Secular trends in Swedish hip fractures 1987-2002: birth cohort and period effects

Recently, a leveling off in hip-fracture incidence has been reported in several settings, but the annual number is nonetheless predicted to increase due to the growing elderly population.

Using Swedish national data for 1987-2002 for all inpatients 50 years or older, we examined the annual number and incidence of hip fractures and explored age, period, and cohort effects. Age adjustment was done by direct standardization, time-trend analysis by linear regression, changes in linear trends by joinpoint regression, and age-period-cohort effects by log-likelihood estimates in Poisson regression models.

Before 1996, the age-standardized hip fracture incidence was stable (0.1% per year [95% confidence interval = -0.2% to 0.5%]), and the annual number of hip fractures increased (2.1% per year [1.8% to 2.4%]). After 1996, both the age-standardized hip fracture incidence (-2.2% per year [-2.8% to -1.6%]) and the number of hip fractures (-0.9% per year [-1.5% to -0.4%]) decreased. The period + cohort effects were more marked among women than men, with a major reduction in hip fracture incidence in subsequent birth cohorts (estimated incidence rate ratio = 2.2 comparing women born 1889-1896 with 1945-1952) or periods (estimated incidence rate ratio = 1.1 comparing women living 1987-1990 with 1999-2002).

The age-standardized hip fracture incidence has decreased since 1996, more than counteracting the effects of the aging population and resulting in a decline in the annual number of hip fractures through 2002. The magnitude of the combined period and cohort effects in women seems to be of biologic importance. If this persists into older age, the annual number of hip fractures will be lower than has been projected.