March 10th, 2014
Can Decision Support Tools Actually Harm Patients?
Christine Sinsky, MD
Christine Sinsky is an internal medicine physician in Dubuque, Iowa. This post originally appeared on her blog, Sinsky Healthcare Innovations, and has also been published on KevinMD.
Have you ever encountered an intrusive, insistent popup decision support screen while trying to take care of patients? Found yourself stuck in a dead end electronic hallway without egress? A situation where you had to choose an option that was inappropriate for your patient just to exit the screen?
This is the situation with my local medical institution’s “DVT Advisor,” a mandatory decision support screen for all patients.
DVT Advisor was implemented in response to meaningful use and out of a laudable desire to improve safety. Unfortunately, as constructed, the Advisor has the opposite effect. In this popup the physician must process through a 5 screen, 9 click process. The first step identifies the patient’s DVT risk level. If the risk level is low, no intervention is required. Yet, inexplicably, the physician must next process through the 4 additional screens to record whether the patient has any contraindications to the interventions which the Advisor has just indicated are not recommended. (!)
I’ve learned how to click through each of these cluttered, non-intuitive screens to get on to the business of taking care of patients, but it is a distraction, and thus hazardous.
I recently had a patient the DVT advisor identified as “high risk.” I was sending him home that day, so no anticoagulation was necessary. Nonetheless I had to wade through each screen, documenting that the patient didn’t have any contraindications to a treatment I wasn’t planning to give. At the end there was a force field where the only option was to choose one of 4 anticoagulant regimens. I could find no other option. There was a button elsewhere on the screen where I could decline the intervention, but doing so simply resulted in the popup reappearing the next time I entered an order, and I had to go back and progress through all 5 screens from the beginning and ultimately choose an anticoagulant for the patient in order to continue writing orders, or finish all of the other orders and then exit when there would be no further orders to trigger the popup yet again. No other way to get around it. No way to be the human being in the equation who had sense and could say this patient doesn’t need this intervention.
One of my colleagues told me that in these (common) situations he selects one of the anticoagulant options, which creates an order for his patient for a medication that is not indicated, and then he goes back in the medication list a few minutes later and discontinues it. Talk about waste. Talk about hazard.
The information in the DVT Advisor can be a useful reference if a physician is uncertain about anticoagulation, but its intrusive and insistent characteristics are based on hope and belief, rather than evidence. The mandatory requirement to prescribe anticoagulation, even when the physician has reasons not to prescribe it, has gone far too far in my view, and has created a systematic occasion for patient harm.
Time spent on such low-value exercises or workarounds is time taken away from other patient care. What 5 minutes of talking with the patient, a colleague or consulting a medical resource, what 5 minutes of deep medical decision making shall I give up in order to move through this poorly constructed, time-consuming low-value work?
The hospital is a large organization, and has heard from many physicians about the form, but still the rigid and mandatory popup persists. Individual physicians do not have control over its design and yet labor under its impact. This is part of the real life harm associated with products that could in theory be helpful. The form was designed in response to the current federal regulatory climate, citing meaningful use and a move to more clinical decision support. The resultant hazard to patient safety may be an unintended consequence of over-reaching regulation but it is a real side effect nonetheless.
Addendum: A week later I had a 91-year-old patient identified as high risk who was already on coumadin. The DVT Advisor would not allow me to exit without ordering enoxaparin or unfractionated heparin. I could find no other way to deal with this except as my colleagues have also done: order the enoxaparin, wait a few minutes for the order to be processed, and then remember to go back in and cancel the order.
EHR’s are decreasing “productivity”, ripe for errors that are hard to correct, and take time away from patient care and generate reams of useless data through the cut and paste method of charting. In addition, if like me you work in your own office, have your own EHR and affiliate with two hospitals, you now have three systems to master.
Someone wrote recently that EHR’s are like texting while driving. The risk of missteps leading to death may not be far off the mark.
Nonetheless, most of us power through these systems, do what is asked and our work days get longer and longer.
Wholeheartedly share the author’s frustration…and am certain many colleagues share the same concern. Instead of being supportive tools, they have become excessive, inappropriate, cumbersome, essentially dumbing of medicine!
As a proponent of decision support tools, the value of a well-designed, tested and useful tool/checklist is notable. Unfortunately, many of these tools are not designed by or with the input of the end user rather only to meet a metric. This often creates tools that feel cumbersome to the end user resulting in work arounds (like you describe) or non-use of the tool (if possible).
I had similar experience with one of these forced function in that as practice guidelines changed the requirement became obsolete but the order function was never changed. Hundreds of providers would click through this to get the work done, moan about the frustration and move one. Receivers of the order actually ignored this field as was often populated with incorrect information or deleted by the ordering provider and absent. Until it happened, an error occurred. Following the field was changed/removed and this was better but for a long time everyone knew it was a potential problem.
Would recommend reporting this to your institution. Enter an incident report because every time you have to order a medication that is inappropriate and then remember to go back and discontinue this before you sign the orders is is a near miss. All it would take is one interruption at the right time and you or a college forgets to remove the medication and the patient gets the medication. Hold your organization accountable for this and demand a change but offer to be part of the change and suggest changes so that the tool becomes useful and not a burden. It is a patient safety issue.
I agree with Dr Cox,
that better design may reduce the distraction. But in some volume the waste of time is inescapable with such algorithms. The fuzzy character of human decision making may not be simply transformed in the yes-no|option algorithms. The option may be to permit to physician to accept the work with the tool, or skip it. But to skip responsibly (Are you sure? It is your responsibility…)