Linking lives through time SCOTTISH LONGITUDINAL STUDY (SLS)

linking lives through time www.lscs.ac.uk linking lives through time www.lscs.ac.uk

SCOTTISH LONGITUDINAL STUDY(SLS)

linking lives through time www.lscs.ac.uk

Why do we need the SLS?• overall mortality rates are higher than the rest

of Britain • fertility rates are lower than the rest of Britain• population ageing is a significant problem for

the future in Scotland• specific causes of death such as lung cancer

and heart disease rank among the worst in Europe

• rates of household deprivation are higher than in the rest of Britain

• These issues, and the inter-relationships between them, are examples of what can be explored using the SLS


Muller-Nordhorn, J. et al. Eur Heart J 2008 0:ehm604v1-11; doi:10.1093/eurheartj/ehm604

Age-standardized mortality from ischaemic heart disease in European regions (men; age group 45-74 years; year 2000)


What is the SLS?• SLS is a replica of the LS (Longitudinal Study of

England and Wales), which started in 1971• The SLS is a large-scale, anonymised linkage study

designed to capture 5.5% of the Scottish population• Sample based on 20 semi-random birthdates• It actually contains information on a 5.3% sample• It is built using data available from…– Census data (initial sample from 1991, then 2001)– Vital Events data (births, deaths, marriages etc…)– National Health Service Central Register (NHSCR) data

(migration into or out of Scotland)– NHS data (cancer registrations, hospital episodes and

maternity data)– Education data


Study structure

SLS database

1991 original sample:

270,385 selected from 1991 Census

plus household members Entries since 1991:

Births 53k, Immigrants 31k

Exits of sample members from SLS:

deaths 48k, emigrants 20k

Events to sample members since their entry to SLS:

Births, widow(er)hoods, marriages, stillbirths, infant mortality (139k in total)

2001 Census sample

265,104 sample members found at 2001

plus household members

Coming soon:

School Census records, inc attendance/exclusions, attainment, qualifications

ISD health data

Project specific extracts of:Hospital admissionsCancer registrationsMaternity data




SLS is one of three UK LS

LS England and Wales (LS)• 1% sampe

Northern Ireland LS (NILS)• 28% sample

Northern Ireland mortality study (NIMS)• 100%


SLS project examples


SLS project - Example 1

Assessing socio-economic inequalities in mortality at the Scottish national level, incorporating a comparison between mortality in Scotland and England

Frank Popham – University of St Andrews Paul Boyle - University of St Andrews

SLS project no. 2009_004– www.lscs.ac.uk/sls/projects/2009_004.htm


Is there a “Scottish effect” for mortality? Scottish effect = excess Scottish mortality

over England after deprivation differences

Found in unlinked ecological data but what about linked data joining SLS and ONS LS?

Popham, F and Boyle, P: Journal of Public Health doi:10.1093/pubmed/fdr023


Source: Mark Hillary

What about those born in Scotland now living in England?


.81

1.2

1.4

Rel

ativ

e m

ort

alit

y ra

te (

log

scal

e w

ith 9

5% C

Is)

Born Eng, live E&WBorn Eng, live Scot

Born Scot, live E&WBorn Scot, live Scot

Before adjusting for SEP differences After adjusting for SEP differences

Source: Scottish Longitudinal Study and ONS Longitudinal Study

Impact on relative mortality rate 2001-2007 of adjusting for socio-economic differences


Is there a “Scottish effect” for mortality? Scottish effect = excess Scottish mortality over

England after deprivation differences

Found in unlinked ecological data but what about linked data joining SLS and ONS LS?

Additionally account for country of birth as Scots born have high mortality generally

Confirms there appears to be a Scottish effect associated with born Scottish• However, unmeasured confounding still possible

explanation• Socio-economic circumstances in childhood

important?

Popham, F and Boyle, P: Journal of Public Health doi:10.1093/pubmed/fdr023


SLS project - Example 2

Does being widowed increase the risk of death?

Paul Boyle - University of St Andrews Zhiqiang Feng - University of St Andrews Gillian Raab - University of St Andrews

SLS project no. 2008_006– www.lscs.ac.uk/sls/projects/2008_006.htm


Research OverviewBackground Does widowhood increases mortality risk? Although commonly observed, this ‘widowhood

effect’ could be due to selection effects as married couples share various characteristics related to the risk of death.

The widowhood effect by different causes of spousal death; some are correlated with these shared characteristics, while others are not.

Methods Using data from the SLS outcomes for men and

women by different causes of death of their spouse, controlling for a range of individual- and household-level characteristics were compared.


