Free to decide

Describe electronic clinical decision support to someone outside of healthcare.

They'll probably start conjuring images of Watson or Clippy - an automated guide who incorporates clinical data and provider habits, to offer suggestions to improve care.

And yet, despite advances in computing power and machine learning, and a massive push to adopt electronic records, clinical decision support remains stuck - so primitive that even 90's-era Clippy seems wise and helpful in comparison.

I think there are many culprits for this, but a big reason clinical decision support lags is that it's been hijacked by well-meaning hospital administrators. Instead of nudging doctors into learning new things, dosing meds appropriately for complex patients, ordering the proper tests, so far it's been mostly inane reminders written by bureaucrats. Instead of helping doctors make decisions, decision support has been used to boost compliance on various metrics of interest.

You see, when compliance with a federal measure at your institution is at 97%, steps need to be taken to get that up to 100%. And those steps often involve popups, and "hard stops" that disrupt your progress through a task, even if it doesn't quite apply to this particular patient or that particular situation.

Of course these metrics are important, and if you don't score well, CMS reimbursement could be withheld, and your institution will look bad on HospitalCompare.

But by hijacking an EHR's clinical decision support system to boost administrative goals, you've conditioned a generation of providers to ignore CDS popups and warnings. The alerts are often not really relevant to providing the best care. They're often not appropriate to the situation. And they get really annoying to busy physicians, when they delay appropriate care and add to their workload.

What's happened is that modern decision support has taken the most inquisitive, hardworking, and self-sacrificing group of people you're likely to meet, and turned them off to the idea that the EHR can be a teaching tool.

And I don't think there's a popup alert to reverse that clinical situation.