euroheis 2 dr linda beale october 2007 – september 2010

19
EUROHEIS 2 Dr Linda Beale October 2007 – September 2010

Upload: buddy-bruce

Post on 28-Dec-2015

216 views

Category:

Documents


0 download

TRANSCRIPT

EUROHEIS 2

Dr Linda Beale

October 2007 – September 2010

EUROHEIS2 objectives

General objective

• Develop further methods for integrating and analysing information on environmental exposure and human health

Strategic objectives

• Linking data on environmental pollutants to (routinely collected) health data

• Collaboration with other EU funded projects (INTARESE, HEIMTSA et al.)

Small Area Health Statistics Unit (SAHSU)

Developed methods and eventually a tool for SAHSU staff to analyse UK’s routinely collected health/population data

Used to investigate environmental and other factors in explaining local geographic variations in disease with respect to other factors such as demographic, environmental, socio-economic risk factors.

The RIF

The RIF is a tool that allows users to assess relationships between the environment and health

• Links spatial and non-spatial data

• Embedded in ESRI® ArcGIS

• Risk analysis around putative hazardous sources

• Disease mapping

Data requirements

• Accurate health event data, located geographically to a place of residence or small geographical area

• Population data (e.g. from a national census) by small geographical area, and by age and gender

• Spatial data of area boundaries/point locations that link to the health and population data

Optional:• Covariate data e.g. socio-economic status, income or

ethnicity

• Exposure data

Linking spatial and non-spatial data in the RIF

ACCESS/ORACLEdatabase

Geographical areas(administrative/ hierarchical geography)

Numerator data(cancer registrations, mortality data, hospital admissions, congenital malformation registrations

Denominator data(population census output)

Covariate data(SES, ethnicity, income…)

Define study area

ArcGIS

ODBC

Define comparison area

spatial select

db select

Spatial DataGeographical boundariesExposure data (land use, TRI sites,…)Contextual Information

Define investigation

Do study

View data

Study report

WinBUGS

SaTScan

Maps

Output & Export

Run the RIF

Run external models

Types of analysis

1. Risk analysisAllows assessment as to whether a risk factor has a statistical association with a health outcome in a local population selected by:

• distance bands around one or more user defined point or area sources

• modelled exposure

2. Disease mappingAllows a user to visualise mortality or morbidity rates and spatial patterns of health outcomes, selecting by:

• Variables stored in the database

• Spatially selected areas

Output: Rates and risks

Directly standardised rates • Apply the study area stratum-specific rate to the comparison

area population+ Can be directly compared between exposure groups - Can be unstable if small populations/rare diseases

Indirectly standardised risks• Apply the comparison area stratum-specific rate to the study

area population+ More stable as based on larger comparison population rates - Not directly comparable between different exposure groups (esp

where population structure significantly different).

Directly Standardised Rates (DSR)

•DSRi is a weighted average of the specific rates, using as weights the population of the comparison region

•Calculation of DSRi can be seen as a projection of the area specific rates of the study region onto the population of the comparison region.

•Confidence intervals

5

*

*

10j

j

iji N

NrDSR

ddistributenormally ely approximat

are that assumption the on based is ncalculatio If )(DSRO ii log,100

Poisson the of tables lstatistica the from obtained are CI95% the If ,100iO

Standardised Mortality/Morbidity Ratio (SMR)

•SMRi provides a measure of the relative risk of area i compared to that of the comparison region.

•Confidence intervals:

i

ii E

OSMR

region comparison the in asmortality same the dexperience had area

that if area in cases expected adjusted of number total iNrEj

ijji :*

ddistributenormaly

ely approximat are that assumption the on based is If )(SMRO ii log100

on)distributi the withiprelationsh almathematic its (via

Poisson the of tables lstatistica the from obtained are CI95% the If2

,100iO

Further analysis

• Empirical Bayes smoothing» Low counts of observed cases/ small populations

» Both rates and SMRs become numerically unstable (rates even more than SMRs)

• Chi square tests for homogeneity and linear trend (with accompanying p values)» test global association between a distance/exposure and relative risks

• Graphs of the risks as a function of exposure per band (risks plotted on a log-scale)

• Full Bayes smoothing (WinBUGs)• Spatial scan - Statistically significant clusters (SatScan)

EUROHEIS2 specific objectives

To enhance and test the RIF user interface further to make it more user friendly and readily transferable to other EU countries

To arrange workshops in partner countries to discuss methodology and suggest enhancements to the RIF within the EUROHEIS framework

• comprehensive workshop reports will be produced

Technical and statistical qualities of the RIF

• Enhance the import and export functions within the RIF» These should include additional ability to export selected data from the

RIF» Import and export of a range of commonly used EU data types and

sources will be ensured, including country specific denominator data and a range of local geographies

» This work will extend compatibility with other approaches and methods

• Include spatio-temporal methods for disease mapping in RIF

• To add measures of uncertainty to disease mapping, and visualise this uncertainty in the maps

User interface and test cases in new countries

• To incorporate the capability to include EU country specific indices of socio-economic status (SES)

» to enable the user to choose from a selection of indices to standardise for in analyses of environmental health risks

• To use data on SES and environmental pollution to allow users to assess inequalities in health as well as environmental equity

• Test the user interface and the expanded RIF software

• To set up a web-based support tool (web-forum) assisting member countries in implementing and operating the system

Dissemination

• Disseminate the RIF software as freeware via the internet

• Supply training courses and material to interested EU countries

• Organise an end of project conference showing the advances made during the project and summarise the overall project strategic developments

• Identify dissemination mechanisms for reaching target audiences

Involvement of policy makers

• To interact with stakeholders at relevant workshops, ensuring the policy relevance of project work

• To raise awareness of the policy implications of the issues and trade-offs surrounding data governance, data protection, privacy and data quality issues

• Raise awareness of accurate (health) data collection, across the EU as an input to spatial epidemiological analyses

Good practice recommendations and future work

• Recommend data quality indicators to aid interpretation of the results

• Identify issues in integrating the RIF into existing spatial data infrastructures, such as SMASH and the Health Atlas

EUROHEIS2 Work packages

WP 1. CoordinationWP 2. DisseminationWP 3. EvaluationWP 4. Adaptation and enhancements of the current RIF to EU conditionsWP 5. Evaluation of RIF for integrated assessment of environment and health risksWP 6. Spatio-temporal methods for disease mappingWP 7. Exposure databases and GIS methodsWP 8. Health and Environment Information System in PolandWP 9. Health and Environment Information System in HungaryWP 10. Integration of RIF into existing spatial data Infrastructures

Partners

Organisation Town / City Country

University of Valencia Valencia Spain

National Public Health Institute

Kuopio Finland

National Institute of Environmental Health

Budapest Hungary

Dublin City University Dublin Ireland

National Institute for Public Health and the Environment

Bilthoven The Netherlands

Nofer Institute of Occupational Medicine

Lodz Poland

Lund University Lund Sweden

Imperial College London London UK