The Ontario Health & Environment Integrated Surveillance (OHEIS) - - PowerPoint PPT Presentation

The Ontario Health & Environment Integrated Surveillance (OHEIS) Project

• E. Holowaty, P. Brown, T. Norwood, S

. Wanigaratne Population S tudies and S urveillance, Cancer Care Ontario May 2, 2008

Presented to: The Association of Public Health Epidemiologists in Ontario (APHEO)

Overview

• Introduction to the RIF
• Relevance to APHEO
• Technical details & requirements
• RIF database
• RIF demo
• Opportunities and Challenges
• Questions & Discussion

2 2

The Rapid Inquiry Facility

S

mall Area Health S tatistics Unit at Imperial College London

User friendly tool, used to rapidly address public

health questions using routinely collected health and population data

• risk analysis around putative hazard sources and across

covariate categories

• sophisticated risk mapping

Calculates standardised incidence ratios and directly

standardised rates for any health outcome, for specified age and year ranges, for any given geographic area

3 3

Disease Mapping with the RIF

Addresses problems of small populations and counts by

calculating indirectly standardised incidence ratios

Covariate-adj usted S

IRs using Poisson assumption for rare health outcomes

Allows for spatial dependence between neighbouring

regions using random effects models

Interfaces to other readily available statistical

programs to perform Bayesian inference and cluster analysis

4

5

6

slide of income

7

8

9

Risk Analysis: point source exposures

Calculates standardised incidence ratios and directly

standardised rates for each exposure band or covariate category, does test for homogeneity and linear test for trend

10

tables and graphs are

• utput from risk analysis

11

How is this relevant to APHEO?

Ontario Public Health Standards

… establish requirements for fundamental public health programs and services, which include assessment and surveillance… … key component of the requirements… is to identify and work with local priority populations… identified by surveillance, epidemiological, or other research studies… 12

Population Health Assessment Surveillance Protocol

The board of health shall: … adopt, adapt, or develop techniques, tools and/ or systems for… collection, management, and integration of data… … make comparisons by person, place and time to consider the relationships among these elements… includes analysis of health unit data and how data are spatially distributed… … consider the following aids to interpret and understand various aspects of a health issue… by… maps…

13

Identification, Investigation and Management of Health Hazards Protocol

Purpose: to prevent or reduce the burden of illness from environmental health hazards. The board of health shall: … monitor and collect data on the health status of residents in the board of health area, including but not limited to epidemiological studies focusing on adverse health

• utcomes potentially related to health hazards in the

environment … conduct analysis and evaluation of the data collected to identify potential human health risks from health hazards in the environment

14

RIF – Technical Details

RIF Version 3.11

• An extension for ArcGIS

9.0 or higher. Primarily, hardware requirements are determined by ArcGIS requirements:

• 1.6 GHz processor or better Intel Duo, Pentium or

Xenon processors

• 1 GB RAM minimum; 2GB recommended
• Disk space 1.2 GB
• 64 MB video card recommended
• Windows XP, 2003, or Vista
• Microsoft Office 2003 or higher
• Microsoft Access or Oracle 10 or higher

15

Technical Details - RIF application architecture

OCR & Census data (Access or Oracle)

Read Write SQL Commands

ArcGIS

Geographic Information System (GIS)

RIF Database

16

RIF Technical Details

CCO’ s OHEIS

S erver Architecture

ArcGIS Database Server

Web map Service

Firewall RIF (ArcGIS Desktop) Internal Users LAN RDMS Web Server External Users (Project collaborators)

Web server: Open GIS compliant service (e.g. WMS) Web map Service

RIF (ArcGIS Desktop) WWW

ArcGIS Server

17

RIF Database

Designed on basis of geographical hierarchies. For Canadian data:

PR* CMA PR* CD* CSD* DA/EA CT DA LHIN* DA PR*

SGC

(Standard Geographical Classification)

SAC

(Statistical Area Classification)

LHIN

(Local Health Integrated Network)

Hierarchy name: Hierarchical levels: * Note: intercensal population estimates available for these geographical levels. +Note: if applicable PHU

(Public Health Unit) PR* PHU DA PHU Custom Geography+

18

RIF Database

RIF Database contains 48 tables required by the

application to run

You add tables for population, health events,

covariates and geographic hierarchy lookup

Of these tables that you add to the database,

generally, you can think of them as:

• numerator data –

your health outcome file

• denominator data –

population by age-sex groups

19

RIF Database

Numerator data: coded by age-sex groups, and

health event coding if applicable

Denominator data: population by geographic

• area. You may use multiple geographic

hierarchic levels. For each geographic level:

• lookup table
• table of population by age-sex groups
• covariate table (optional)

20

RIF Demonstration

S

IRs by 2001 DA for Windsor CS D

WinBUGS

– Full Bayesian S moothing using the Besag, York and Mollié (BYM) model 21

Opportunities

• Interest and use of GIS

for health data

• Computing power and software availability
• Availability of georeferenced data
• Expectations for rapid hazard appraisal and

more granularity in community health profiling

• General calls for widespread use and easy

access

• S

ignificant advances in spatial analysis

22

How can GIS help Public Health?

