Visibility in Hong Kong has deteriorated significantly over 40 years with visibility below 8km in the absence of fog, mist, or precipitation, increasing from 6.6 days in 1968 to 54.1 days in 2007. We assessed the short-term mortality effects of daily loss of visibility. During 1996-2006, we obtained mortality data for non-accidental and cardiorespiratory causes, visibility recorded as visual range in kilometers, temperature, and relative humidity from an urban observatory, and concentrations of four criteria pollutants. A generalized additive Poisson regression model with penalized cubic regression splines was fitted to control for time variant covariates. For non-accidental mortality, an interquartile range (IQR) of 6.5km decrease in visibility at lag0-1 days was associated with an excess risk (ER%) [95% CI] of 1.13 [0.49, 1.76] for all ages and 1.37 [0.65, 2.09] for ages 65 years and over; for cardiovascular mortality of 1.31 [0.13, 2.49] for all ages, and 1.72 [0.44, 3.00] for ages 65 years and over; and for respiratory mortality of 1.92 [0.49, 3.35] for all ages and 1.76 [0.28, 3.25] for ages 65 years and over. The estimated ER% for daily mortality derived from both visibility and air pollutant data were comparable in terms of magnitude, lag pattern, and exposure-response relationships especially when using particulate matter with aerodynamic diameter < or = 10 microm to predict the mortality associated with visibility. Visibility provides a useful proxy for the assessment of environmental health risks from ambient air pollutants and a valid approach for the assessment of the public health impacts of air pollution and the benefits of air quality improvement measures in developing countries where pollutant monitoring data are scarce.
Copyright 2010 Elsevier Inc. All rights reserved.