"To survey or to register" is that the question for estimating population incidence of injuries?

Measuring the true incidence of injury or medically attended injury is challenging. Population surveys, despite problems with recall and selection bias, remain the only source of information for injury incidence calculation in many countries. Emergency department (ED) registry based data provide an alternative source.The aim of this study is to compare the yearly incidence of hospital treated Home and Leisure Injuries (HLI), and Road Traffic Injuries (RTI) estimated by survey-based and register-based methods and combine information from both sources in to a comprehensive injury burden pyramide.

Data from Luxemburg's European Health Examination Survey (EHES-LUX), European Health Interview Survey (EHIS) and ED surveillance system Injury Data Base (IDB) collected in 2013, were used. EHES-LUX data on 1529 residents 25-64 years old, were collected between February 2013-January 2015. EHIS data on 4004 other residents aged 15+ years old, were collected between February and December 2014. Participants reported last year's injuries at home, leisure and traffic and treatment received. Two-sided exact binomial tests were used to compare incidences from registry with the incidences of each survey by age group and prevention domain. Data from surveys and register were combined to build an RTI and HLI burden pyramide for the 25-64 years old. This project was part of the European Union project BRIDGE-Health (BRidging Information and Data Generation for Evidence-based Health Policy and Research).

Among 25-64 years old the incidence of hospital treated injuries per thousand population was 60.1 (95% CI: 59.2-60.9) according to IDB, 62.1 (95% CI: 50.6-75.4) according to EHES-LUX and 53.2 (95% CI: 45.0-62.4) according to EHIS. The incidence of hospital admissions was 3.7 (95% CI: 3.5-4.0) per thousand population from IDB-Luxembourg, 12.4 (95% CI: 7.5-19.3) from EHES-LUX and 18.0 (95% CI: 13.3-23.8) from EHIS. For 15+ years-old incidence of hospital treated HLI was 62.8 (95% CI: 62.1-63.5) per thousand population according to IDB whereas the corresponding EHIS estimate was lower at 46.9 (95% CI: 40.4-54.0). About half of HLI and RTI of the 25-64 years old were treated in hospital.

The overall incidence estimate of hospital treated injuries from both methods does not differ for the 25-64 years old. Surveys overestimate the number of hospital admissions, probably due to memory bias. For people aged 15+ years, the survey estimate is lower than the register estimate for hospital treated HLI injuries, probably due to selection and recall biases. ED based registry data is to be preferred as single source for estimating the incidence of hospital treated injuries in all age groups.