Research, S

urveillance and Planning

• Hypothesis generation –

maps, correlations, clusters

• Models of disease risk
• S

ervice planning and optimisation

• Making predictions e.g. Health Impact Assessment

S

patial Decision S upport S ystems

• Infrastructure –

roads, towns, HC services/ availability

• Census –

population statistics; socio-demographics

Emergency/ Pandemic Response S

ystems

• 911 services
• Disease/ event registers, including infectious diseases

23

Special challenges

Accuracy, granularity and completeness of exposure,

health and population data, and boundary files

Geocoding, i.e., accurately assigning spatial

coordinates to location info.

Current place of residence may not be good proxy

for exposure

Problems adj usting for known confounders Necessity of aggregating counts S

cale/ zone translation problems (MAUP)

S

patial autocorrelation

Data access and confidentiality restrictions

24

Measuring and reducing disclosure risk

Estimating risk

• Population uniques
• Probabilities of identification
• Absolute vs. relative risk of disclosure

Institutional approaches

• Full public access if no disclosure risk
• Licensing of users
• Data enclaves

Technical approaches

• Aggregation
• Data limitation –

cell suppression, rounding

• Geo-masking –

swapping, random perturbation

25

Aims of OHEIS Proj ect

To design, implement and test an enhanced

spatial surveillance system for mapping and risk assessment

Rapid assessment (3-6 mos. <1 week) S

cientifically sound, robust, understandable and transparent

Easily export data to more sophisticated spatial

analysis systems – WinBUGS ; S aTS can; EXCEL

To share data, software and/ or RIF outputs

with PH partners thru web-based portal and/ or stand-alone, secure PC applications. 26

Resources

“The GIS Primer” web-based

http:/ / www.innovativegis.com/ basis/ primer/ primer.html

“Health and Environment Information Systems for

Exposure and Disease Mapping and Risk Assessment”

Jarup et al. Environmental Health Perspectives. 2004. Vol. 112: 995-525.

“GIS and Public Health”

Cromley EK and McLafferty S

• L. Guildford Press. 2002.

“ Putting People on the Map : Protecting

Confidentiality with Linked Socio-Spatial Data”

Gutmann MP et al. National Research Council. 2007. http:/ / books.napedu/ catalog/ 11865.html

27

Resources (cont’ d)

“Spatial Epidemiology : Methods and Applications”

Elliott P. et al. Oxford University Press. 2000.

“Applied Spatial Statistics for Public Health”

Waller LA and Gotway CA. Wiley Interscience. 2004.

“ Geographic Information Systems and Public Health”

Richards TB et al. Public Health Reports Vol.114.1999. http:/ / www.healthgis-li.com/ library/ phr/ phr.htm

“ Public Health and GIS”

Rushton G et al. Annual Review of PH. Vol.24.2003.

28

Appendix A – RIF Demonstration S lides

These slides are intended to allow teleconference users to follow the live demonstration

• f the Rapid Inquiry Facility (RIF).

29

Disease Mapping: Lung Cancer Incidence in Windsor, Ontario

• The RIF is an extension to ArcGIS

and adds a “ RIF” menu to the ArcMap application. To start a new study, within the menu, simply go to “ S tudy” ->” New” :

30

S tep 1 – Define S tudy Type

For this demonstration, we will be showing a

disease mapping analysis:

Click “ Next”

31

S tep 2 – Define S tudy Area and Geographic Resolution

• There are two components to S

tep 2:

1. Define t he geographic unit s of analysis – the geographic boundaries or spatial units for which the rates and risk ratios will be calculated 2. Define the study area resolution – the extent of your study area. This you can do by selecting a region on the map, from a list, writing S QL or loading a previously created text file. Click “ Next”

32

S tep 3 – Define Comparison Population/ Area

In step 3, we define what population will be used to

calculate indirectly standardized rates. Here, we are going to select Ontario:

Click “ Next”

33

S tep 4 – Investigation details

• At step 4, we define one or more investigations by selecting age

groups, sex, covariates for adj ustment, and health event:

Note that we may do more than one invest igation at a t ime. Click “ Next” .

34

Run the S tudy… .

35

The RIF

calculates and saves the

• bserved

counts, expected counts, rates, S IRs and confidence limits for each geographic unit in the comparison population and study areas

Disease Mapping – results of analysis

A “ Disease Mapping”

menu is added to ArcMap

Within this menu,

there are options to map:

• “ rates” –

standardised rates; or,

• “ relative risk” –

S IRs

• adj ustment for

covariates may be included or excluded on the

• map. (If no

covariates were selected those

• ptions do not

appear in the menu).

36

Windsor, Male Lung Cancer Incidence 1999-2003

37

Windsor, Male Lung Cancer Incidence 1999-2003

S IRs adj usted for CMA/ CA household income quintiles

38

Interface to WinBUGS – implements Besag, York and Mollié (BYM) Model (1991)

39

The RIF calculates the adj acency matrix, saves files

to folder specified and runs the BYM Model in WinBUGS , with three Markov chains. Results are returned as map layer of “ smoothed” S IRs and exceedence probabilities.

Interface to WinBUGS – implements BYM Model (continued)

• For Windsor 2001 Dissemination

Areas, it takes WinBUGS about 10 min. to complete the model and return results to ArcMap.

40

Windsor, Male Lung Cancer S IRs 1999-2003

Adj usted for spatial dependence and CMA/ CA Household Income quintiles covariate:

41

End of Demonstration

Thank You!

42