Winter hospital admissions and weather types in the Auckland Region

Recent studies suggest that extreme climatic conditions (Kalkstein and Valmont, 1987), as well as poor air quality (Goodman et al., 2004) impact significantly on human health. Mortality and the rate of admissions to hospital are often used as measures of 'health impact' with 'temperature' (either the daily maximum or minimum) being the most commonly-used climate variable (Braga and Schwartz, 2002; Donaldson and Keatinge, 1997). International studies show that the effects on health of cold events during winter are generally more pronounced in regions of temperate climates, probably because of the tendency for poorer house insulation and home heating compared to regions that experience very cold winters (The Eurowinter Group, 1997), as exemplified by an early New Zealand study reporting a 35% increase in winter mortality compared to summer (Marshall et al., 1988). All of the studies carried out to date in New Zealand have used individual climate variables when considering the impact of climate on health. One study looked at climate and mortality in Auckland and found significant negative associations between total mortality and respiratory and coronary heart disease mortality in relation to temperature (Cockburn and Salinger, 2001). The study was subsequently extended to consider hospital admissions rather than mortality and found a 40% rise in admissions in winter relative to summer (Gosai and Salinger, 2007). Although many studies in the literature suggest that the elderly are the most affected (Schwartz et al., 2004), this Auckland study found that infants are also disproportionately affected by climate extremes (Schwartz et al., 2004). This is supported by a more detailed study showing that for admissions due to respiratory infections and inflammation, there is a dramatic increase in admissions for the very young (<1 year and 1-4 years), and for whooping cough and acute bronchitis for the <1 year age group on both a seasonal basis and on a daily basis (Gosai and Salinger, 2007). Maori and Pacific Island populations were found to be the most sensitive groups, suggesting that socioeconomic and/or cultural factors influence the impact of climate on health (Gosai and Salinger, 2007 Gosai et al., 2008), as supported by * Corresponding author address: Marina Baldi, IBIMET-CNR, Via Taurini 19, Roma, Italy; email: m.baldi@ibimet.cnr.it other studies carried out overseas (Schwartz et al., 2004). Also, the very young are more affected by climate in Auckland than suggested by the current international literature (Wilkinson et al., 2001). Studies comparing mortality and/or hospital admission rates with individual climate variables (such as all of those carried out in Auckland so far) are limited in some sense because climate variables are generally not independent of each other, so uncertainty remains as to whether it is the temperature (or moisture) per say that is the factor, for example, or whether it is a combination of factors, including other variables such as the ambient air pollution levels. To overcome this problem, many more recent studies in the international literature have based their assessment on a 'weather type' classification system (where the weather is classified based on the synoptic-scale pressure and wind patterns) or an 'air mass' classification scheme (de Pablo et al., 2008; McGregor et al., 1999; Morabito et al., 2006) where the weather conditions are classified based on patterns of surface measurements of the climate variables. A suitable weather classification scheme has been developed for New Zealand (known as the Kidson Weather Types) (Kidson, 2000) which allows such a study to be carried out in Auckland. This paper investigates the relationships between weather types (based on the Kidson Weather Types), air pol