Estimating Census Health Geographies: using synthetic estimation with secondary survey and census data ES/K003046/1 Data set one Filename: Great Britain Small Area Health Estimates Content: Small area estimates of poor self-rated health and limiting long-term illness, associated 95% Bayesian credible intervals, 2011 Census data on the same topics for comparison. Model-based estimates for England, Wales and Scotland based on linked survey data from respective national surveys. Spatial resolution: Middle Layer Super Output Areas (England and Wales), Intermediate Datazones (Scotland) Variables in data set one:(column number, variable name, content and derivation) Column 1: Country (England, Wales, Scotland) Column 2: MSOA_Name (ONS/GRoS Middle Layer Super Output Area Name (Intermediate Datazone name in Scotland)) Column 3: MSOA (ONS/GRoS Middle Layer Super Output Area Code (Intermediate Datazone code in Scotland) Column 4: CensusPSRH (2011 Census Poor Self-Rated Health %; Downloaded for England and Wales from the ONS official labour market statistics website (NOMIS) and for Scotland from the Scottish Census data warehouse website. ‘Fair’, ‘bad’ or ‘very bad’ general health was coded as Poor Self-Rated Health. Age 16+) Column 5: SAEPSRH (Small area estimate of Poor Self-Rated Health (%), erived from multilevel model of 23,374 participants (16+years) from the linked and combined 2011 waves of the Health Survey for England (n=8,603), the Scottish Health Survey (n=7,537) and the Welsh Health Survey (n=7,234). Modelled variables were age, sex (individual level), and, at the area level, disability benefits (defined as combined rate of Disability Living Allowance and attendance allowance) per 1,000 adults per MSOA, and working age benefits (defined as combined rate of incapacity benefit plus Severe Disablement Allowance plus Employment and Support Allowance benefits) per 1,000 adults per MSOA and a UK-wide Index of Deprivation (ID) score, calibrated from the English, Scottish and Welsh versions of the index. Additional fixed effect dummy variables identified the different countries. Both benefit measures were derived from the Department of Work and Pensions (DWP) neighbourhood statistics website) Column 6: SAEPSRHCIlow (Small area estimate of Poor Self-Rated Health, lower 95% Bayesian Credible Interval) Column 7: SAEPSRHCIhi (Small area estimate of Poor Self-Rated Health, upper 95% Bayesian Credible Interval) Column 8: CensusLLTI (2011 Census Limiting Long-term Illness % Downloaded for England and Wales from the ONS official labour market statistics website (NOMIS) and for Scotland from the Scottish Census data warehouse website. ‘Yes, limited a lot’ and ‘yes, limited a little’ were coded as having a LLTI. Age 16+) Column 9: SAELLTI (Small area estimate of Limiting Long-term Illness (%), derived from multilevel model of 23,374 participants (16+years) from the linked and combined 2011 waves of the Health Survey for England (n=8,603), the Scottish Health Survey (n=7,537) and the Welsh Health Survey (n=7,234). Modelled variables were age, sex (individual level), and, at the area level, disability benefits (defined as combined rate of Disability Living Allowance and attendance allowance) per 1,000 adults per MSOA, and working age benefits (defined as combined rate of incapacity benefit plus Severe Disablement Allowance plus Employment and Support Allowance benefits) per 1,000 adults per MSOA and a UK-wide Index of Deprivation (ID) score, calibrated from the English, Scottish and Welsh versions of the index. Additional fixed effect dummy variables identified the different countries. Both benefit measures were derived from the Department of Work and Pensions (DWP) neighbourhood statistics website) Column 10: SAELLTICIlow (Small area estimate of Limiting Long-term Illness, lower 95% Bayesian Credible Interval) Column 11: SEALLTICIhi (Small area estimate of Limiting Long-term Illness, upper 95% Bayesian Credible Interval) Details of underlying models are provided in the accompanying User Guide. Data Set Two Filename: Small Area Health Estimates Using Aggregated and Linked Area Data Content: Small area estimates of limiting long-term illness (severe and not severe), 2011 Census data for comparison. Model-based estimates for England based on survey data from the Crime Survey for England and Wales. This data set shows how small area estimates vary with respect to differences