Abstract
Background: In the last two decades, Hong Kong experienced reduced visibility, which is primarily attributed to air pollutants such as respirable suspended particulates (PM10) and nitrogen dioxide (NO2). These air pollutants are associated with adverse health effects including increased doctor consultations, hospital admissions, and premature deaths.
Objective: To assess the short-term effects of variation in visibility on daily mortality in Hong Kong.
Methods: Daily 24-hr visibility data was recorded as visual range at an urban observatory and the airport. Meteorological data and daily all natural mortality counts for 1996-2002 were obtained from the Hong Kong Observatory and Census and Statistics Department. Poisson regression with non-parametric smoothing was used to obtain a core model of daily mortality counts with adjustments for trend, seasonality, meteorological variables, days of the week and influenza epidemics. The visibility was then entered as a linear term.
Results: Visibility showed a strong inverse correlation with PM10 (r = -0.72) and NO2 (r = -0.59). The mean visibility for the period was 13.37km. We found an excess risk per 10 km change in visibility (ER) of 1.57% (95% CI 0.55%-2.59%) for all natural causes, which translated to about 483 avoidable deaths annually. Sensitivity analysis using non-parametric approach for 2000-2002 showed an ER of 3.57% (0.6%-7.1%) for all natural causes, which translated to about 1,100 avoidable deaths annually. This number was comparable to the estimated 1,400 avoidable deaths in an analysis using parametric approach for the three-year period. Since non-parametric approach generally leads to lower ER and standard errors than parametric approach, the sensitivity analysis validated our results.
Conclusion: A measure of visibility in this setting provided a reasonable proxy for the assessment of environmental health risks from ambient criteria air pollutants. This may provide a valid approach in areas where pollutant estimates are absent or unreliable.
Presentation: http://www.cleanairnet.org/baq2006/1757/docs/SP5_3.ppt
Visibility, mortality, Poisson regression, Hong Kong |