Upload
brianna-vega
View
217
Download
2
Tags:
Embed Size (px)
Citation preview
Enabling obesity research using the Health Surveys
for England: The Obesity e-Lab Obesity e-Lab project
Dexter CanoyThe University of [email protected]
Obesity as a Public Health ProblemObesity as a Public Health Problem
• Increasing prevalence: by age, sex, ethnicity, SES
• A global phenomenon
• Chronic diseases
• Health costs
• Economic costs
Energy equation balanceEnergy equation balance
Social determinants
T I M E
Social determinants
Energy deficiency & excess in Energy deficiency & excess in CubaCuba
0
5
10
15
20
25
30
35
40
1982 1993 1998 1982 1993 1998 1982 1993 1998
Underweight Overweight Obese
Prev
alen
ce (%
)
Men Women
Rodriguez 2002, Public Health Nutr
Role of individual and societyRole of individual and society
World
Nation
Region
Local community
Household
Individual (modifiable)
Individual (fixed)
Challenges in obesity researchChallenges in obesity research• Requires social, behavioural, bio-medical and environmental
perspectives
• Unfamiliarity with other domains of research communities: Lack of understanding of theoretical concepts underpinning the data collected by other disciplines
• Lack of awareness of relevant data sources: Secondary data sources
• Lack of sharing of reproducible packages of research incorporating data and data processes
Time-related changes in excess Time-related changes in excess deaths due to obesity in the USdeaths due to obesity in the US
NHANES: I – 1971 to 1975 II – 1976 to 1980 III – 1988 to 1991
The The Obesity e-Lab Obesity e-Lab projectproject• ESRC-funded project (Sept 2008 to Aug 2011)
• Investigators/project team from diverse background:– Public Health, Informatics, Statistics, Social Science, Epidemiology, Computer
Science, Human Computing Interaction, Software Technical Architecture, Software Engineering, Education & Professional Development
• Work alongside other project teams (e.g., the e-Science community, text mining community, etc)
• Supported by stakeholders representing various groups (local health authority, leaders of obesity-related initiatives, etc.)
• Linking with UK Data Archive, ESDS, CCSR
The The e-Labe-Lab model model
Secure, online environment that brings together DATA, RESEARCH METHODS, ANALYTICAL TOOLS, REFERENCES and INVESTIGATORS
User workbench (software)
DATA ANALYSIS METHODS VISUALISATION REFERENCE
Research Question
Sharing, collaboration
Data and Methods
Obesity e-Lab Obesity e-Lab modelmodel• Data
– Search and browse for relevant datasets and relevant variables across survey periods– Convenient access to supporting meta-data– Support valid comparisons of variables or databases across years– Download relevant dataset for further analyses
• Analysis and methods– Detailed record of user’s research process (from data selection to analysis)– Support users to link analysis with particular variables– Develop an archive of useful scripts and expertise
• Visualisation methods– Incorporate graphical and map-based analysis using geo-visualisation application
• Dynamic reference methods– Allow users to look and discover relevant work across disciplines– Linkage with variables, databases or other relevant survey documentation
User communitiesUser communities
• Academic researchers– Public health/Epidemiology/Medicine– Social Science
• NHS and DH data analysts (Public Health Intelligence Specialists)– Local public health decision making– Application of national directives to local context
The The Obesity e-Lab Obesity e-Lab development development approachapproach
• Engagement with academic research users– Novice and regular users of surveys
• Engagement with data custodians– UK Data Archive, ESDS
• Engagement with expertise in other domains– Technical architecture development
• Engagement with PCTs in the UK Northwest– Understand problems and processes– Identifying good practices– Planning training and support (eg, for professional development)
Potential use of Potential use of Obesity e-LabObesity e-LabHow can I compare local obesity prevalence to national sample?
Are the factors influencing national obesity prevalence similar for my
local area?
(1) Could I do it myself?
(2) Has anyone else done it? Or, I’ve done
something – can I share this to my
colleague but not my supervisor just yet?
Dat
a (e
.g.,
Hea
lth S
urve
y fo
r Eng
land
)
Anal
ysis
tool
s &
Res
earc
h M
etho
ds
(2)
(1)
My Obesity e-Labbeta
Building the prototypeBuilding the prototypeTesting the technical architectureTesting the technical architecture
• Search & download variables
Variables are grouped into categories
Browsing a category lets youview and select associated variables
Where are we now?Where are we now?Supporting the individual researcherSupporting the individual researcher
• Focus on the HSE
• Making it easier to find relevant variables within and across surveys
• Making it easier to understand what a variable means
• Automatically making a trail of what you’re doing and what you’ve selected
Where are we now?Where are we now?Developing a communityDeveloping a community
• Sharing expertise and creating dialogue
• User profiles (similar to Facebook, MySpace)
• Sharing scripts, calculations, paper references
• Add comments to surveys, survey years or variables
• You choose who sees what – yourself, trusted colleagues, everyone
age Search
Variable Year Description Type
Select variables
Select all Invert selection
Selected datasets: 1998, 2003, 2006Change selected datasets
Survey CartYou have no variables in your cart
age Search
Variable Dataset Description Type
CurrAge 2006 Respondent’s current age in years DerivedChildSchoolAge 2006 Number of children of school age DirectMonthAge 2006 Respondent’s current age in months DerivedDaysAge 2006 Respondent’s current age in days DerivedCurrAge 2006 Respondent’s current age in years Derived
Select variables
Select all Invert selection
ChildSchoolAge 2006 Number of children of school age DirectMonthAge 2006 Respondent’s current age in months DerivedDaysAge 2006 Responsents’s current age in days Derived
SPSS Derivation
Sjdflskdfjlsdkfk jlkjl k lasiodaoilklkj
Sdlfkjsdlfkjsdlfkjss daslkd aslkd ajslkd
No. of respondents: 3879Response rate: 99.67%Mean:46.39
Selected datasets: 1998, 2003, 2006Change selected datasets
Survey CartYou have 1 variable in your cartView cart
1
New search
Related information to the variable can be read or accessed in the same page
Next stepsNext steps
• Continue working with users– Explore ‘what users want’– Testing the next prototype– HSE training course
• Continue dialogue and engagement with other users of HSE– Is this concept interesting to you?– Can we ask about how you use HSE or other survey data?
Long-term visionLong-term visionHelp promote use of secondary analysis of data Help promote use of secondary analysis of data
sourcessources
• Database from research projects can be created and shared– Project protocol and documentations– Workflow for extracting relevant subsets of the database– Structured container for the subsequent uses of the data extract
(statistical scripts, result summaries, slides, manuscripts, reference lists, etc)
• Capturing, sharing and re-using research processes becomes routine
• Beyond obesity• Beyond survey data
THANK YOU!THANK YOU!