Short-term effects of ambient air pollutionon public health in Hong Kong - an APHEA-2 study

Short-term effects of ambient air pollution
on public health in Hong Kong - an APHEA-2 study

A consultancy report submitted to the Environmental Protection Department, Hong Kong

May 1998

Image of The University of Hong Kong

Department of Community Medicine The University of Hong Kong

Department of Community Medicine
The University of Hong Kong
Environmental Health and
Biostatistics and Computing Groups

Project Team

Dr CM Wong
(Data analysis and report writing)

Mr Stefan Ma
(Computation and statistical advice)

Professor AJ Hedley
(Head of Department)
Professor TH Lam
(Epidemiological advice)

(Acknowledgement: Ms Marie Chi for secretarial support)




Executive Summary

1.0 Background and introduction
2.0 Scope and objectives

Materials and methods

3.1 Study design
3.2 Databases
3.3 Statistical modelling


4.1 Descriptive statistics


Statistical modelling

4.2.1 Core models for hospital admissions
4.2.2 Core models for hospital deaths
4.2.3 Models for pollutants on hospital admissions
4.2.4 Models for pollutants on hospital deaths
4.2.5 Effects of pollutants adjusted for copollutants
4.2.6 Models with adjustment for autocorrelation
4.2.7 Non-linear effects
4.2.8 Models for interactions between pollutants
4.2.9 Models with interaction between pollutants and seasons


Validity and reliability of models

4.3.1 Predictive validity of 1995-97 models
4.3.2 Comparison of 1995-97 with 1995-96 models



5.1 Validity and reliability of the models
5.2 Comparison with APHEA studies
5.3 Limitations of the Hong Kong study
5.4 Significance of study results
6.0 Conclusions



A. Figures

B. Basic Tables


Akaike's Information Criterion (AIC): Akaike's Information Criterion is an index for assessing the fitness (as reflected by small deviation from the observed or large likelihood) of a model, taking into account the number of parameters in the model.

APHEA: Air Pollution on Health: a European Approach. This is a research organization formed by researchers from several European countries, to study the short term effects of air pollution on health according to guidelines from an established protocol. The protocol has been adopted by researchers in Hong Kong.

Collinearity: An effect due to some strong relationship between pollutants and/or other covariates when studying their effect on an outcome such as hospital admissions and deaths.

Covariates: Variables which are possibly predictive of the outcome under study. A covariate may be of direct interest to the study, a confounding variable, or an effect modifier.

95% confidence interval: An estimate in terms of an interval which has 95% of the chance to include the true value.

Confounding: A situation in which the effects of two processes are not separated. The distortion of the apparent effect of an exposure on risk brought about by the association with other factors that can influence the outcome.

Co-pollutant: A pollutant which coexists with the main pollutant under study.

Estimate: A measurement or a statement about the value of some quantity is said to be an estimate if it is known, believed, or suspected to incorporate some degree of error.

Exposure: Proximity to and/or contact with a source of a health hazard (e.g. air pollutants) in such a manner that effective transmission of the agent or harmful effects of the agent may occur.

FSP: Fine suspended particulates.

International Classification of Diseases (ICD): The classification of specific of conditions determined by an internationally representative group of experts who advise the World Health Organisation. The different groups of diseases are subdivided into different sections or rubrics with unique code numbers.

IPAS: Integrated Patient Administration System. It was established by the Hospital Authority for management and retrieval of hospital patient records. It was being gradually phased out and replaced by the Medical Record Abstracting System (MRAS).

Lag effect: Effect due to pollutants and covariates which have occurred in previous days.

Linear relationship: A relationship between exposure levels and a health outcome (or a transformation of it) which can be represented by a straight line.

ug/m3: Unit for measuring air pollutant concentration in micro grams per cubic metre.

um: Micrometre or micron; one millionth of a metre.

MRAS: Medical Record Abstracting System. This system was developed to replace the IPAS system for management and retrieval of hospital patient records.

p-value: Probability value; the probability that a test statistic would be as extreme as, or more extreme, than that observed, if the null hypothesis were true.

Poisson regression: A regression method which models counts of events, which occurred at some point in time, as the outcome variable and relates them (in this study) to some pollutants and covariates.

Relative risk: The ratio of the risk of disease or death among the exposed to the risk among the unexposed.

Risk: The probability that an event (e.g. admission to hospital or death) will occur.

RSP: Respirable suspended particulates.

Time-series analysis: An approach to estimate acute health effects of air pollution by statistical modelling of daily counts of a health outcome on daily average air pollutant concentrations. The analysis is usually done on a series of several years of data including hospital admissions, air pollutant concentrations, meteorological conditions, holidays, day after holidays and other cyclical factors.

Wave length: The duration for each seasonal cycle in the health outcome under study.

Back to Content


Back to topdot_clear.gifTable of ContentNext
User defined date2: 
Friday, 28 April, 2006