Comparing 90% to 5% would be very silly. What percentage of cancer-related deaths are attributable to tobacco use? (In the same sense of "attributable" as for the alcohol?)