Spring 2015 Coast to Coast Seminar Series: How do I love data? Let me count the ways… Using linked databases for work and health research
You may experience this event in one of the following ways:
- In person at the University of British Columbia (Point Grey Campus), Earth Sciences Building, 2207 Main Mall, Room 4127.
- Via videoconference: find location nearest to you
- Via live webstream: access link will be provided upon registration
Dr. Mieke Koehoorn
School of Population and Public Health, University of British Columbia
The key focus of my program of research is the use of workers’ compensation claims data, linked to other databases, for policy-relevant research on work and health. As co-lead of the Partnership for Work, Health and Safety at the University of British Columbia and professor in the university's School of Population and Public Health, I will share examples of the contribution of research to informing workers' compensation policy and programs in British Columbia, Canada. Research examples include the surveillance of occupational injuries and cancers, the occupational epidemiology of cardiovascular disease among emergency workers, and the evaluation of compensation policies and procedures, including the effect of expedited surgical fees and private surgical clinics on work disability duration and the effect of faller certification on injury rates in the forestry sector. I also hold a research chair in gender, work and health and will discuss the importance of integrating sex and gender based analysis in workers’ compensation research. Finally, the talk will discuss the potential use of the linked health data for comparative research across compensation systems in Canada and with other jurisdictions.
Data sources used
Workers’ compensation claims data, linked to medical services, hospitalizations, cancer registry, vital statistics and other ‘external’ work-related databases.
Strengths and caveats associated with data use
Key strengths are the rich and longitudinal nature of the workers’ compensation data for evaluation of programs and policies on work injury, illness and disability outcome over time in British Columbia, as well as the ability to link the claims data to other data sources for adjustment of key variables such as co-morbidity; caveats are limited employment and work variables (as major determinants of health) at the population-level, beyond claims data for injured and ill workers, for population-based workforce studies and comparisons.
Analytic techniques used
Various different statistical models depending upon the outcome and research question, including logistic and linear regression, quantile regression, random effects Poisson regression, discrete time survival analysis.
Policy implications/key findings
Administrative data matters for having a policy or program impact, in particular in providing population-based evidence in a local context that resonates with the decision maker (in my case, with decision makers in the workers’ compensation system); and supporting public access models to use this data for research purposes is the way forward;
For research evidence to inform and impact policies and programs the researcher-policymaker relationship needs to be a long-term, committed one with a long-term view of the impact of research evidence (i.e. research evidence needs to ‘percolate’ into decision making).