Estimating Census Health Geographies: using synthetic estimation with secondary survey and census data

Lead Research Organisation: University of Southampton
Department Name: School of Geography

Abstract

The 2011 Population Census was almost certainly the last to be undertaken using traditional methods of survey completion. It is acknowledged however that there will continue to be a requirement for comprehensive data on small area variations on a range of topics for service planning, evaluation and many other needs. This continuing need extends across the public and private sectors. Alternative approaches to the provision of small area data exist. Most attention, in the context of post 2011 census debates, has focussed on the potential for record linkage arguing that much present census data is available, in some form, in other databases. An alternative, less-explored approach is small area synthetic estimation (SASE). Essentially two approaches to SASE are available: microsimulation and multilevel SASE (ML-SASE). Both approaches entail the integration of secondary data from survey sources and subsequent statistical modelling.
This proposal focusses, in line with expertise of the investigators, on ML-SASE. It will assess the effectiveness of linking secondary survey data with ML-SASE as an alternative route to the production of census-equivalent small area data. It builds on ESRC's secondary data investment in the 2011 census and exploits ESRC managed access to a range of national surveys. It also builds on earlier work in developing the small area estimation techniques that form the basis for the proposal. We will compare and contrast different approaches to ML-SASE, generating estimates for four health indicators present in all national versions of the 2011 UK census using survey data collected at the same time as the Census. Further comparison will then be made between the generated estimates and the actual results of the 2011 census to examine the extent to which estimates match census 'reality'.
The originality of the proposal lies in addressing the relative neglect ML-SASE in current work on census futures. The proposal will also develop and enhance ML-SASE methodology to move beyond now standard approaches and draw benefit from recent advances in multilevel modelling. The substantive practical and policy contribution flowing from the proposed research will be a significant enhancement to the the evidence base underpinning policy debates on approaches to the replacement of the traditional census. The data generated will be relevant to the key ESRC strategic challenges on influencing behaviour and developing a free and fair society

Planned Impact

The proposed research will achieve academic impact through the completion of high quality research demonstrating the potential of linked survey datasets and ML-SASE in producing small area health indicators. This research will feed into academic and policy level debates on future census replacement strategies. This academic impact will flow from the dissemination of project results in leading journals and at relevant conferences. Key disciplines to be impacted include geography, social statistics, social epidemiology and public health.

We will promote impact with non-academic bodies through public engagement and communication with key stakeholders, and through direct involvement of non-academic agencies in the final phase of the project. The intention will be to ensure wide awareness of alternative future strategies for generating small area data and an appreciation of the potential uses of linked survey data. Non-academic imapct will be a specific emphasis in the end-of project workshop and there will also be specific consideration of non-academic bodies in the design of the project website. Through these strategies we hope to contribute indirectly to future policies and practice concerned with increasing the effectiveness of public services and enhancing health. Specific targets for impact include:
- private sector market research companies seeking alternative approaches to add value to their existing data and target new markets, from data linkage and new forms of catchment data, as well as the opening up of potential new markets.
- public sector agencies using census data and concerned about the future provision of such data, for example public health observatories and equivalents in the reformed NHS. There will be regular contact with the Office for National Statistics (ONS) throughout the lifetime of the project to ensure potential impact on future post census debate.
- international agencies with interests in the provision of small area data; we have preliminary interest from the EU (DG SANCO) and Statistics New Zealand.

Publications

10 25 50

publication icon
Rowlands G (2018) Developing and Applying Geographical Synthetic Estimates of Health Literacy in GP Clinical Systems. in International journal of environmental research and public health

