Sample data for health analysts Welcome! ESDS Government. Samples of Anonymised Records

Transcription

1 Sample data for health analysts Welcome! ESDS Government Samples of Anonymised Records

2 Today Introduce some key data resources for health research Provide examples of how the data can be used Demonstrate different routes for accessing data

3 Introduction to Government Microdata Focussing on some key individual-level sample datasets gathered by government for policy making purposes Health Surveys Census Microdata Introduce the data and its features How to access data, resources and support Future developments Hands-on session before lunch

4 Why should you want to know about the data? Because the data are... Very cost effective: data free of charge to not for profit researchers Saves time: no need to conduct survey Access to high quality, well documented data Can provide nationally representative data - allows generalisation to population Allows historical and geographical comparisons to be made ESRC funded data support services

5 Individual cases are maintained each record contains a wide range of characteristics Microdata?

6 What are they like? Typically... Sample data Snapshot data ( cross-sectional ) Long-running series Large Flexible

7 Government Surveys Conducted by: Office for National Statistics (ONS) NatCen or similar... Collected face to face CAPI (and CASI) Household Surveys Typically exclude communal establishments May include one, all or some residents Limited range of topics

8 Health Surveys Detailed Health Information Health Survey for England from 91 Welsh Health Survey continuous from 03 Scottish Health Survey 95, 98, 03/04 Some useful indicators: general health, gp/hosp visits, smoking/drinking contraception General Household Survey from 71 (less recently) Contains some health data Omnibus Time Use Survey Labour Force Survey

9 Census microdata Wide-ranging but brief Self completion Decennial Samples of Anonymised Records available for 91 and 01 Sample size up to 5% Contains only basic health indicators

10 What data are available? Which datasets might be most suitable for: Comparing the health of unpaid carers across local authority areas Comparing obesity levels across minority groups Comparing the relationship between education and contraception type over 3 decades Looking at the impact of job type on accidents at work

11 Why use...? SARs HSE Detailed health data Basic indicators only LTLI, General Health, Caring Health focussed inc. nurse visit Sample size Up to 5% (n=3 million) Varies e.g. c. 15k full interviews in 2007 Lowest Geography Down to LAD Region/ SHA (published modelled figures for lower levels see Scholes et al 2008) Currency 2001 data available 2008 data available Other Links to other census products Easy to use Provides key indicators

12 From the census... Can be used alongside other census products Promise of confidentiality microdata is not public data Gaps due to non-response were filled using imputation (borrowing data from similar respondents) Simple random sample (approximately) Very large samples 0.6 to 3 million cases

13 Census topics 2001 On Employment Education Household structure Housing Transport Ethnicity & Religion Health & Care Not on Type of illness or disability Income

14 Health in the census Long term limiting illness in 1991 and 2001 New to 2001: Caring General health

15 What do you want? Do you want smaller than local authority geography? Area tables Range of census files Individ -ual licence d EUL Samples of Anonymised Records Special licence household SAR Small Area Microdata Individual CAM Household CAM Longitudinal Studie s - - Local authority Individual flexibility Individual detail Household members linked to each other Ability to follow individual over time Less V.G Good V.G Excellent Excellent Excellent

16 SARs Use Carers by local authority area (Buckner and Yeandle) Care homes in 1991/2001 (Banks et al 2006) High Levels of Ill health in Scotland exist once key socio-economic factors are controlled (Popham 2005) Dahlberg et. al. 2007

17 SARs Access Registration for End User Licence files Academic (online registration using federated access) Non-academic, form on site. No charge for notfor profit. Name up to 10 users Approved Research files Start at SARs site should get routed to right place Approved Researcher legal status

18 Health Survey for England Executed by the Joint Health Surveys Unit of NatCen and UCL C. 2.5k accesses in 2009 but nearly all academic Design varies to reflect topic: 2004 ethnic minority groups 2005 older people 2006 cardiovascular disease 2007 knowledge and attitudes 2008 physical activity and fitness Data collected directly from 13+ yos Regional/SHA geography Published reports most recently from NHS Information Centre

19 HSE 08: Physical Activity & Fitness More interesting design, requires weighting For main interview and nurse visit if consent obtained Main sample (max 2 households) 64% response rate. Main sample 15k adults and 3.5k children (0-15) Child boost 73% response rate n=4.5k Nurse visit for 10k adults and 2.4k children Sub-sample of c. 1/3 of sampling points Up to 1 adult (+ 1 child) per hhd asked to wear accelerometer perform step test during nurse visit

20 Scottish Health Survey 95, 98, 03/04 (08-11 in progress) Closely modelled on HSE Sample large enough to distinguish 7 regions 2003 focus on CVD: saliva, urine and ECG measures taken Health Board Geography

