Estimating the population burden of injuries: a comparison of household surveys and emergency department surveillance.

BACKGROUND: Injuries represent an important public health problem but their incidence is difficult to estimate.
METHODS: We conducted a population-based household survey in Greece covering 4079 interviewed individuals. The interviewees reported, for themselves and for cohabitating adults (age 15 years and older; n = 7157), injuries that occurred during the preceding year. Major injuries were defined as those requiring contact with a health institution. We compared these survey data with data obtained through a national Emergency Department Injury Surveillance System (EDISS).
RESULTS: For the month closest to the survey interview, the incidence reported for the responders was 5.9 per 100 person-year, whereas the incidence for cohabitating adults was 3.7 per 100 person-years. These incidence rates declined for months more remote to the interview. Comparison of survey and EDISS data suggested that survey reporting was less accurate for nontraffic-related injuries. Taking into account possible recall and telescoping biases, the best survey estimate of the national annual number of major injuries is 525,000 (5.9 per 100 person-year), whereas the EDISS data yielded an estimate of 1,150,000 major injuries (12.9 per 100 person-years)
CONCLUSIONS: Comparison of survey and EDISS data systems provides quantitative assessment of accuracy of the survey data in relation to time of injury before report date, to severity of injury, and to whether the injury is to the interviewee or to a cohabitant. The 2 systems could be used in a complementary way, although EDISS generates information that is medically more accurate and is a more cost-effective data collection system. Copyright 2004 Lippincott Williams and Wilkins