Effects of extreme weather on human health: methodology review
Abstract
This work critically evaluates current methodology applied to estimate the effects of extreme weather events (EWE) on human health. Specifically, we focus on uncertainties associated with: a) the main statistical approaches for estimating the effects of EWE, b) definitions of health outcomes and EWE, and c) possible sources of errors and biases in currently available data sets. The EWE, which include heat waves, cold spells, ice storms, flood, drought and tornadoes, are known for their massive effects on ecosystems, economies, infrastructures. In particular, human lives and health are frequently impacted by EWE; however, the estimate of such effects is complex and lacks a systematic methodology. An accurate and reliable estimate of health impacts is critical for developing preparedness and effective prevention strategies, better allocating scarce resources for mitigating negative impacts of EWE, and detecting vulnerable populations and regions in a timely manner. We reviewed 82 manuscripts published between 1993 and 2011, selected from MedPub and Medline databases using predetermined sets of keywords, such as extreme weather, mortality, morbidity and hospitalization. We classified publications based on their geographical locations, types of included health outcomes, methods for detecting EWE and statistical methodology employed to determine the presence and magnitude of EWE associated health outcomes. We determined that 57% of the reviewed manuscripts applied time-series analysis and the associations analysis and were conducted in temperate regions of the US, Canada, Korea, Japan and Europe respectively. About 60% of reviewed studies focused primarily on mortality data, 30% on morbidity outcomes and 9% studied both mortality and morbidity with respect to direct effects of extreme heat waves and cold spells. A wide range of EWE definitions were employed in those manuscripts, which limited the ability to compare the results to a certain degree. We observed at least three main sources of uncertainty, which may lead to an estimate bias: potential misrepresentation and misspecification of the biological causal mechanism in statistical models, completeness and quality of reporting EWE-specific health outcomes, and incomplete accounting for spatial uncertainties in historical environmental records. Finally we show that some of those systematic biases can be reduced by performing proper adjustments, while some of them still need further studies and efforts. Reducing bias provides more accurate representation of disease burden. Better understanding of EWE and their impacts on human health, combined with other preventive strategies, can provide better protection from EWE for vulnerable populations in the future.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2012
- Bibcode:
- 2012AGUFMED33A0741W
- Keywords:
-
- 1616 GLOBAL CHANGE / Climate variability;
- 1630 GLOBAL CHANGE / Impacts of global change;
- 3275 MATHEMATICAL GEOPHYSICS / Uncertainty quantification