Cold, Hard Cash as a Handwashing Incentive

Whenever you write a book, it’s interesting to see which parts of it people respond to en masse. With SuperFreakonomics, the global-warming chapter has certainly gotten its fair share of attention, and Levitt noted a lot of feedback about the perils of drunk walking.

But there’s a third strong contender: hand hygiene, or the lack thereof, especially in hospitals. Here’s one fascinating reply from Jeffrey R. Starke, a physician in Houston:

I am in charge of infection control at Texas Children’s Hospital in Houston, the largest children’s hospital in the U.S., and am on the faculty of Baylor College of Medicine. We also had difficulty getting hand hygiene rates where we want them to be. We tried all the usual methods that you mention in the book, and a few others, but had a difficult time getting the rates above 70 percent. (We measure rates by use of a “secret shopper,” an expert in infection control from outside the institution who observes the actual hand hygiene behavior of the workers.)

We decided to use a more direct incentive. Our employees participate in a bonus program called P3. Previously, all the required performance measures were financial. However, we made hand hygiene rates part of the program; employees had to achieve and sustain >96 percent compliance with hand hygiene to get their full bonus. (This is a pooled bonus plan — either everyone gets it or no one gets it.) We did much better, but still not quite good enough until we hit on a second idea: we made the hand-hygiene performance part of the hospital executives’ performance bonus, even though they don’t care for patients. Magically, we have attained and sustained a rate of hand hygiene >98 percent, and won a national award for quality improvement from the Children’s Hospitals Corporation of America. Equally interesting is that the rate of hand hygiene among physicians, who are not hospital employees and do not participate in any performance bonus program, also has a sustained rate >98 percent. I guess this is a positive externality, perhaps pressure from employees on physicians to make everyone look good.

Starke added a note about the general use of statistics in SuperFreak:

It is a real problem in modern society that most folks don’t understand statistics at all, and they are used to scare folks all the time. (Mark Twain was right that there are lies, damned lies, and statistics.) My favorite recent example is the H1N1 influenza virus. I lecture quite a bit about this to both medical and lay groups. My standard beginning is the following: “If I told you that 99.9 percent of people who get the H1N1 influenza will suffer no significant complications, would that make you feel better?” (Almost everyone says, “Yes.”) “If I told you that 1 million people in Houston will get the H1N1 influenza, and 1,000 of them will suffer significant complications, would that make you feel better?” (Almost everyone says, “No.”) Of course, both examples are the same number expressed in different ways. These are difficult concepts for most folks to grasp, and so many “experts” take advantage of this.


They're not the same number. 99.9 percent doesn't give me a sample size.


The H1N1 examples aren't exactly the same. The percentage is the same, but the scenario is not. They are only comparable if when considering the first example, you think the contagion rate will be high enough to hit 1 million cases in Houston. That contagion rate is guaranteed in the second example--which would in itself be enough to give me pause.


The point is that both are *accurate* statistics if used in describing the same phenomenon. Including the contagion rate is a strategic decision - if you are trying to downplay the risks of H1N1, you would say "99.9% do not suffer complications". If you're trying to do the opposite, you'd mention both the 1,000,000 number, and the 1,000 number.


Brian, the percentage is the same in each scenario.

Dr. Manak

Did the hospital execs cheat? That is the play I would expect coming from the standard CEO playbook.

Jeffrey Starke

Dr. Manek, there is no cheating. The measurement of hand hygiene is blinded to all hospital workers and administrators, and is performed independently by persons from outside the institution who are unknown to the employees. The data are fed back to everyone. Our administration is very supportive of this, especially after seeing the results.

William G

I have an idea to encourage hand washing, since financial incentives are on the table:

When someone successfully washes their hands, a dollar bill is dispensed from the sink, and the person who washed their hands can immediately take it.


Right up there with telling you something increases breast cancer rates by 30%. That means from 10% lifetime risk to 13%, if you live to be 90 years old, but saying 30% makes it sound more threatening.


Hopefully you're familiar with -- it's hard to achieve hygiene success when regulatory groups actively fight against it.


@ #7...with what's found on a dollar bill, that would actually defeat the purpose.


William G: How filthy is the dollar bill? :)

Leland Witter

William G (#7) - That dollar bill had better be disinfected, since paper money is (widely presumed to be) filthy with "germs".

Mark S.

"I have an idea to encourage hand washing, since financial incentives are on the table:

When someone successfully washes their hands, a dollar bill is dispensed from the sink, and the person who washed their hands can immediately take it."

Coins would be better, they are easier to sterilize w/o destroying them. Someone even found that the high nickel content euro coins are very nearly sterile in every day use.


"A judicious man looks on statistics not to get knowledge, but to save himself from having ignorance foisted on him." Thomas Carlyle


Now if only you could apply the concept to restaurants, there would be a lot less food borne illnesses....


Number 7 is quite funny. You are giving them the dirtiest thing I can think of as a reward for washing up.

I face palmed.


Did the hand washing improve the hospital's effectiveness?


The problem with the H1N1 example is that the numbers express the situation as a positive (99.9% are Ok) and a negative (1000 suffer significant complications). It's not just a misunderstanding of the numbers that's causing the the differing responses of concern; it's the sign. It would be interesting to do an experiment to see how much ignorance of statistics and phraseology each contribute to the response.


man who washes hands after passing (sterile!) urine may be touching a tap touched by faeces contaminated finger of previous toilet user.


In almost every case, making executives part of the incentive program is the path to success. Maybe we can incentivize our elected leaders at all levels the same way?