Friday, January 18, 2008

Blunder

Not long ago, a survey in Massachusetts General Hospital told us that nearly half of doctors admitted to witnessing a serious medical error but not reporting it. According to another survey published this week, most doctors in three United States teaching hospitals agreed that reporting errors improves quality of care, although only a little over half said they understood how to do so; only 18% said they had actually reported a minor error and 4% a major error.

How can this be, I muse, that reporting medical errors lead to improved quality of care? As appealing as such contention seems, I am loath to suggest that reporting medical errors is the reason for improved medical care. This reminds me of a project Early Childhood Longitudinal Study that I learned from the book Freakonomics. This monumental project was undertaken in the late 1990s when more than twenty thousand children from kindergarten through the fifth grade were followed up in the United States, with measurement of the students' academic performance. As it turns out, a spanked child is associated with better test scores. Taken at face value, this would have appeared to advocate parenting with the dogma "spare the rod and spoil the child". Upon second thought, the survey included direct interviews with the children’s parents. Imagine a parent who would have to sit knee to knee with a government researcher and admit to spanking his child. This would suggest that a parent who does so is simply more honest. Honesty, a so-called confounder in statistical term, is in reality more important to good parenting than spanking is to bad parenting. After all, I reason, we should pride ourselves on a hospital where error discussions are valued. The "incidence" of medical errors is going to be the lowest when the system for reporting problems is stacked against the whistle-blower who punishes and ostracizes the doctors reporting medical mistakes.

On the other hand, one might often hear that the hospital with the highest number of medical error reporting is the most malicious one overwhelmed with medical blunders. Such wayward thinking reminds us of another folktale of the czar, who learned that the most disease-ridden province in his empire was also the province with the most doctors. His solution? The almighty czar promptly ordered all the doctors shot dead.

2 comments:

Edmond Chow said...

Statistic is indeed difficult to interpret. Great article.

KM Chow said...

The tricky bit of statistics is that they are like swim-wear - what thy reveal is suggestive but what they conceal is vital.