Errors That are Truly Fatal

During lunch at the International Python Developers Conference this month, I found myself at a table with several application developers who work for medical service providers. We were talking about handheld devices, following that mornings announcement that a version of the Python scripting language was now running on Palm OS.

I suggested that it was only a matter of time before a test case hospital death—caused by misunderstood written instructions—spurred wholesale medical adoption of handheld data terminals. These could ensure more accurate transmittal of doctors orders.

Around the table, heads were shaken at my naiveté. “It happens too often,” one medical software consultant replied, “to get anyone that excited.” I took his comments to be playful exaggeration, but statistics show him to be—pardon the expression—deadly serious. Medical errors outnumber automobile accidents, by a factor of two to three, depending on your source, as a cause of death in the United States. A hundred thousand people every year, give or take 20,000, die from mistakes that are largely matters of misinformation.

The fundamental failure of medical IT is not so much technical as social: I refer to the widespread failure to gather, index and offer convenient access to data that might elicit common patterns of error. Other application domains also exhibit this problem; if failure is treated as shameful, instead of being regarded as a learning opportunity, valuable data will be suppressed. Every IT professional can take a leadership role in this area.

The IT community has a substantial head start on many others in the development of nonproprietary standards for instruction and explanation. Relevant data, consistently packaged, can save lives.