Objectives A number of data sources help inform policy decisions regarding the risk of work-related ill-health and sickness absence. This study aimed to compare self-reported and medically reported data from a range of sources to'triangulate' and validate estimates of the incidence of work-related ill-health in the UK.

Methods Sources consisted of the The Health & Occupation Reporting network (THOR) and the survey of Self-reported Work-related Illness (SWI). Results from SWI and of THOR from GPs, rheumatologists, psychiatrists, dermatologists and respiratory physicians (2006–2009) were compared. THOR-GP data also included patient referrals from primary to secondary care.

Results THOR-GP (85%) and SWI (76%) reports were mainly musculoskeletal and mental ill-health. Overall, SWI incidence rates were greater than GPs'; however, these varied by diagnoses. SWI rates were higher for mental ill-health (GP 500, SWI 790 per 100 000 persons employed) and ‘other’ diagnoses (GP 42, SWI 368) whereas rates for musculoskeletal (GP 684, SWI 670) and skin diagnoses (GP 153, SWI 38) were higher from GPs. Very few cases of musculoskeletal and mental ill-health were referred to clinical specialists (<1%). Skin (15%) and respiratory (29%) cases were referred more frequently.

Conclusions SWI data generally capture more cases than THOR-GP; however these are unsubstantiated by a medical practitioner. Clinical specialist reports are subject to biases, for example, severity and referral patterns. GP data may give a better reflection of the incidence of work-related disease such as dermatitis. Triangulation of these data sources can help to obtain more reliable estimates of work-related ill-health.