in the disclosure of area-level covariates. Spatial resolution: Middle Layer Super Output Areas (England only). Variables in data set:(column number, variable name, content and derivation) Column 1: MSOA (ONS Middle Layer Super Output Area Code) Column 2: SAELLTIa (Small area estimate (%) of limiting long-term illnesss derived from multilevel model of 46,597 participants (16+years) from the 2010/11 sweep of the Crime Survey for England and Wales, geocoded version available via special licence from the UK Data Service. Modelled variables were age, sex (individual level), and, at the area level, the provided deciles of the English Deprivation Index (ID). Many secondary data sets routinely attach ID quantiles) Column 3: SAELLTIb (Small area estimate (%) of limiting long-term illness derived from multilevel model of 46,597 participants (16+years) from the 2010/11 sweep of the Crime Survey for England and Wales, geocoded version available via special licence from the UK Data Service. Modelled variables were age, sex (individual level), and, at the area level, linked raw values of the English Deprivation Index (ID). Few data sets routinely disclose the spatial identifiers that enable the linkage of raw ID data) Column 4: SAELLTIc (Small area estimate (%) of limiting long-term illness derived from multilevel model of 46,597 participants (16+years) from the 2010/11 sweep of the Crime Survey for England and Wales, geocoded version available via special licence from the UK Data Service. Modelled variables were age, sex (individual level), and, at the area level, the percentage of people living in the MSOA who were unemployed, calculated as the percentage of residents unemployed from the 2011 Census, linked to the survey data via geocoding. Few data sets routinely disclose the spatial identifiers that enable the linkage of specific census data on where respondents live) Column 5: SAELLTId (Small area estimate (%) derived from multilevel model of 46,597 participants (16+years) from the 2010/11 sweep of the Crime Survey for England and Wales, geocoded version available via special licence from the UK Data Service. Modelled variables were age, sex (individual level), and, at the area level, the percentage of people living in the MSOA who were unemployed, calculated as the percentage of crime survey respondents unemployed for each MSOA. Many surveys routinely provide survey cluster indicators that can be used as a basis for aggregating survey respondents to create pseudo area indicators) Column 6: CensusLLTI (2011 Census limiting long-term illness (%), individuals (aged 16+) with an LLTI; downloaded for England and Wales from the ONS official labour market statistics website (NOMIS). ‘Yes, limited a lot’ and ‘yes, limited a little’ were coded as having a limiting long-term illness) Details of underlying models are provided in the accompanying User Guide. Data Set Three Filename: Small Area Health Estimates Comparing Approaches to Modelling Content: Small area estimates of limiting long-term illness, 2011 Census data for comparison. Model-based estimates for England and Wales based on survey data from the Crime Survey for England and Wales. This data set shows how small area estimates vary with respect to differences in the modelling of outcome variables and the parameterisation of age. Spatial resolution: Middle Layer Super Output Areas (England and Wales). Variables in data set: (column number, variable name, content and derivation) Column 1: MSOA (2011 ONS Middle Layer Super Output Area Code) Column 2: LLTIsevere (Small area estimate of limiting long-term illness (%), derived from multilevel model of 46,597 participants (16+years) from the 2010/11 sweep of the Crime Survey for England and Wales, geocoded version available via special licence from the UK Data Service. Modelled variables were banded age, sex (individual level), and, at the area level, linked raw MSOA-level values from a UK-wide Index of Deprivation (ID) score calibrated from the English, Scottish and Welsh versions of the index, combined MSOA-level receipt of Attendance Allowance and Disability Living Allowance (per 10,000 people) sourced from the UK Department of Work and Pensions, and the ONS urban-rural typology classifying MSOAs as urban, town/urban fringe, or rural (village, hamlet or isolated dwellings). Outcomes were modelled as multinomial variables; this variable reports the percentage of