The correspondence of self-reported accidents with company records

Self-reporting is a common data collection method in accident research, Based on e.g. postal enquiry or telephone interview, it is a Fairly time-saving and cheap method. The reproducibility of this method has usually remained unverified. The aim of our study was to determine the reliability of self-reporting as a measure of accident occurrence among loggers using power saws. The self-report data were collected by a questionnaire mailed to 450 loggers, of whom 228 responded. Non-response was often because the loggers in question had retired From work, or did some other type of work not involving the use of power-saw. The study focused on accidents that had led to a report to the insurance company and which had occurred in the preceding year. All the accident reports to the insurance company concerning the subjects were then traced from the employer's records. Each of these reports was then compared with the corresponding self-report, and the Kappa coefficient was calculated as the measure of concordance. Altogether, 35 subjects reported 41 accidents and, according to the company records, 31 subjects had had 35 accidents. The reports of 95% of the subjects were concordant with the information on the accident occurrence. According to both sources the number of the common cases was 168. The Kappa coefficient was 0.81 and suggested excellent concordance, of roughly of the same level as found earlier in farmers. Despite the excellent congruence of the Kappa coefficient, the proportion of common cases should be higher for studies in which accident data are crucial. Self-reporting can be cautiously applied in studies where accident information is not of utmost importance. More studies are needed to confirm the general relevance of the result. (C) 1998 Elsevier Science Ltd. All rights reserved.