Update 2010-01-19: I received a thoughtful note from Jane Sarasohn-Kahn at Health Populi — more thoughtful than perhaps I deserve, in light of my snarkiness — thanking me for the observation and letting me know she’s replaced the chart in her original post. I confess that I still have some trouble making the direct connection between this data point and EHR, but as a fellow believer in the potential for EHR and PHR to improve the healthcare landscape on many dimensions, I appreciate her trying to connect the dots. Original post below…

–

I’ve done a bit of work related to Electronic Health Records (EHR) and Personal Health Records (PHR), and as a result, I follow several blogs that cover these issues. I’m also very interested in data visualization, good and bad. Today, these worlds collide.

I’m a believer in EHRs, but today Health Populi takes a tiny data point (from a somewhat off-kilter outfit called the National Center for Policy Analysis) and not only places it entirely out of context, but also commits a major dataviz sin: screwing around with the origin point.

In context, NCPA says that in Minnesota, for treatments of colds and sore throats in children, MinuteClinic (the largest retail-based walk-in clinic) follow evidence based-guidelines 91% of the time, versus an average of 86% of the time for all primary care clinics. A 5% difference can certainly be meaningful, although it might not be. They provide this helpful chart to illustrate the information:

NCPA's Original Chart; not great but not bad

It’s not a great chart; best practice suggests that 3D bar charts can be slightly misleading. But it’s mostly honest, and shows that the difference is effectively tiny.

For reasons passing understanding, Health Populi recreates this chart:

ZOMG! Huge difference!

And abra cadabra! The difference in conformity to guidelines now appears to be about 2.5 times greater among retail clinics, an effect achieved by setting the bottom of the chart to the 83% point.

What’s especially weird here is that this massaging of the chart doesn’t even bolster the argument the Health Populi post tries to make, which seems to result from this leap of logic: MinuteClinics show better compliance with guidelines (explicitly true) -> better adherence to guidelines equals better outcomes (probably true, but not explicitly shown by the source materials) -> MinuteClinics have adopted EHR and this must somehow be responsible for these amazing results (wait… what?).

Maybe I’m missing something, but I’m having an awfully hard time seeing how this argument hangs together, or finding any possible justification for juicing the chart like that. I’d welcome comments.