engleski

Weekly all-cause mortality and outdoor NO_2 in Zagreb

Introduction. Many recent studies analyze the short-term effects of air pollution on human health (morbidity and mortality). In some cases association between the concentration of air pollutants and number of hospital emergency room visits or deaths was found even for concentrations below the limit and/or recommended values. In European Union there is a collaboration group APHEA, which works on the assesment of the short-term effects of air pollution on mortality and morbidity using epidemiologic time-series data. Our results show that there is an association between the weekly number of emergency room visits due to asthma in children and adults and outdoor NO_2 concentrations. In this paper we have used the same methodology aiming to assess the possible association between the number of deaths due to all causes and outdoor NO_2 concentration. Methods. Data on the number of deaths in Zagreb for 1994 and 1995 were provided by the Croatian Central Bureau of Statistics. The 24-hour avarage levels of NO_2 were measured by the staff of the Department of Environmental Hygiene, Institute for Medical Research and Occupational Health, Zagreb. We had measurements for the period from July 1994 to December 1995. This time series had 11 missing values. Information on average temperature, air pressure and relative humidity was collected by the Hydrometeorological Institute of Croatia. We have used the weekly averages, to enable comparison with previous results. This resulted in time series with 78 data points. Average weekly concentrations of NO_2 for weeks with missing values were estimated using the avaliable data. Except for the first and the last week, we had 5 weeks with 6, and one week with 5 available values. Time series of the number of deaths approximately followed the Poisson distribution and was overdispersed and positively autocorrelated. We have used logarithmic transformation on the count data. In order to account for the observed and some of the unobserved confounding variables, we have adjusted both NO_2 and death count series with respect to trend, seasonal variation and weather variables. Effects of trend and seasonal variation were estimated using the locally weighted regression (LOESS) with a bandwith of 12 weeks. For number of deaths complete series (104 data points) was used for trend and seasonality assessment. Association between the residuals of NO2 and the number of deaths was analyzed using autoregressive linear regression. All analyses were performed on a PC 486 using SAS 6.12 for Windows. Results. Average weekly number of deaths in Zagreb in 1994 and 1995 was 152.2 (SD 17.3, Var 300.8) with skewness 0.64 and kurtosis 0.41. It ranged from 119 to 204. Concentration of NO_2 was within the limit (Cavg=50 micro g/m^3, C98=120 micro g/m^3), but above the recommended daily value (Cavg=30 micro g/m^3, C98=60 micro g/m^3 ; our daily: mean 45, SD 18.4, range 5.5 to 169.8 ; weekly: mean 45, SD 9.1, range 24.5 to 64.9). Autoregressive linear regression revealed no autocorrelation of residuals, and the coefficient for weekly NO_2 concentration was 0.002769 (SE 0.00134, p=0.0423). This coefficient corresponds to a relative risk associated with the increase of 10 micro g/m^3 in average weekly NO_2 concentration equal to 1.028 (95% CL 1.001 to 1.056). Discussion. We have found a weak association between weekly NO_2 concentration and the number of deaths in Zagreb. The associated relative risk was lower than that found for emergency room cases of hyperreactive lung diseases in children and adults in the same period in Zagreb (asthma in children RR=1.137, CL 1.005 to 1.287, in adults RR=1.098, CL 1.002 to 1.203 ; all respiratory causes in children RR=1.067, CL 1.002 to 1.136, in adults RR=1.019, CL 0.990 to 1.049). Stronger relationship may be found for cause specific mortality.

air pollution; mortality; time series

nije evidentirano