Research results indicate a wide variation in belt use effectiveness (reduction of the number of casualties in percents of belt use increases from 0% to 100%). Results are reported from an increase to a decrease of more thut 50%.
Accepting some variation in different countries due to different accident circumstances, weakness in research designs could form a part of the explanation of this variation.
Analyses have been made of trends in fatality rates of car occupants in three European countries; the Netherlands, the Federal Republic of Germany and Creat Britain. The aim wes to assess seat belt use effectiveness using common, plausible assumptions.
It turned out that is not recommendable to use the before and after comparison, because it is hard to assume that there are no other influences than changes in seat belt use. Before and after comparison with a control group is a methodologically seen strong design but its application could be unnoticed weak, despite "plausible" assumptions. So a dangerous design. The use of time series analysis is recommended but the nature of the seat belt intervention and other influences and their course in time have to be known. Specific and careful, research and good data are needed.
On the basis of the actual knowledge we think that the best estimation of belt use effectiveness is 35-40Z%.