Part of a series on primary care challenges and their solutions.
I missed a drug interaction warning the other day when I prescribed a sulfa antibiotic to Barton, a COPD patient who is also taking dofetilide, an uncommon antiarrhythmic.
The pharmacy called me to question the prescription, and I quickly changed it to a cephalosporin.
The big red warning had popped up on my computer screen, but I x-ed it away with my right thumb on the trackball without reading the warning. Quite honestly, I am so used to getting irrelevant warnings that it has become a reflex to bring the cursor to the spot where I can make the warning go away after a quick glance at it. Even though I have chosen the setting “Pop up drug interaction window only when the interaction is severe”, I get the pop up with almost every prescription.
Today I went back to Barton’s chart and looked at his interaction screen.
With the Bactrim DS no longer there, the first of the red boxes was a major interaction between his 81 mg aspirin and his Pradaxa (dabigatran) – two blood thinners are more likely to make you bleed than one. That is basic knowledge, even common sense.
The next red box was a moderate interaction between his baby aspirin and his lisinopril. Theoretically, higher doses of NSAIDs can interfere with the blood pressure lowering properties of ACE inhibitors. That is very basic knowledge, too.
The third red box, another moderate interaction, was between the aspirin and his steroid-bronchodilator inhaler. Theoretically, steroids and aspirin can increase the risk for stomach irritation and supposedly, the pharmacologic effect of aspirin may be decreased by the inhaler.
After these came several warnings labeled “extreme caution” and some that were “not recommended”. The scrolling seemed endless, so I printed out the warnings instead. They filled eight pages. I counted 61 “extreme caution” warnings, from metoprolol and diabetes to the poor man’s steroid-antifungal cream and his diabetes. Beta blockers can, at least theoretically, decrease the tremors and other warning symptoms of low blood sugar, and oral steroids can raise blood sugars, but a mild steroid cream doesn’t do that.
There were 32 “use cautiously”, many of them quite tangential, like the blessed fungus cream and Barton’s history of hepatitis C.
On the last two pages were the dietary warnings, including not to swallow your atorvastatin with grapefruit juice, or to mix your pain pills with alcohol.
I hate to sound uppity, but no amount of pop-up interaction alerts or other forms of “decision support” can replace basic medical education. In Barton’s case, the only warning I needed was the one about his dofetilide, which he gets from his cardiologist, and the antibiotic I wanted to prescribe. The aspirin-Pradaxa interaction is common sense, and the baby aspirin-Symbicort interaction is nonsense. And if I were to even read through the eight pages worth of precautions and “use with caution”, I would have doubled the 15 minutes it took to assess and document his infection in the first place. Or I could have listened to a tutorial about evaluating lung sounds – how much coaching do the EMR designers think we need?
So, here is my suggestion: Make these warnings behave like some computerized card games – let users decide based on their skill level whether to get all the warnings or only the critical ones that are not generic class effects we all learned in pharmacology class. Because when everything is a red alert, alarm fatigue sets in and all the warnings are wasted.
It reminds me of the story I heard as a child about the boy who cried wolf…