A time-series analysis of mortality and air temperature in Greater Beirut

https://doi.org/10.1016/j.scitotenv.2004.02.027Get rights and content

Abstract

The literature on the association between health and weather in the temperate to semi-arid cities of the Eastern Mediterranean is scarce. The quantification of the relationship between temperature and daily mortality can be useful for developing policy interventions such as heat-warning systems. A time-series analysis of total daily mortality and weather data for the city of Beirut was carried out. The study covered the period between 1997 and 1999. Poisson auto-regressive models were constructed, with mean daily temperature and mean daily humidity as explanatory variables. Delayed effects, up to 2 weeks, were accounted for. The regression models were used next to assess the effect of an average increase in temperature on yearly mortality. The association between temperature and mortality was found to be significant. A relatively high minimum-mortality temperature (TMM) of 27.5 °C was calculated. A 1 °C rise in temperature yielded a 12.3% increase (95% confidence interval: 5.7–19.4%) and 2.9% decrease (95% confidence interval: 2–3.7%) in mortality, above and below TMM, respectively. Lag temperature variables were found to be significant below TMM but not above it. Where the temperature change was less than 0.5 °C, annual above-TMM losses were offset by below-TMM gains, within a 95% confidence interval. TMM for Beirut fell within the range usually associated with warm climates. However, the mild below-TMM and steep above-TMM slopes were more typical of cities with temperate to cold climates. Our findings suggest that heat-related mortality at moderately high temperatures can be a significant public health issue in countries with warm climates. Moreover, at the projected climate change over the next 50 years, heat-related losses are unlikely to be offset by cold-related gains.

Introduction

A strong association between temperature and total daily mortality has been identified for a number of cities around the world. Deaths that are directly related to heat, such as hyperthermia, can occur as a result of cardiovascular, respiratory or cerebro-vascular malfunctions (Kalkstein, 1993, Martens, 1998). More generally, the number of deaths from circulatory and respiratory causes has been found to be strongly associated with weather conditions. While the effects of heat waves on daily mortality have been a subject of study for some time, researchers have turned their attention more recently to the impacts of smaller variations in temperature (Kunst et al., 1993, Saez et al., 1995, Ballester et al., 1997, Curriero et al., 2002).

The so-called V-shaped (Martens, 1998) and J-shaped (McMichaels et al., 1996) curves refer to a pattern in which daily mortality decreases, as mean temperature increases from its winter lows, to reach a mortality minimum. Mortality then rises, sometimes quite abruptly, beyond a certain threshold of mean summer temperatures. The minimum-mortality temperature, TMM, could be as low as 16 °C in the Netherlands (Kunst et al., 1993) and as high as 27.2 °C in Miami (Curriero et al., 2002) and 29 °C for cerebral infarction mortality in Taiwan (Wen-Harn et al., 1995). For a more complete review of the literature on temperature–mortality relationships, the reader is referred to Martens (1998).

The literature on the association between mortality and temperature in Lebanon, and more generally, the Middle-east and North Africa is scarce. Khojali (1983) noted the high prevalence of heat-related illness during the Mecca pilgrimage in Saudi Arabia. Green et al. (1994) used monthly death counts in Israel to study the seasonality of ischaemic heart disease and found strong evidence of a causal association with temperature. Douglas et al. (1991), analyzing monthly death counts in Kuwait, concluded that socio-economic progress had reduced the seasonality of infant and total mortality. No analysis of the relationship between temperature and total daily mortality in the temperate to semi-arid cities of the eastern coast of the Mediterranean appears to have been carried out. Only five countries in this region, with no more than 16% of the total population, supplied any mortality data to the World Health Organization between 1985 and 1990 (United States Bureau of the Census, 1999).

This paper presents the results of a time-series analysis of daily death counts for the city of Beirut, which aims to identify any significant associations with temperature and humidity, including lag effects. In addition, the constructed regression models are used to evaluate the likely impact on annual total mortality of a future average increase in temperature.

Three policy-relevant questions can be better addressed as a result of this research. The degree of importance of heat-related deaths as a public health issue, in comparison with other health problems encountered in the region, remains to be assessed. Second, if data suggest a strong association between same-day temperature and mortality and show little evidence of compensatory effects, policy measures partly based on daily temperature indicators, such as heat-warning systems, would be more likely to succeed in reducing mortality. Finally, the analysis of the relationship across relevant age categories allows for a more targeted prevention policy.

Section snippets

Materials and methods

Beirut is situated on the eastern coast of the Mediterranean, at 33.5° latitude and 35.5° longitude. Its climate is temperate to semi-arid with hot and humid summers and mild winters. It is adjacent to a high mountainous range, the Mount Lebanon, which experiences cold winters and heavy snowfall at its peaks. Most of the city's infrastructure, destroyed by the 1975–1990 civil war, has been rebuilt. Although vital statistics have been resumed at the end of the war, Lebanon suffers from a lack of

Results

Table 1 shows overall statistics of daily mortality and weather parameters. Mean daily mortality during cold months (November–April) was found to be significantly higher than in warm months (May–October) at P<0.001. Mean daily mortality was higher on Mondays and Tuesdays, in comparison with the rest of the week (P<0.1), and higher in 1998 than 1997 (P<0.05) and 1999 (P<0.001).

The graphic determination of TMM yielded a value of 27.5 °C (Fig. 1). An identical TMM value for the over-65 was

Mortality–temperature slopes

The relationship between total daily mortality and mean daily temperature for Beirut follows the V shape observed in many other cities around the world. However, while the value of TMM (27.5 °C) falls within the range usually associated with warm climates, the relatively mild below-TMM slope and the steep above-TMM slope are akin to a J shape that is more typical of cold climates. Table 4 shows a comparison with results for selected European and North American cities.

A number of authors have

Conclusions

The scope of this study is limited by the lack of cause-specific death counts, the inability to correct for influenza and the short time span of the dataset. However, the analysis has shown that steep mortality–temperature slopes may be obtained, even when the minimum-mortality temperature is relatively high. Given the weak lag effects in summer, our study shows that heat-related mortality at moderately high temperatures can be a significant public health issue in countries with warm climates.

Acknowledgements

The study was funded by a grant from the University Research Board of the American University of Beirut. The authors are grateful for the valuable input of Dr Abla Sibai. Helpful advice on statistical analysis was received from Dr Marwan Khawaja, Dr Theo Stinjen, Dr Hala Tamim and Ms Eza Al-Zein. Dr. Iman Nuwayhid kindly read the manuscript and provided his feedback. The acquisition of data was greatly facilitated by Dr Abdo Bajjani of the Department of Weather Forecasting, Dr Marwan Houri from

References (24)

  • Intergovernmental Panel on Climate Change. Climate Change 2001. Impacts, adaptation and vulnerability. Contribution of...
  • M. Khojali

    Epidemiology of heat illness during the Makkah pilgrimage in Saudi Arabia

    Int J Epidemiol

    (1983)
  • Cited by (0)

    View full text