Marine-shipping related exposures and mortality in Hong Kong : a time-series study

Abstract

Ship emission has been one of the most crucial components of ambient air pollution in recent years because of its various environmental health effects on habitants of port regions. However, attentions were mainly paid to ocean-going vessels, while the health impact due to river vessels has rarely been assessed and in particular, there has not been any study quantifying the relationship between mortality and vessel volume including river vessels.

This study aimed to assess the temporal association between marine-shipping exposure and monthly mortality in Hong Kong from 1999 to 2011.

I received 491,824 death records from the Census and Statistical Department in defining the health outcomes, and monthly data for Net Registered Tonnage (NRT) of ocean-going vessels (OGVs) and river vessels (RVs) from the Marine Department for the exposures for the period from 1999 to 2011. The NRT for OGVs, RVs, and their total, namely Total Vessels (TVs), were employed to assess the temporal marine-shipping exposure indicators and their association with mortality. Poisson regression model with natural cubic spline smooth function was introduced to control for seasonal confounding variations and time trend in the mortality counts, and with adjustment for temperature and relative humidity. Partial autocorrelation plots were applied to examine discernible patterns in residual plots. The analysis also included stratification by sex, age group, and interaction model for cool and warm season. Excess risks from the model were assessed.

For the associations between NRT of OGVs and monthly mortality for all-natural, cardiovascular and respiratory causes for all-age, in the whole-year analysis, an increase of 1000 NRT of ocean-going vessels was associated with an increase of 1.00% [95% CI: 0.79 to 1.21%], 1.35% [0.95 to 1.76%] and 1.10% [0.62 to 1.58%], respectively. In the cool season, the estimates were 1.00% [0.76 to 1.19%], 1.30% [0.89 to 1.70%] and 1.00% [0.51 to 1.49%]; while in the warm season the estimates were 1.19% [-0.59 to 3.00%], 1.70% [-1.04 to 4.52%] and 1.49% [-0.14 to 3.14%], respectively. For the associations between the capacity of RVs and monthly mortality for all ages, in the whole-year, the estimates were statistically significant (with the 95% CI > 0) only for all-natural causes and cardiovascular, but not for respiratory diseases (with the 95% CI containing zero).Particularly for analyses of interaction model for cool season the estimates were consistent with those of the whole-year analyses. The associations between the capacity of TVs and monthly mortalities were similar with those for OGVs. For the three marine-shipping exposure indicators, the estimates for the older than 65 population were greater than for the all-age population, and greater for females than for males.

It is evident that there is an association between marine-shipping exposure and mortality, which is stronger for the older than the all-age population. The associations were robust in cool season for all the mortality outcomes. It is the first study worldwide to explore the associations between health outcomes and monthly capacity of marine vessels. However, the study design was based on aggregated data, making it less powerful to assess the association than designs based on individual subjects. In the future, we should use study designs with recruitment of individuals and estimate their exposure by means of spatial and temporal variations of marine emissions in the real-time using GIS technology.