A new study by Castlight Health has found that while nearly all hospitals have implemented CPOE systems, those systems are far from perfect. And that may be because too many clinicians find system alerts to be a distracting annoyance.
The research, based on an analysis of data collected by The Leapfrog Group, found that 96% of hospitals reported use of a CPOE system, up from 33% in 2010 in a scant 2% in 2001. This data is drawn from the 2015 Leapfrog Hospital Survey of 1,750 U.S. hospitals.
But while the high adoption rate might be good news, it comes with bad news as well. The Castlight analysis found that even where hospitals had CPOE systems in place, 39% of possibly harmful drug orders and 13% of potentially fatal orders weren’t flagged by the system in place.
The most common errors that didn’t get flagged included when clinicians prescribed the wrong meds for the patient’s condition, or the wrong dose or meds entirely inappropriate for kidney function, and the failure to display a reminder to test drug levels after issuing medication.
These errors are occurring despite the fact that many of the hospitals studied by Leapfrog (64%) met its CPOE standard. To do so, the hospitals had to alert physicians about a minimum of 50% of common, serious prescribing errors. Also, physicians had to order at least 75% of inpatient medication orders through a CPOE system.
So if the CPOE system is being used actively, and performing as it should in most cases, why would nearly 40 percent of potentially harmful drug errors slip by? The answer may be that fairly or not, CPOE alerting is still seen as a hindrance rather than a help by many physicians.
While I don’t have hard statistical evidence to this effect, the anecdotes doctors share suggest that some click through alerts as quickly as possible. One physician blogger shared that he was quite frustrated by the alert generated when he wanted to prescribe 81mg baby aspirin tablets, which patients can buy over the counter. I understand his frustration (and even what seems like wounded pride). And if it took several clicks to dismiss the related prompts, I’m sure it was indeed annoying.
On the other hand, as my colleague John Lynn rightfully notes, doctors aren’t going to blog or tweet about the time the CPOE system alert saved them from making a major prescribing error. So there is a bias to comments and blog postings since they only cover the negative side of CPOE and not the positive side. Perhaps the doctors who are working with these alerts successfully are simply going about their business and feel no need to vent. (Please note: I’m not suggesting that those who do vent are out of line in some way.)
Still, it seems quite clear that there’s considerable work to do in improving the workflow around physician alerting. If hospitals with CPOE in place are still seeing this level of potentially harmful or fatal prescribing, after many years to adapt to alerts, they need to do more to accommodate physicians.
P.S. They might want to start with a look at how Montefiore Medical Center succeeded with its CPOE rollout.