Where the management of a risk involves changing behaviours, it can be very difficult to design an effective control. Behavioural change usually involves some kind of incentive control and these are notoriously hard to get right.
In the mid-nineteenth century, childbirth was frequently fatal to the mother (1 in 10), and often to the child as well. Ignatz Semmelweis was a young Hungarian doctor at the General Hospital in Vienna, and he decided to analyse the data on childbirth to try and understand what was behind such a high death-rate. He deduced that the main cause of death was infection caused by doctors not washing their hands (death-rates for midwife-only deliveries were much, much lower).
Semmelweis got doctors in his own hospital to agree to use a chlorine handwash regularly, and the death-rate fell from 10% to 1%. He then went on a mission to get doctors generally to adopt this behaviour for the good of their patients. However, as a profession, doctors were rather arrogant; they refused to believe that they were actually infecting their patients, and they managed to get Semmelweis put away in an asylum where he was physically restrained and also beaten, dying two weeks later.
In 1999, a US report found that up to 90,000 deaths a year were due to hospital error; this included patients picking up infections from poor hygiene. More recently, in an Australian exercise, doctors were asked to self-report on how frequently they washed their hands. They self-reported an average of 75% against the frequency with which they should have washed their hands; however, unbeknown to them, nurses had been asked to observe the exercise, and they reported only 9% doctor adherence to hand-washing.
One highly respected hospital in Los Angeles reported a 65% level of hand-washing and decided that needed improving significantly. They tried all sorts of incentives but they did not move the dial. Eventually, at a lunch meeting of senior medical staff, cultures were taken from their hands. The microscope photos of those cultures were then loaded as screensavers on all the hospital computers to provide an ever-present reminder of the need to wash hands. Only this worked as an incentive control.
Story sources: Super Freakonomics – Steven D. Levitt & Stephen J. Dubier; also Wikipedia.