21 Welsh Health Survey Quite different! 1995 (not available) 1998 self completion (n=30k) WHA geography 2003 onwards continuous survey household interview and self completion in 2007, n=14k adults and 3k children general health, specific illness, NHS use, subjective lifestyle Regional geography

22 General Health Survey (GHS) aka General Lifestyle Survey (GLF) multi-purpose continuous survey since 1971 face to face interviews all adults in 13,000 addresses annually (72% response) regular health-related lifestyle topics smoking prevalence, consumption, dependency drinking - amount & frequency occasional lifestyle topics e.g. sport & leisure extensive socio-economic data household composition, education, employment, housing tenure Change to panel design in 2005

23 Not the only data! Other surveys in the ESDS Government contain health data as we ve seen. e.g. Omnibus (aka Opininions) Labour Force Survey British Social Attitudes...

24 What you can do with the data practical uses and examples Looking at change over time Look at sub-populations Using the flexibility of the data to look at alternative definitions

25 SMOKING AND SOCIAL CLASS - MEN Marmot, M (2003) % Source:HSE year all sc I&II sc IV&V Marmot, M (2003)

26 Using successive cross-sectional data over time Pros Reasonable amount of comparability Can you pool years/quarters to look at periods? Data is representative at each time point Good at looking at impacts on groups (not individuals) Cons Limits to continuity in the data (e.g. ethnicity, SOC) Cannot establish individual change

27 Small populations are often too small Survey data is subject to sampling error! Example: Pregnancy and Employment Using General Household Survey data alone there are only 168 pregnant women aged % Confidence interval for % pregnant women economically inactive % Combined 3 years data to obtain sample of 465 pregnant women Confidence interval using 3 years data: %

28 Using the flexibility of the data to look at alternative definitions What are hours worked? Is it just paid work? Or unpaid as well? Hours usually worked, or actually worked last week? In main job, or in any job? What about students? Overtime paid? Overtime unpaid? Lunch hours? Do non-workers work zero hours or should they be excluded?

29 Limitations Cannot attempt to identify respondents so not possible to do direct matching under End User Licence matching on known area is OK probabilistic approaches OK PCT geography not available region lowest level on most data some LAD geography Not public data Must register

30 Accessing Data through the ESDS All users can access study descriptions, online documentation, including questionnaires, free of charge without registering with ESDS In order to access the datasets you need to register with ESDS Register online using your UK Federation username and password Simple online form, takes about 10 minutes You need to register a usage of the data as part of this process Non-commercial users: free of charge Commercial users: 500 Charge for per study and will need to apply for a UK Federation username and password You need to agree to the End User Licence when you register

31 Longitudinal Census longitudinal studies: LS, SLS, NILS See ESDS longitudinal studies e.g.: cohort studies NCDS, BCS70, MCS British Household Panel Studies See English Longitudinal Study of Ageing

32 and more... Drug use, Smoking and Drinking among Young People in England Psychiatric Morbidity Surveys National Survey of NHS Patients Adult Dental Health Infant Feeding Survey National Diet & Nutrition Survey National Survey of Sexual Attitudes & Lifestyles (NATSAL) Active People Survey Taking Part Survey...

33 What data is available for my topic? Census.ac.uk has a search portal for data SARs team about to publish a health topic guide ESDS Government health theme guide survey pages survey finder omnibus module pages variables search publications and registered usage pages nesstar ESDS Data Catalogue ESDS theme page on health: Survey Question Bank (Survey Resources Network)

34 Finding out about the data! Unless... you can track your variable back to the question(s) asked on the questionnaire, know who the questions were asked of and what was done with the raw data to turn it into the final data and what you must do to be able to use it You don t understand the data

35 Documentation Questionnaires Codebooks Information about derived variables Technical information on sampling, weighting etc.

36

37

38

39 Future developments Census 2011 Range of new questions on identity and language Less on socio-economic characteristics Northern Ireland Scotland information on specific disabilities ONS Integrated Household Survey GHS, Labour Force Survey, Expenditure & Food Survey, English Housing Survey, Omnibus Surveys, Life Opportunities Survey Approx 500k individuals in c. 250k households Core questionnaire Economic Activity Education Health & Disability Identity Income PCT level estimates? Outputs expected IC Survey Consultation - a new Health and Social Care Survey Fieldwork Jan 2011, reporting late 2012 Join IC s Health Surveys Programme Network surveys.queries@ic.nhs.uk

40 Help and Support SARs Annual User Meeting ESDS Government GHS/GLF and Health Surveys Annual User Meetings