O42-1Occupational self coding and automatic recording (OSCAR): an innovative validated web-based tool to collect lifetime job histories in large population-based studies

• Published in: Occupational and environmental medicine (London, England) Vol. 73; no. Suppl 1; pp. A80 - A81
• Main Authors: De Matteis, Sara, Jarvis, Deborah, Young, Heather, Young, Alan, Darnton, Andy, Rushton, Lesley, Cullinan, Paul
• Format: Journal Article
• Language: English
• Published: 01.09.2016
• Subjects: Automation; Categories; Coders; Coding; Manuals; Recording; Texts; World Wide Web; British Isles
• ISSN: 1351-0711
• DOI: 10.1136/oemed-2016-103951.216

BackgroundThe standard approach for assessing lifetime individual occupational exposure is the manual collection and coding of self-reported job-histories. However this process is time-consuming and unfeasible in large population-based studies such as the UK Biobank cohort. Our aim was to develop a new valid and efficient web-based tool to collect, and automatically code, individual lifetime job-histories in the UK Biobank cohort to investigate work-related COPD.MethodsUK Biobank is a population-based cohort of 502,682 subjects, aged 40-69 years, recruited in 2006-2010. We developed an online job-questionnaire, termed OSCAR (Occupations Self Coding Automatic Recording tool), based on the hierarchical structure of the UK Standard Occupational Classification (SOC) 2000 to allow participants to self-collect and automatically code their lifetime job-histories via a decision tree model. Each participant was asked to identify each job through lists of job categories ending in a job title linked to a hidden 4-digit SOC-code. Also for each job a job title in free text was collected to allow OSCAR validation. Cohen's kappa coefficient (k) was used to estimate inter-rater agreement between SOC codes assigned by OSCAR and by an expert manual coder (gold standard).ResultsOSCAR was submitted in June-August 2015 to the 324,653 Biobank participants with an available email. Lifetime 4-digit SOC-coded job-histories for 108,784 responders were obtained (participation rate 34%). A random sample of 400 job titles in free text was used to perform OSCAR validation. The agreement between the 4-digit SOC codes assigned by OSCAR and the manual coder was moderately good (k = 0.46; 0.42-0.49). The agreement improved when considering job categories (0.70 at a 1-digit SOC-code level).ConclusionsOur web-based job questionnaire OSCAR is a new efficient and valid tool for collecting and automatically coding lifetime job histories in large population-based studies and is adaptable for use in other health research projects.