Creating a postcode history from medical sources for longitudinal analyses
Everington, D., Huang, Z. & Feng, Z. (2017) UK Administrative Data Research Network Annual Research Conference, Royal College of Surgeons, Edinburgh, UK, 1 - 2 June 2017 [SLS]
Other information:
Abstract:
The Scottish Longitudinal Study (SLS) is a large-scale linkage study created using data from administrative and statistical sources. These include: census data from 1991 onwards; vital events data (births, deaths, marriages); NHS Central Register data (migration into or out of Scotland); and education data (Schools Census and SQA data).
There are many advantages to using these data: they are a large, representative sample of the Scottish population with a low attrition rate. The SLS includes a range of variables describing demographic, economic, health, education, cultural, housing, social and ecological data. Our sample is further linked to ISD health data including cancer registrations, hospital admissions etc. One of the main disadvantages of this study is that although the vital event, education and health data can be regularly updated, the Census variables are only known at the 10 year time points. Trying to determine cause and effect without full histories is clearly more difficult.
To address this, NHS Scotland have provided postcode data obtained from GP registrations and other health records since 2000. Although the data will not be provided to users in their fine geographical details, as far as we know, this is the first time that these data are available for longitudinal analyses on a small area scale. These data will be of particular interest to researchers wishing to study migration. They will also allow area effects such as deprivation and urban/rural classification to be looked at over time.
This presentation will describe the characteristics, difficulties and processing of these data. The data are validated by comparison to the enumeration postcodes in 2001 and 2011 which are highly accurate. Future analyses will investigate how the match rate varies by other characteristics such as age, gender, economic activity and geographical area.
Available online: Link