Can a smaller than expected crown–rump length reliably predict the occurrence of subsequent miscarriage in a viable first trimester pregnancy?