Swedish Longitudinal Occupational Survey of Health (SLOSH)

Summary

With the exception of the Finish Public Sector study (FPSS), SLOSH is the only large-scale occupational cohort in the Nordic countries which follows a nationally representative sample of individuals with frequent repeat measures over an extended period of their working career. In contrast to FPSS, SLOSH includes also private and central government employees as well as self-employed, and has more frequent measurements. This makes the study uniquely suited for analyses of how the labour market as a whole impacts on health and well-being. In addition it enables analyses in groups which are rarely studied, like private sector employees. With its unique data SLOSH allows for a better under¬standing of how working life and well-being are linked, what the causal mechanisms might be, and how cumulative exposures to psychosocial factors are associated with disease risk. This makes it possible to study long-term outcomes such as drug therapy, hospitalisation, and cause-specific death. Furthermore, it allows researchers to study (and adjust for) a range of objectively measured exposures, risk factors and confounders, not the least in working life. By combining such registry data with self-reports, researchers can study the development of disease over long time periods and relate this development to work-related, socioeconomic, psychosocial and physical exposures, as well as to health behaviours. In addition, as the cohort is ageing, it will provide important data for the understanding of the timing of labour market exit and of how working life and retirement impact on health and well-being in early old age. In order to utilise and strengthen these advantages an effective strategy for data utilisation is needed. As with all cohort studies SLOSH is less well suited to study secular as opposed to intra-individual changes. Another weakness is that drop-out threatens the generalizability of the research. However this is not unique to SLOSH but common to all longitudinal studies.

Type of data

Data Source
Registry + Survey

Type of Study
Survey same

Data gathering method
Registries
Self administered questionnaire

Access to data

Conditions of access
Data is available for the scientific community in line with Swedish laws and agreements with Statistics Sweden.

Type of available data (e.g. anonymised microdata, aggregated tables, etc.)
Anonymised microdata.

Formats available
Normally SAS, STATA, SPSS, but other formats could be provided if necessary

Coverage

Coverage Years of collection, reference years, and sample sizes
Wave 1: Data was collected in 2006 and included a sample of 9,214 individuals. Wave 2: Data was collected in 2008 and included a sample of 18,915 individuals. Wave 3: Data was collected in 2010 and included a sample of 17,738 individuals. Wave 4: Data was collected in 2012 and included a sample of 17,409 individuals. The main representative cohort is comprised of 18,915 individuals. In 2010, a boost sample from those living in the greater Stockholm or Gothenburg area was added. Thus, the overall SLOSH contains information on 21,489 respondents.

First year of collection
Data was first collected in 2006. The baseline data from the Swedish Work Environment surveys are available for 2003 and 2005. Register data is available retrospectively in some cases back until 1967.

Stratification if applicable
The SLOSH sample consists of all respondents to the Swedish Work Environment Surveys (SWES) 2003 (n=9,212) and 2005 (n=9,703), forming the main representative cohort of 18,915 individuals, plus those participants from the 2007 SWES living in Stockholm or Gothenburg areas (n=2,572). SWES is conducted biennially by Statistics Sweden (SCB) and consists of subsamples of gainfully employed people, aged 16-64 years, from the Labor Force Survey (LSF). These individuals were first sampled into the LFS through stratification by country of birth, sex, citizenship, and occupation.

Base used for sampling

Geographical coverage and breakdowns
Sweden. At present county level is the smallest geographical area for analyses.

Age range
19-74 (due to the fact that first basic sample from 16 – 64 has been older during the longitudinal process)

Statistical representativeness
Population representative

Coverage of main and cross-cutting topics
The primary focus of SLOSH is on a wide range of prospective longitudinal research questions about the relationships between labour market participation, working life, social environment, personal agency, and health. Further, the purpose of the database is to better understand the aetiology of illnesses and functional limitations of public health relevance, and to increase the understanding of ‘the causes behind the causes’. This includes studying socioeconomic factors, occupational exposures, and health behaviours, as well as factors in private life. SLOSH respondents choose from two questionnaire versions, one for those currently in work, and one for those who are not in gainful employment. The former contains questions about work, health, social factors outside work, and health behaviours. The non-worker version replaces the work section with questions about the re-spondents’ current life situation (retired, unemployed, studying etc.), e.g. the reasons for not working, positive and negative aspects of not working, and, where applicable, how rehabilitation is experienced.

Linkage

Standardisation
SLOSH uses several well-established international validated scales. Moreover ISCO and SEI are available.

Possibility of linkage among databases
Register data regarding demographic information, income from work and old-age pension, receipt of disability pension, unemployment and social benefits, main workplace (including information about staffing and finances), hospitalisation, sickness absence, purchases of prescribed drugs, cancer and CVD cases, and deaths, are linked both prospectively and retrospectively to all respondents. All data are retrieved using personal identification numbers thorough a key file stored at SCB and delivered and stored with personal identification numbers replaced by serial ID numbers. The key is stored only by Statistics Sweden, ensuring that the researchers cannot connect the data to specific persons without their explicit permission.

Data quality

Entry errors if applicable
Internal attrition differs considerably between the two versions of the questionnaire, i.e. those in gainful employment and those who are not, but also between items. For many items, internal attrition is between zero and 5 per cent, but for several items, it is as high as nearly 70%. This seems, however, not based on applicability of those questions, thus participants didn’t answer and a no-alternative was not available. The questionnaires are scanned by SCB and consistency is checked ensuring good quality of the data. Register data are of high quality.

Breaks
SLOSH was initiated by the Stress Research Institute in 2006. Prof. Hugo Westerlund took over as PI for SLOSH from Prof. emeritus Töres Theorell on 1 January 2012 after having been co-PI since the inception of the study. SLOSH is run by a steering committee, which consists of among others Hugo Westerlund as PI and chairman, Ass. Prof. Constanze Leineweber (data manager), Dr. Linda L. Magnusson Hanson (SLOSH study director), and Dr. Martin Hyde (deputy head of unit). All these researchers have been involved in SLOSH since 2007 or earlier. Depending on different collaborations, research question, etc., some scales were omitted, other were added and others were changed during the years. However, there are no dramatic changes regarding the content of the questionnaires. Overall, there is a fairly stable core set of items that cover the main research topics (outlined above), which are repeated in each of the waves.

Consistency of terminology or coding used during collection
The complexity of the database results in some inconsistencies, including different formats for the same data from different years, different original variable names, as well as different coding systems across years (e.g. for education and diagnoses). However, we are continuously working to build up a library of SAS scripts which merge original data and transform them into a standard form with standardised names according to our uniform naming convention. All variables are given short names which are evocative of their meaning in English (e.g. ‘workfast’ for an item asking ‘Do you have to work very fast?’) followed by an underscore and the number of the wave (thus ‘workfast_3’ denotes the responses to this item in the 2010 SLOSH survey). In addition to being convenient, this convention facilitates the transformation of data between ‘wide’ and ‘long’ formats.

Governance

Contact information
Hugo Westerlund
Stress Research Institute
Frescati Hagväg. 16A
106 91 Stockholm Sweden Phone: +46-8-159 868
Email: hugo.westerlund(at)stress.su.se
Url: http://www.stressforskning.su.se/slosh-studien/slosh-start

Timeliness, transparency
The initial data is typically available in the technical reports provided by Statistics Sweden within a couple of months after the finalised data collection.