Do Electronic Records "Cause" Better Diabetes Care? Who Cares?

Persons with diabetes should look for this?

High functioning primary care sites are more likely to purchase an EHR. 

That's the contrarian conclusion of the Disease Management Care Blog after reading this New England Journal article by Randall Cebul and colleagues titled "Electronic Health Records and Quality of Diabetes Care."

Briefly, this was a study of persons with diabetes who were being served by the primary care sites of seven health care organizations in Cleveland and surrounding Cuyahoga County in Ohio.  There were 21 primary care sites in a not-for-profit system, 12 owned by a safety-net hospital, 1 in a university hospital and the remainder were federally qualified health centers.  The sample was made up of 27,207 patients cared for by 569 primary care physicians in 46 practices.  13 practices with 53 providers used paper, the remainder used EHRs. 

The impressive results can be found here.  Basically, if you had diabetes and was cared-for in a clinic with an EHR, you were more likely to have your HbA1c tested, kidney disease addressed, eyes evaluated for retinopathy and be immunized against pneumonia.  You'd also have better control of your blood glucose, blood pressure, cholesterol, weight and be less likely to use tobacco.  For most of the measures, the differences were in the double digit range, and, except for glucose control and blood pressure control, remained significant after statistical adjustment.

So why is the DMCB being a nattering nabob of negative nitpickiness? 

This is an observational study involving physicians who independently chose whether to purchase an EHR.  It is quite possible that high functioning and financially successful clinics that were already taking good care of their patients with diabetes committed to an EHR, which was otherwise an innocent bystander.  That's called "confounding" or "selection bias" and cannot be ruled out.  We cannot say for certain that if the paper-using clinics were forced to purchase an EHR that their numbers would have improved.

That being said, the DMCB doesn't really care.  If the EHR caused better diabetes care or is merely associated with better diabetes care, the "signal" is the same: having computer screens in a clinic means higher quality.

The policy implications:

1.  Based on this study, we still don't know for sure if the EHR causes better care, but

2. It makes sense to "steer" patients with diabetes toward clinics that have EHRs.