Climate change and human health in search of magic numbers… NCAR Summer colloquium 28 July 2004 R Sari Kovats Centre on Global Change and Health Dept of Public Health and Policy London School of Hygiene and Tropical Medicine
STRATOSPHERIC OZONE DEPLETION -Global problem -Health and environmental impacts -Skin cancer -Cataracts Information from epidemiological studies
Impact models Estimates of populations at risk or attributable burden of disease Greenhouse gas emissions scenarios Defined by IPCC Global climate scenarios: Generates series of maps of predicted future distribution of climate variables 30 year averages Modelling impacts of climate change 2020s 2050s 2080s 2020s2050s2080s
High child, high adult High child, very high adult M F Both MFMF (000) Addictive substances Tobacco Alcohol Illicit drugs Environmental risks Unsafe water, sanitation hygiene Urban air pollution Indoor smoke from solid fuels Lead exposure Climate change Occupational risks Risk factors for injury Carcinogens Airborne particulates Ergonomic stressors Noise WorldAfrica Deaths, 2000
Deaths (thousands)DALYs (millions) Estimated death and DALYs attributable to climate change. Selected conditions in developing countries Floods Malaria Diarrhoea Malnutrition
Health-impact models Process-based/Biological models –Malaria/vectorial capacity [MIASMA] –Heat budget models Empirical statistical –Temp-mortality (Kalkstein, Moser, etc.) –Temp –Diarrhoeal disease –Rainfall -flood-death –Temp/rainfall- Dengue, Malaria [spatial correlations]
TRANSMISSION POTENTIAL Temperature (°C) Incubation period (days) Biting frequency Temp (°C) (per day) Survival probability (per day) Temp (°C) Martens et al. 1999, van Lieshout et al. 2004
Can global models reveal regional vulnerability? Increase: East Africa, central Asia, Russian Federation Decrease: central America, Amazon [within current vector limits] A1 B2 A2 B1
Potential distribution of Aedes aegypti in the North Island based on 10°C midwinter isotherm limit for a mid- and high-range climate change scenario. Source: Hotspots dengue fever risk model developed by the International Global Change Institute, University of Waikato, with the assistance of funding from the Health Research Council Present Present Mid-range scenario (SRES B2 greenhouse gas emission scenario, best guess climate sensitivity) High-range scenario (SRES A2 greenhouse gas emission scenario, high climate sensitivity)
Empirical-stats models EXTRAPOLATION –Can you extrapolate the exposure-response relationship beyond the bounds of the observed temperature range? VARIATION –Can you extrapolate the exposure-response relationship derived from a different population. ADAPTATION –Responses to climate change - acclimatization MODIFICATION –What is likely?– –changes to exposure response relationship
Predicted distribution of the malaria vector (mosquito Anopheles atroparvus) in present day Europe, and in the 2080s with SRES A2 climate scenario. [Kuhn, LSHTM, 2002] Current climate2080s
Temperature-salmonellosis [fully adjusted models] England & Wales Scotland Switzerland Netherlands
Netherlands: time series Total weekly cases
Climate change and air pollution, UK Health Assessment 2002 Pollutant2020s2050s2080s ParticlesLarge decrease Ozone (assuming no threshold) Large increase (by about 10%) Large increase (by about 20%) Large increase (by about 40%) Ozone (assuming a threshold) Small increase Nitrogen dioxideSmall decrease Sulphur dioxideLarge decrease
Outcomes... Shift in “climate envelope” Additional population at risk –Definitions of risk Relative risk Absolute risk –additional/excess cases/deaths –Disability-adjusted life-year [DALY] COSTS
Simplified causal web linking exposures and outcomes WHO model
Attributable fractions vs attributable deaths/cases Population change –Growth –Ageing –Countries have national projections Which baseline disease incidence used to estimate attributable cases. –Current or future?
Scenarios Climate –Averages, extremes Population –Population growth ✔✔ –Population ageing ✔ –Urbanisation, coastal migration “socio-economic”
Non climate scenarios Vector presence/abundance Baseline disease prevalence –Cardiovascular disease –HIV/AIDS Millennium Development Goals Population Income/GDP per capita/PPP per capita Technology –Malaria vaccine Qualitative “Knowledge is King, Big is Beautiful”
Relevance of attributable vs avoidable burden Avoidable burden more policy-relevant Why calculate attributable burden?
WHO Definitions… A health impact assessment is a combination of procedures or methods by which a proposed policy, programme or project may be judged as to the effects it may have on the health of a population. The basic principles underlying such an assessment are democracy, equity, sustainable development and evidence-based advice.
Uncertainty Climate scenario –>1 climate model –>4 emissions scenarios –Regional model –Downscaling Exposure response relationship –Key uncertainties/assumptions in the models –Confidence intervals –Monte Carlo simulation/Bayes
Qualitative Low High Established but incomplete Speculative Competing explanations Well-established Amount of evidence Level of agreement, consensus
Past[climate/weather-healthrelationships]Future [map malaria] Present [highland malaria] learn?analoguesmechanismsdetectionattribution predictive modelling three research tasks Empirical studies [epidemiology]
CountryReference Antigua and BarbudaO'Marde and Michael, 2000 – UNEP Country Study AustraliaMcMichael et al, 2002 CameroonUNEP/ Ministry of Environment and Forestry, Cameroon, 1998 CanadaDuncan et al., 1997 Fijide Wet and Hales, 2000 JapanAndo et al, 1998 KiribatiTaeuea, de Wet and Hales, 2000 New ZealandWoodward et al PanamaSempris E and Lopez R, eds ANAM/UNDP PortugalCasimiro and Calheiros, 2002 South AfricaUNEP Country study 2000 Sri LankaRatnasari 1998 St LuciaSt Lucia National Communication, chapter 4. United KingdomDept of Health, 2002 United StatesPatz et al., various documents ZambiaPhiri amd Msiska, 1998