Effect of widowhood on the mortality hazard ratio in a Cox model, compared with the effect of age at the 1991 census, with age 50 as the baseline, adjusted for socio-economic, LLTI and ethnicity.


Results The widowhood effect is greater than has been found

from other recent studies, especially for older women. The risk is highest shortly after widowhood, but

remains significant for over ten years. These broad results hold regardless of the cause of

death of the spouse.

Conclusions The analysis of the widowhood effect used three

methods of classifying the spouse’s cause of death in an attempt to control for potential selection effects.

The results are highly consistent and suggest that this is a causal effect, rather than a result of selection.

Boyle, P, Feng, Z. & Raab, G, (2011) Does widowhood increase mortality risk? Comparing different causes of spousal death to test for selection effects. Epidemiology, 22 1-5


SLS Future developments


School Education data – 3 datasets:

Pilot linkage exercise carried out using 2007 & 2008 data. 2009 linkage to be carried out summer 2011.

Names not available so linkage carried out on DOB, sex and postcode only - >80% linkage success

Data to be provided annually, and successive linkages will increase proportion of linked pupils (since each has up to 13 chances of being linked)

- School Census data- SQA attainment data - Attendance/ absence & exclusions

data available


2011 Census – Questions added

5 new questions included:• national identity • ability in spoken English • languages other than English used at home • long term health conditions • month/year of arrival into the UK (for people not

born in the UK)


Summary: strengths of the SLS Sample size much larger than most surveys

BHPS started with ~10,000 people in GB SLS has ±270 thousand traced members + ±505 thousand

household members in Scotland (1991) This means that it’s possible to analyse relatively small sub

populations Covers several fields of interest – demography, economic

and social issues, health, housing, mortality, migration, fertility etc

Can be linked with detailed geographical information Census compulsory so attrition rates low and linkage rates

high Includes those in communal establishments Ability to link hospital episodes data to socio-economic

characteristics Potential to link to other datasets in the future – (eg 2011

Census, HESA)


Summary: weaknesses of the SLS

Restricted range of variables • Smoking• Income

Census information only collected every decade Not possible to return to the sample to ask

extra questions The data are highly confidential Immigration and emigration data not very

reliable We rely on census data - will there be a 2021

Census? Difficult dataset to analyse, both technically,

and logistically!


The SLS team – contact detailsBased in Edinburgh: Based in LSCS, St

Andrews:

Joan Nolan Database Administrator0131 314 [email protected]

Zengyi HuangProgrammer0131 314 4253 [email protected]

Claire BoagSLS Manager0131 314 [email protected]

Gillian RaabStatistician 0131 314 [email protected]

Christine Lamb Chris Dibben Administrator SLS Director

SLS Support Officers

Zhiqiang Feng Peteke FeijtenPlus:Frank Popham & Lee Williamson

Interested in using the SLS?Contact:[email protected]



Information on datasets


Publication Hub


Downloads Page (includes relevant legislation)


PDF Researcher packs


Guidance


Registration



Living circumstances

0.0%

10.0%

20.0%

30.0%

40.0%

50.0%

60.0%

70.0%

80.0%

90.0%

100.0%

15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30

alone

husband kids (parents others)

husband (parents others)

partner kids (parents others)

partner (parents others)

kids (parents others)

I just kids

parents (others)

just others

missing

• Young women no births.

• Picture for young men looks very similar see next slide

• Over ages 15 to 30 gradual move away from parents to live alone or with husband or partner

• Others is an intermediate state


Living circumstances at 2001

• Young men no births.

• Picture very similar to that for young women

0.0%

10.0%

20.0%

30.0%

40.0%

50.0%

60.0%

70.0%

80.0%

90.0%

100.0%

15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30

alone

wife kids (parents others)

wife (parents others)

partner kids (parents others)

partner (parents others)

kids (parents others)

I just kids

parents (others)

just others

missing

0.0%

10.0%

20.0%

30.0%

40.0%

50.0%

60.0%

70.0%

80.0%

90.0%

100.0%

17 18 19 20 21 22 23 24 25 26 27 28 29

0.0%

10.0%

20.0%

30.0%

40.0%

50.0%

60.0%

70.0%

80.0%

90.0%

100.0%

17 18 19 20 21 22 23 24 25 26 27 28 29

Teen mums Teen dads


Economic position at 2001

Linking lives through time SCOTTISH LONGITUDINAL STUDY (SLS)

Documents

linking lives

scottish effect

sls outcomes

1093pubmedfdr023 slide

excess scottish mortality

sls project example

sls project examples

st andrews sls project