Traffic-related exposures and all-cause and cause-specific mortality of general and older population in Hong Kong

Abstract

Background Epidemiological studies have shown that air pollution was associated with both mortality and morbidity of various diseases including cardiovascular diseases, respiratory diseases and cancers.

However, the various traffic-related exposure indicators are difficult to compare due to the diversity of study areas, populations, measures of traffic exposures and confounders. Moreover, most of the studies were conducted in the western and European countries. Few studies using the traffic density as surrogate of traffic-related exposure for effects on mortality risk have been performed in Asia and none has been performed in Hong Kong.

Objective This study aims to assess the association between the traffic-related exposure and the all-cause and cause-specific mortality risk in the densely populated city of Hong Kong, where traffic emission plays an important role on the ambient air quality.

Methods Three traffic-related indicators were employed including: Road Density (RD) in terms of total length of roads divided by Tertiary Planning Unit (TPU) area; Traffic Density (TD) defined by road lengths times the Annual Average Daily Traffic (AADT) and divided by the TPU area; and Vehicle Density (VD) means average AADT over TPU area. Each exposure was divided into three groups according to tertiles. An ecological study was conducted first with population census and mortality data. The age-sex standardized all-cause and cause-specific mortality rate was calculated for each TPU with the whole Hong Kong death rate in 2010 as the standard. Poisson regression models were performed to estimate risks of traffic-related exposure with adjustment of the marital status, race, education, housing tenure, and median household income. After the TPU-level analyses, a cohort with 64,888 elderly subjects being followed up from 1998 to 2012 was used to assess the association with further control for the individual-level factors including age, sex, education, income, housing, and smoking. The Multilevel Cox proportional regression models were built with adjustment for both the individual level confounders and TPU-level covariates. Excess risks from both models were reported.

Results Higher exposure areas were found in the northern part of Hong Kong Islands and the inner city of Kowloon peninsula. Statistically significant association between traffic-related exposure and mortality was observed. For the investigation in the general population, the all-nonaccidental cause mortality was associated with 43% (95% confidence interval 37-48%) and 50% (44-56%) excess risk for areas with the middle and high level TD exposure compared with the low level group. The association was similar with measures of RD and VD. For the cause-specific mortality, the respiratory deaths showed a higher risk when compared with the cardiovascular and cancer deaths. For the elderly subjects, the excess risk of all-nonaccidental causes relative to the low level exposure of 13% (1-26%) and 12% (0-25%) for the middle and high level exposure were smaller when compared with the risk in the general population.

Conclusion There is an association between traffic-related exposure and mortality in the general and older population of Hong Kong. In future comprehensive investigations with the individual-level exposure measure are needed. Assessment on the younger population should also be studied.