475 Fatal occupational injuries in Norway: surveillance data are biassed and underestimated risk

Background The Norwegian Labour Inspection Authority (NLIA) compiles and publishes statistics on fatal occupational injuries. Other institutions also register such information on a national level: Statistics Norway (SN) (from Cause of Death Registry (CDR), recently transferred to the Norwegian Institute of Public Health), the National Insurance Administration (NIA) and Finance Norway (FN) (from private insurance companies). The aim of this study was to examine completeness and quality of NLIA statistics, and see if use of additional sources could improve surveillance of risk. Methods Residents in Norway have a unique personal identification numbers. This was used to compare cases of death from occupational injuries 2000–2003 registered in NLIA, NIA and FN – with information in CDR. Results NLIA had registered 171 deaths from occupational injuries 2000–2003. 75 more deaths were identified from the three other sources. Of all the 246 deaths, NLIA had information on 171 (70%), NIA 158 (64%), SN 141 (57%) and FN 50 (20%). NLIA was most complete, but completeness varied between industries, from 24% for Public administration and defence, compulsory social security to 81% for Construction (Standard Industrial Classification (SIC2002)). Completeness also varied according to external cause of death, and was particularily low (32%) for transport accidents with car (ICD-10 V4). All 246 deaths were found in CDR, but due to incomplete information in death certificates, only 57% were identified as occupational injuries. Conclusions The NLIA registry was most complete, but biassed, and grossly underestimated risk in some subgroups. Administrative changes could improve completeness and quality and make it a better tool for surveillance of risk. The CDR is potentially the most complete source of information. If death certificates for accidental deaths have incomplete data, the certifier could more often be asked to provide missing information. Validation by casewise comparison between NLIA and CDR with regular intervals is recommended.