OBJECTIVE: To analyse differences in the trend in childhood bicycle-related injuries in Sweden between geographic areas - north/south and urban/rural.
METHODS: The study population consisted of children 0-14 years of age (both genders). National hospital-discharge data (ICD-9) for bicycle-related injuries were used for the period 1987 to 1996. Multivariate analysis was performed to examine linear and quadratic time trends in a statistical model.
RESULTS: There were no differences in incidence rate between the south and the north, except for children 7-14 (higher rate). There are substantial differences between urban and rural areas, and no possible methodological errors could explain this finding. But the picture that emerged from the north/south comparison is not as clear.
CONCLUSION: From methodological point of view the statistical method used with both linear and quadratic modelling of the panels gives a better understanding of the time trends than the traditional linear regression model alone. The study gives a strong support for launching preventive programmes, especially in the urban and suburban (county) areas compared to the rest of Sweden.