Meta-analysis of studies examining long-term construction injury rates.

BACKGROUND The construction industry is one of the employment sectors with the highest risk of injuries. AIMS To evaluate the injury trend in the construction industry from data published from 1987 to 2010. METHODS All papers with at least two measurements of injuries within a medium- to long-term period were included. The numbers of fatal and non-fatal injuries were examined in two separate groups: 100,000 workers per year and 200,000 worked hours per year. RESULTS All injuries significantly decreased between the first and the second measurement, with fatal injuries decreasing by 35% and non-fatal ones by 33% in workers/year and by 22% in worked hours/year. There was high heterogeneity among the sources of data for workers/year index (I(2) = 49% for fatal injuries, 99% for non-fatal injuries) but no heterogeneity for worked hours/year index (I(2) = 0). Meta-regression analysis showed a significant linear relationship between time and risk reduction for fatal injuries (r = 0.63; P < 0.001; a 6% reduction per year); trend reduction for non-fatal injuries was not related to the time taken between the measurements. CONCLUSIONS Fatal injuries have a reduction trend that depends on large interventions, whereas non-fatal injuries are more prone to episodic changes. Furthermore, while the workers/year index allows easier evaluation of the injury rate variation in a single working environment, the worked hours/year index is better at comparing the injury rate variation in different working environments because it reduces the sources of heterogeneity.