Estimates of home and leisure injuries treated in emergency departments in the adult population living in metropolitan France: a model-assisted approach

Abstract

Background: Home and leisure injuries (HLIs) are currently a major public health concern, because of their frequency, associated consequences, and considerable medical costs. As in many other countries in Europe, in France the population coverage of the surveillance system of HLIs is low. In this study, a model-assisted approach is developed to estimate the incidence rates of HLIs in adults treated in emergency departments (EDs) in metropolitan France between 2004 and 2008.

Methods: Using a sample of the hospitals participating in the French ED-based surveillance system, a generalized linear mixed model was applied, which describes the relationship between the numbers of ED visits for HLIs and the sex and age of the patients on the basis of the number of injury-related stays recorded by the hospitals. Statistics on hospital stays were provided by the French hospital discharge databases in the participating hospitals. The same statistics were available at the national level, which made it possible to extrapolate national incidence estimates.

Results: Over the 2004-2008 period, the estimated incidence rate of HLIs age-standardized on the European population aged 15 years and over was 48.7 per 1,000 person-years (95% confidence interval: 39.4-58.0), and displayed little variability over time. This rate corresponded to an average of 2.5 million emergency hospital visits each year due to an HLI in people aged over 15 in France.

Conclusions: The method made it possible to use medico-administrative datasets available nationwide to provide informative estimates despite the small number of participating EDs. The consequences and costs generated by hospital emergency visits can sometimes be onerous, and these estimated rates confirm the scale of the problem and the need to continue investing in preventive actions.

Figure 1

Geographical distribution of the sample of participating hospitals from 2004 to 2008 in the EPAC network used to estimate the rate of HLIs treated in EDs in metropolitan France. The area of the spot is proportional to the number of stays for injuries (aged over 15, reference area = Blaye).

Figure 2

Log-scaled numbers of HLI versus log-scaled numbers of hospital stays observed in the eight hospitals of the EPAC network for each combination of sex and age category from 2004 to 2008, according to algorithm 1 (ICD-10 chapter 19 diagnostic codes) and algorithm 2 (extended selection).

Figure 3

Estimates and 95% confidence intervals of the number of HLIs treated in EDs in metropolitan France for each year from 2004 to 2008, according to age category and sex. (Points are drawn at the center of five-year age categories).

Figure 4

Estimates and 95% confidence intervals of metropolitan age-specific incidence rates of HLIs treated in emergency departments from 2004 to 2008, for women and men. Horizontal segments represent age-specific rates for the period 2004–2008 (solid line: estimates, dotted line: 95% confidence intervals).

Figure 5

Forest plots of estimates of random effects (sigma1 = σ₁ (hospital)), sigma2 = σ₂ (sex-age category within hospital level), auto-correlation (rho = ρ (lag1)), total numbers of HLIs over the 2004–2008 period, and related rates with 95% confidence intervals, when excluding the hospital named in the first column. The vertical line indicates the central estimates for the full sample of hospitals (last row).