 
Description This research tested the utility of routine survey data as a basis for the generation of small area estimates of health indicators currently provided in the UK Census. It also advanced methods for small area estimation.
The 201 UK Census provided information on four health topics on a UK-wide basis (additional information was collected for Scotland and Northern Ireland alone). The UK-wide topics were limiting long-term illness (LLTI), self-assessed health status (SAHS), (in)activity through disability, and participation in the provision of informal social or health care. The research has discovered that:
- Despite commitments to common question wordings and formats, there remain significant differences in the measurement of the four health topics between routine official surveys and between such surveys and the UK census. There are also differences in coverage, sampling design and the availability of spatial identifiers. These variations mean that the utility of routine surveys for small area estimation at a UK-scale is essentially restricted to LLTI and SAHS. Even the merging of separate national surveys can be problematic. Small area estimates of disability would be possible but are unnecessary as improving access to administrative data now offers an alternative source of information.
- Where available, either routinely or through secure access, it appears that knowing the geographical location of survey respondents can enhance small area estimates by allowing local contextual data to be linked into multilevel estimation models. This enhancement is not however always to be found and approaches to small area estimation need to test for its utility. The research has developed methods to do this.
- Subject to the above, it is possible to generate robust small area estimates using routine survey data down to the MSOA level for LLTI and SAHs for the nations of the UK. These estimates show a strong convergence with the UK Census but issues of question wording and the limitations of the small area estimation method ensure that the match is not perfect. Small area estimates of the provision of informal care in Scotland differ significantly from the census for similar reasons. It should be noted that the UK Census for all its strengths is itself not a gold standard.
- Small area estimation methodology can be extended to recognise multinomial outcomes, allowing for example estimation of varying levels of LLTI. Models can also be reformulated to use orthogonal polynomials to measure ordered categorical predictors; this can improve model parsimony for example with age effects. Geodemographic indicators (area typologies) can significantly enhance small area estimates.
An original intention was to link the research to debates over the future of the UK Census. Policy developments during the grant resulted in a decision to retain the current form of the UK census for 2021 and shifted the focus of the research to inter-censal change. Delays in data access resulting from the care.data embargo was recognised by a three month no-cost extension till February 2016.
Three further academic papers are in progress disseminating the above findings and our two working papers have been resubmitted to academic journals.
Exploitation Route We are collaborating in a bid to NCRM to build on our participation in the NCRM Methodological Network and further advance small area estimation methodology seeking synergies between statistical approaches (as used in our work) and microsimulation. Future academic advances are likely to focus on methodological advances, for example in accommodating distributional assumptions and dealing with missing data in survey models. These developments will of necessity need to be accompanied by more uniform survey wording and enhanced (secure) access to geographical identified data.
Our strategy for non-academic extensions to the research has focussed on achieving impact, facilitating understanding and promoting uptake within the health sector both nationally and internationally. Through our workshops with civil servants and public health practitioners we are aware of and are addressing significant opportunities in health needs assessment and service benchmarking. Robust, consistent small area estimates can show both how otherwise unmeasured phenomenon vary at the small area level and also provide an indication of expected outcomes against which actual levels can be set. Our work has shown that achieving this ambition requires careful training of non-academic users and senior policy level commitment. We are specifically targeting Public Health England as a first step to other national equivalents.
Sectors Healthcare

URL http://www.synthetic-estimation.moonfruit.com
 
Description To underpin methodological development of small area estimates of adolescent smoking for Public Health England and NICE. These estimates have been disseminated to health care planners and practitioners via the NHS 'Local health' web tool and have been used to formulate local youth smoking strategies.To train and develop understanding of small area estimation in the Netherlands public health practitioner community. Two CPD courses have been run and delegates have developed small area estimates of access to care / need for care in the Netherlands. These estimates have been used to assess local care needs and plan resource allocation. To provide public health practitioners with intelligence on local variations in health literacy between small areas in Inner London. This work is feeding in to service developments regarding access and use of emergency services.To provide public health practiioners with small area data on variations in adolescent and adult smoking and adult obesity. This work feeds Public Health England's Local Health Tool.
First Year Of Impact 2016
Sector Healthcare
Impact Types Societal

 
Description Development and production of small area estimates of adolescent smoking for NICE/PHE
Geographic Reach National 
Policy Influence Type Citation in other policy documents
URL https://www.gov.uk/government/news/new-tool-models-youth-smoking-variation-across-england
 