individuals reporting severely limiting long-term illness) Column 3 LLTInotsevere (Small area estimate of limiting long-term illness (%) derived from multilevel model of 46,597 participants (16+years) from the 2010/11 sweep of the Crime Survey for England and Wales, geocoded version available via special licence from the UK Data Service. Modelled variables were banded age, sex (individual level), and, at the area level, linked raw MSOA-level values from a UK-wide Index of Deprivation (ID) score calibrated from the English, Scottish and Welsh versions of the index, combined MSOA-level receipt of Attendance Allowance and Disability Living Allowance (per 10,000 people) sourced from the UK Department of Work and Pensions, and the ONS urban-rural typology classifying MSOAs as urban, town/urban fringe, or rural (village, hamlet or isolated dwellings). Outcomes were modelled as multinomial variables; this variable reports the percentage of individuals reporting long-term illness that is NOT severely limiting. Column 4: noLLTI (Small area estimate of limiting long-term illness (%) derived from multilevel model of 46,597 participants(16+years) from the 2010/11 sweep of the Crime Survey for England and Wales, geocoded version available via special licence from the UK Data Service. Modelled variables were banded age, sex (individual level), and, at the area level, linked raw MSOA-level values from a UK-wide Index of Deprivation (ID) score calibrated from the English, Scottish and Welsh versions of the index, combined MSOA-level receipt of Attendance Allowance and Disability Living Allowance (per 10,000 people) sourced from the UK Department of Work and Pensions, and the ONS urban-rural typology classifying MSOAs as urban, town/urban fringe, or rural (village, hamlet or isolated dwellings). Outcome calculated as 100-(variable 2+variable 3); this variable reports the percentage of individuals reporting NO limiting long-term illness) Column 5: CensusLLTIsevere (Limiting long-term illness from 2011 Census, downloaded for England and Wales from the ONS official labour market statistics website (NOMIS). Percentage aged 16+ reporting ‘Yes, limited a lot’) Column 6: CensusLLTInotsevere (Limiting long-term illness from 2011 Census, downloaded for England and Wales from the ONS official labour market statistics website (NOMIS). Percentage aged 16+ reporting ‘Yes, limited a little’) Column 7: CensusnoLLTI (Limiting long-term illness from 2011 Census, downloaded for England and Wales from the ONS official labour market statistics website (NOMIS). Percentage aged 16+ not reporting any limiting long-term illness) Column 8: LLTIsevere2 (Small area estimate of limiting long-term illness (%), derived from multilevel model of 46,597 participants (16+years) from the 2010/11 sweep of the Crime Survey for England and Wales, geocoded version available via special licence from the UK Data Service. Modelled variables were banded age, sex (individual level), and, at the area level, linked raw MSOA-level values from a UK-wide Index of Deprivation (ID) score calibrated from the English, Scottish and Welsh versions of the index, combined MSOA-level receipt of Attendance Allowance and Disability Living Allowance (per 10,000 people) sourced from the UK Department of Work and Pensions, and the ONS urban-rural typology classifying MSOAs as urban, town/urban fringe, or rural (village, hamlet or isolated dwellings). Outcome modelled as binomial contrast of severe LLTI versus no LLTI; this variable reports the percentage of individuals reporting severely limiting long-term illness) Column 9: LLTInotsevere2 (Small area estimate of limiting long-term illness (%), derived from multilevel model of 46,597 participants (16+years) from the 2010/11 sweep of the Crime Survey for England and Wales, geocoded version available via special licence from the UK Data Service. Modelled variables were banded age, sex (individual level), and, at the area level, linked raw MSOA-level values from a UK-wide Index of Deprivation (ID) score calibrated from the English, Scottish and Welsh versions of the index, combined MSOA-level receipt of Attendance Allowance and Disability Living Allowance (per 10,000 people) sourced from the UK Department of Work and Pensions, and the ONS urban-rural typology classifying MSOAs as urban, town/urban fringe, or rural (village, hamlet or isolated dwellings. Outcome modelled as binomial contrast