA and non-CA estimates of temperaturemortality association from our study • Assumed baseline conditions for population size and demographic distribution
Estimated Excess Mortality for the Year 2034
1400 1200 1000 800 600 400 200 0
CA Estimate 477 783
Non-CA Estimate 734 1236
9 counties All of CA
Projected increase in warm season mortality associated with the rise in temperatures: 0.98-1.55%
Implications
• ~3% increase in all-cause mortality associated with 10°F increase apparent temperature • Increased risk also found for cardiovascular mortality, elderly, young children, Blacks • Mortality effect of apparent temperature is immediate • Temperature effect appears independent of air pollutants • Case-crossover and time-series estimates similar • Heat wave not necessary to find a temperaturemortality association in CA
Future Research
• Other temperature definitions (min/max temp) • Further analysis of July 2006 heat wave • Additional analyses of vulnerable subgroups and interaction with air pollutants • Morbidity studies to include hospitalizations
Acknowledgements
OEHHA Bart Ostro Wen-Ying Feng Rachel Broadwin Brian Malig Lindsey Roth Janice Kim Shelley Green Harvard Antonella Zanobetti Joel Schwartz