Description Three workshops on small area estimation
Geographic Reach Europe 
Policy Influence Type Influenced training of practitioners or researchers
 
Description ESRC NCRM Methodology Research Programme
Amount £648,831 (GBP)
Funding ID ES/N011619/1 
Organisation Economic and Social Research Council 
Sector Public
Country United Kingdom
Start 01/2016 
End 12/2019
 
Description Kings College London
Amount £7,428 (GBP)
Organisation King's College London 
Sector Academic/University
Country United Kingdom
Start 03/2015 
End 05/2015
 
Description NICE/Public Health England Contract
Amount £25,943 (GBP)
Organisation Public Health England 
Sector Public
Country United Kingdom
Start 08/2014 
End 11/2014
 
Description Public Health England
Amount £69,133 (GBP)
Organisation Public Health England 
Sector Public
Country United Kingdom
Start 04/2017 
End 01/2018
 
Description University of Southampton ESRC DTC
Amount £21,000 (GBP)
Organisation University of Southampton 
Sector Academic/University
Country United Kingdom
Start 09/2012 
End 06/2016
 
Title Estimating census health geographies: using synthetic estimation with secondary survey and census data 
Description Deposited data comprising small area estimates of census-equivalent health variables using different estimation models 
Type Of Material Database/Collection of data 
Provided To Others? No  
Impact None as yet. Data deposited 2015 
URL https://discover.ukdataservice.ac.uk/catalogue/?sn=851972&type=Data%20catalogue
 
Description Evaluating and improving small area estimation methods: ESRC NCRM Network for Methodological Innovation 
Organisation Imperial College London
Country United Kingdom 
Sector Academic/University 
PI Contribution Contributed to application for funding, participated in meetings, contributed to and edited final report to NCRM
Collaborator Contribution As above under leadership of University of Sheffield
Impact Three collaborative workshops with materials available on website above
Start Year 2012
 
Description Evaluating and improving small area estimation methods: ESRC NCRM Network for Methodological Innovation 
Organisation Office for National Statistics
Country United Kingdom 
Sector Public 
PI Contribution Contributed to application for funding, participated in meetings, contributed to and edited final report to NCRM
Collaborator Contribution As above under leadership of University of Sheffield
Impact Three collaborative workshops with materials available on website above
Start Year 2012
 
Description Evaluating and improving small area estimation methods: ESRC NCRM Network for Methodological Innovation 
Organisation Ritsumeikan University
Country Japan 
Sector Academic/University 
PI Contribution Contributed to application for funding, participated in meetings, contributed to and edited final report to NCRM
Collaborator Contribution As above under leadership of University of Sheffield
Impact Three collaborative workshops with materials available on website above
Start Year 2012
 
Description Evaluating and improving small area estimation methods: ESRC NCRM Network for Methodological Innovation 
Organisation Teagasc
Department Teagasc Rural Economy Research Centre
Country Ireland 
Sector Public 
PI Contribution Contributed to application for funding, participated in meetings, contributed to and edited final report to NCRM
Collaborator Contribution As above under leadership of University of Sheffield
Impact Three collaborative workshops with materials available on website above
Start Year 2012
 
Description Evaluating and improving small area estimation methods: ESRC NCRM Network for Methodological Innovation 
Organisation University of Canberra
Country Australia 
Sector Academic/University 
PI Contribution Contributed to application for funding, participated in meetings, contributed to and edited final report to NCRM
Collaborator Contribution As above under leadership of University of Sheffield
Impact Three collaborative workshops with materials available on website above
Start Year 2012
 
Description Evaluating and improving small area estimation methods: ESRC NCRM Network for Methodological Innovation 
Organisation University of Leeds
Country United Kingdom 
Sector Academic/University 
PI Contribution Contributed to application for funding, participated in meetings, contributed to and edited final report to NCRM
Collaborator Contribution As above under leadership of University of Sheffield
Impact Three collaborative workshops with materials available on website above
Start Year 2012
 