of non-severe LLTI versus no LLTI; this variable reports the percentage of individuals reporting non-severe limiting long-term illness) Column 10: noLLTI2 (Small area estimate of limiting long-term illness (%), derived from multilevel model of 46,597 participants (16+years) from the 2010/11 sweep of the Crime Survey for England and Wales, geocoded version available via special licence from the UK Data Service. Modelled variables were banded age, sex (individual level), and, at the area level, linked raw MSOA-level values from a UK-wide Index of Deprivation (ID) score calibrated from the English, Scottish and Welsh versions of the index, combined MSOA-level receipt of Attendance Allowance and Disability Living Allowance (per 10,000 people) sourced from the UK Department of Work and Pensions, and the ONS urban-rural typology classifying MSOAs as urban, town/urban fringe, or rural (village, hamlet or isolated dwellings). Outcome calculated as 100-(variable8+variable9); this variable reports the percentage of individuals reporting no limiting long-term illness) Column 11: LLTIsevere3 (Small area estimate of limiting long-term illness (%), derived from multilevel model of 46,597 participants (16+years) from the 2010/11 sweep of the Crime Survey for England and Wales, geocoded version available via special licence from the UK Data Service. Modelled variables were banded age, sex (individual level), and, at the area level, linked raw MSOA-level values from a UK-wide Index of Deprivation (ID) score calibrated from the English, Scottish and Welsh versions of the index, combined MSOA-level receipt of Attendance Allowance and Disability Living Allowance (per 10,000 people) sourced from the UK Department of Work and Pensions, and the ONS urban-rural typology classifying MSOAs as urban, town/urban fringe, or rural (village, hamlet or isolated dwellings). Outcomes were modelled as multinomial variables; this variable reports the percentage of individuals reporting severely limiting long-term illness. The model uses an orthogonal parameterisation of the age variable to reduce the modelled parameters from 15 categorical bands to a single value. Column 12: LLTInotsevere3 (Small area estimate of limiting long-term illness (%), derived from multilevel model of 46,597 participants (16+years) from the 2010/11 sweep of the Crime Survey for England and Wales, geocoded version available via special licence from the UK Data Service. Modelled variables were banded age, sex (individual level), and, at the area level, linked raw MSOA-level values from a UK-wide Index of Deprivation (ID) score calibrated from the English, Scottish and Welsh versions of the index, combined MSOA-level receipt of Attendance Allowance and Disability Living Allowance (per 10,000 people) sourced from the UK Department of Work and Pensions, and the ONS urban-rural typology classifying MSOAs as urban, town/urban fringe, or rural (village, hamlet or isolated dwellings). Outcomes were modelled as multinomial variables; this variable reports the percentage of individuals reporting long-term illness that is not limiting. The model uses an orthogonal parameterisation of the age variable to reduce the modelled parameters from 15 categorical bands to a single value) Column 13: noLLTI3 (Small area estimate of limiting long-term illness (%), derived from multilevel model of 46,597 participants (16+years) from the 2010/11 sweep of the Crime Survey for England and Wales, geocoded version available via special licence from the UK Data Service. Modelled variables were banded age, sex (individual level), and, at the area level, linked raw MSOA-level values from a UK-wide Index of Deprivation (ID) score calibrated from the English, Scottish and Welsh versions of the index, combined MSOA-level receipt of Attendance Allowance and Disability Living Allowance (per 10,000 people) sourced from the UK Department of Work and Pensions, and the ONS urban-rural typology classifying MSOAs as urban, town/urban fringe, or rural (village, hamlet or isolated dwellings). Outcome calculated as 100-(variable 11+variable 12); this variable reports the percentage of individuals reporting no limiting long-term illness. The underlying models used an orthogonal parameterisation of the age variable to reduce the modelled parameters from 15 categorical bands to a single value. Details of underlying models are provided in the accompanying User Guide.