Timeseries studies for daily

1
HEI Public Health and Air Pollution in Asia
(PAPA) Project
Time-series studies for daily mortality and
hospital admissions Interaction between air
pollution and respiratory viruses (Hong Kong
Study)
CM Wong1 , JSM Peiris2, TQ Thach1, PYK Chau1 KP
Chan1 , GN Thomas1, TW Wong3,TH Lam1, AJ
Hedley1, 1 Department of Community Medicine and
2 Department of Microbiology, The University of
Hong Kong 3 Department of Family Community
Medicine, The Chinese University of Hong Kong
2
Background
Hong Kong

• Located in a sub-tropical region with average
  temperature 24oC (range 11-34) and relative
  humidity 78 (range 45-95)
• Population 6.8 million in 1092 km2 in urban and
  rural areas
• Primary care 85 private practitioners
• Hospital care 94 in public hospitals
• Computerized mortality and hospital information
  systems

3
The problems of air pollution arise mainly from
burning fossil fuels.
4
In the past 15 years NO2 and O3 were on the
increase while PM10 was relatively stable, and
SO2 levels remained low after implementation of
sulphur restriction regulations in 1990 and in
subsequent years.
5
The University of Hong Kong Department of
Community Medicine since 1989 has carried out a
series of air population and health studies,
including

• Cross-sectional studies on children respiratory
  symptoms
• Bronchial responsiveness in children exposed to
  atmospheric pollution
• Respiratory health of primary school children
• Health benefit of the 1990 sulphur restriction
  intervention
• Cardiorespiratory and all-cause mortality after
  the restrictions
• Comparison between two districts of the effects
  of an air pollution intervention on bronchial
  responsiveness in primary school children
• Effects of an ambient air pollution intervention
  and environmental tobacco smoke on children's
  respiratory health
• Daily time-series studies on mortality and
  morbidity
• Effects of air pollution on hospital admissions
  in Hong Kong and London compared
• Effect of air pollution on daily mortality

6
PAPA Project
Hong Kong, Bangkok, Shanghai and Wuhan were
selected in the 1st phase. While all cities are
to perform time-series studies on mortality, Hong
Kong will also study effects of air pollution on
hospitalization.

• Average 30,000 deaths per year, including 26
  cardiovascular and 19 respiratory.
• Average 1.3 million hospital admissions,
  including 8 cardiovascular and 11 respiratory.

7
Daily counts of health outcome will be modeled by
Poisson regression with adjustment for
seasonality, trend and covariates.
Temperature, Humidity, Holidays
8
Daily air pollutants will then be assessed in the
model after the adjustment.

• Hourly data in each monitoring station
• Definition 8 hourly mean of O3 24 hourly mean
  of SO2, NO2 PM10
• Missing data lt 75 of hourly data

9
In addition, the Hong Kong team will assess the
effects of air pollution and influenza, as well
as interaction between these two agents on
mortality and morbidity.
Identification of epidemics based on
virological data
10
Progress
We had visited all 8 air pollutant monitoring
stations to be used in the study.
11
The surrounding environments are representative
of the living environments of Hong Kong with
residential areas close to commercial complexes
and traffic routes.
Picture of Kwun Tong site
12
We collected documents for site selection
criteria, data processing policy and procedure,
data validation and performance audit procedure
from the Environmental Protection Department.
13
We obtained an outline of data collection and
quality control for isolation of respiratory
viruses.
14
We assessed the procedure in retrieving data for
death from the Death Registry, Department of
Health and Census Statistics Department.
15
Clinical data are captured in real time by a
Clinical Management System. After the accuracy
and coverage of the Clinical Data Format were
assessed to be satisfactory in an ad hoc study,
the system was extended to all major public
hospitals.
16
All the data for year 1996-2002 have been
collected and are ready for analysis.
Air pollutants NO2, SO2, PM10 and O 3 Mortality
All natural causes, Cardio-pulmonary,
Cardiovascular, Stroke, Cardiac or heart disease,
Respiratory, Lower respiratory infections,
Chronic obstructive pulmonary diseases,
Tuberculosis, Digestive and genitourinary,
Neoplasm excluding lung cancer
Hospital admissions Cardiovascular, Stroke,
Ischaemic heart disease, Respiratory, Acute
respiratory diseases, Acute lower respiratory
infections, Chronic obstructive pulmonary
diseases, Asthma
Virology Specimens, Flu AB and Respiratory
syncytial virus isolations Meteorological
Temperature, Relative humidity
17
We first assessed a proxy measure for influenza
epidemics according to weekly numbers of
respiratory deaths above a cut-off. The validity
of the measure against the gold standard
according to virology data was high.
18
Way forward
We will proceed with the following in the coming
5 months.

• Standardization of data structure
• Statistical modeling
• Definition for influenza epidemics
• Documentation of research procedure and outputs
• Development of common protocol for the PAPA
  studies

19
Acknowledgement

• Health Effect Institute
• Environmental Protection Department
• Hospital Authority
• Department of Health
• Census and Statistics Department