Description Evaluating and improving small area estimation methods: ESRC NCRM Network for Methodological Innovation 
Organisation University of Liverpool
Country United Kingdom 
Sector Academic/University 
PI Contribution Contributed to application for funding, participated in meetings, contributed to and edited final report to NCRM
Collaborator Contribution As above under leadership of University of Sheffield
Impact Three collaborative workshops with materials available on website above
Start Year 2012
 
Description Evaluating and improving small area estimation methods: ESRC NCRM Network for Methodological Innovation 
Organisation University of Sheffield
Country United Kingdom 
Sector Academic/University 
PI Contribution Contributed to application for funding, participated in meetings, contributed to and edited final report to NCRM
Collaborator Contribution As above under leadership of University of Sheffield
Impact Three collaborative workshops with materials available on website above
Start Year 2012
 
Description Evaluating and improving small area estimation methods: ESRC NCRM Network for Methodological Innovation 
Organisation University of St Andrews
Country United Kingdom 
Sector Academic/University 
PI Contribution Contributed to application for funding, participated in meetings, contributed to and edited final report to NCRM
Collaborator Contribution As above under leadership of University of Sheffield
Impact Three collaborative workshops with materials available on website above
Start Year 2012
 
Description Public Health England - Small Area Estimation 
Organisation Public Health England
Country United Kingdom 
Sector Public 
PI Contribution Development of small area estimates of obesity and smoking
Collaborator Contribution Data analysis and report authoring
Impact Internal report to PHE
Start Year 2017
 
Description Training collaboration with Public Health Netherlands 
Organisation GGD Amsterdam
Country Netherlands 
Sector Public 
PI Contribution Delivery of training in small area estimation; two workshops and ongoing online consultancy
Collaborator Contribution Participants at training workshops; coordination of participation by other Netherlands public health service researchers and municipal agencies
Impact Two training workshops
Start Year 2014
 
Description Amsterdam Workshop for Public Health Netherlands 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? Yes
Geographic Reach International
Primary Audience Professional Practitioners
Results and Impact Development of knowledge and awareness of small area estimation among public health service staff in the Netherlands

Follow-on workshop
Year(s) Of Engagement Activity 2014
 
Description End of Grant Impact Workshop and Conference 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Professional Practitioners
Results and Impact Activity targeted national civil service departments and ensured policy level awareness of research

Contacts for advice from Welsh Government and DCLG
Year(s) Of Engagement Activity 2014
 
Description Kings Fund presentation 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Policymakers/politicians
Results and Impact Further contact with Public Health England

Links to follow on funding to develop small area estimates of adolescent smoking now available at www.localhealth.org.uk
Year(s) Of Engagement Activity 2014
URL https://syntheticestimation.files.wordpress.com/2014/10/kings-fund-presentation.pdf
 
Description Project Website 
Form Of Engagement Activity A magazine, newsletter or online publication
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Other academic audiences (collaborators, peers etc.)
Results and Impact Project website at http://www.synthetic-estimation.moonfruit.com - to be maintained beyond project lifespan subject to funding and staffing

Attendees at training courses are routinely referred to this website, which also acts as a repository of outputs
Year(s) Of Engagement Activity 2012,2013,2014,2015
URL http://www.synthetic-estimation.moonfruit.com
 
Description Southampton small area estimation workshop for Public Health Netherlands 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? Yes
Geographic Reach International
Primary Audience Professional Practitioners
Results and Impact Very positive feedback and possible future funding applications

None as yet
Year(s) Of Engagement Activity 2015
 
Description University of Southampton Geography Centenary Showcase 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach Regional
Primary Audience Public/other audiences
Results and Impact Poster viewed over two days by c200 members of public

None as yet
Year(s) Of Engagement Activity 2014
URL https://syntheticestimation.files.wordpress.com/2014/08/moon-et-al-poster.pdf
 
Description Using Geodemographic Indicators in Small Area Estimation - International Medical Geography Sympoisum 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Other audiences
Results and Impact Garnered initial reactions to a new paper
Year(s) Of Engagement Activity 2017