HealthSeptember 27, 2016

Your alert management questions answered

Drug dosing errors account for nearly 40% of preventable medical errors in the U.S., and yet healthcare professionals routinely miss valuable dosing warnings from electronic health records (EHRs) due to the glut of irrelevant alerts that cause alert fatigue.

Author: David Kaelber, MD, PhD

EHR screening can help alert professionals to potentially inappropriate or contraindicated drug dosing choices, but in order to overcome alert fatigue and draw real value from relevant, targeted alerts, users need to dive into their data, analyze, and design the optimal decision support system.

That’s what Dr. David Kaelber, CMIO and Vice President of Health Informatics for the MetroHealth System in Cleveland, Ohio, did at his institution, to positive results. He discussed his experience in the RxPerts Academy webinar “Optimizing Drug-Dose Checking to Minimize Alert Fatigue.”. Here, Kaelber answers some of your alert management questions:

Q: What are some of the types of clinical decision support that can be implemented to try and guide professionals’ decision making — consciously or even unconsciously?

A: Some of the most effective methods include:

  • Limited choices — For example, only show formulary drugs, or only show routes of medication administration that are possible. And use picklists that don't allow somebody to choose or order other options, like drugs that doesn't make sense for a route. These can be baked into design and implementation of your electronic health record and a computerized physician order entry.
  • Action order — We know from studies in informatics that, all things being equal, the first item on a list will be disproportionately picked. So, be very strategic about what you put as that first option in a picklist.
  • Default configuration — How configurations and buttons are designed could affect what is selected or overridden. Make the right thing to do the easy thing to do. Think about different colors strategically. One of the things we've done in our system recently is color code alerts based on their importance, so a red alert is something that you really, really, really should pay attention to, whereas green or gray alerts probably do not have the same clinical impacts.
  • Passive vs. interruptive alerts — Passive alerts are real-time alerts that occur in the workflow but do not stop the workflow, whereas an interruptive alert is the classic pop-up alert where the provider literally cannot proceed until they address that alert. Think strategically about what is passive alert appropriate and what is interruptive alert appropriate.
  • Order validation alerts — These occur at the time of signing or approving an order. This is an example of the classic type of CPOE interruptive alert to catch potential errors at that last step.
  • Restricting medications — Most EHRs allow you to take a whole class of medications, often very high-risk medications, and restrict them so that can only be ordered by a specific group of people. You can designate some medications that have to be ordered by two people or “cosigned” before they can go through.

Q: Is an optimized drug-drug allergy alert setting aimed for pharmacists or prescribers?

A: In our system, drug-drug alerts and drug-drug allergy alerts are set up the same for various clinicians, but for drug-dose checking, it's a little bit more strict for pharmacists. But the big picture idea is that within your electronic health record, you likely have the ability to set up alerting based on types of users. You could set up alerting differently in different areas for different types of screening, and differently for pharmacists or prescribers or nurses or others. What we do is we turn on alerts in our systems silently, so that we can analyze them and try to get a sense of false positives versus false negatives. Then we know what we can turn off. And really, when I say “turn off,” all that means is we decide not to show them to end-users.

Q: Are there best practices for screening for duplicate therapies?

A: It will depend on your institution’s strategy and experience. At our organization, the lessons learned around drug-drug duplication were that we found drug-drug exact duplication alerting moderately helpful, because a lot of times, particularly in the inpatient setting where there are lots of physicians taking care of a patient, a physician who is cross-covering may not be aware that the same medication was already prescribed to a patient. However, our experience looking at drug-drug class duplication found those alerts to be very unhelpful. Most of the time if you're prescribing, for example, a short-acting opioid and a long-acting opioid, you actually wanted to do that. To have the computer tell you, “Hey, did you know you're prescribing two opioids?” is not always helpful.

Q: What are some specific ways you reduced number of alerts without sacrificing patient safety?

A: First of all, we turned off all of the minimum dose alerts. There aren't big patient safety issues in terms of adverse drug events for giving too low a dose. They're basically all associated with too high a dose. We then increased the maximum drug-dose single and daily thresholds from 100% of the recommended dose, which was the default, to 125%. That knocked out almost 50% of those alerts. My view and the view of our pharmacist is that there’s not much chance of adverse drug events occurring by going up to 125% of recommended dose. Usually, the adverse events are going to occur at much higher thresholds, like 200% or 500%.

We studied our alerting data over time, and we identified the top 1% of all of the individual medications that were still causing alerts. This ended up being 22 individual medications. And what we saw was that those 22 medications accounted for 75% of drug-dose alerts. So, we knew where to focus our attention.

Q: What were those 22 medications?

A: See the chart below for the medications and some details about them.

Drug name Baseline max single dose alerts, % (n) Baseline max daily dose alerts, % (n) Baseline single dose limit Single Drug-Dose optimization Baseline daily dose limit Daily Drug-Dose optimization
Aluminum & magnesium hydroxide (simethicone) 3% (2,992) 1% (1,199) 20mg 30mg 60mg 180mg
Acetaminophen 2% (1,966) 1% (1,430) weight base 1,000mg weight based 4,000mg
PPI1 1% (1,466) 1% (1,751) 20mg 80mg 20mg 120mg
Vancomycin 1% (1,128) 0.5% (556) 1,000mg 2,000mg 2,000 (weight base) 3,000mg
Diatrizoate meglumine-sodium (contrast) 0.6% (665) 0% (0) 90 ml 1,000ml 90 ml 1,000ml
Albuterol (inhaler, solutions) 0.6% (598) 0.5% (539) 4 mg or 2 puff 15 mg or 8 puff 8 puff or 30 mg 48puff or 288mg
Prednisolone, Prednisone, triamcilone 0.6% (575) 0.3% (309) varied 1,000mg IV or 100mg PO varied 3,000mg IV or 100mg PO
Lorazepam 0.5% (540) 1% (444) weight based 30mg weight based 96mg
Sodium polystyrene 0.4% (376) 0% (2) 15 g 40g or 30mmol 60 g 180g or 30mmol
Zolpidem 0.3% (336) 0.3% (327) 5 mg 10mg 5 mg 10mg
Amoxicillin 0.3% (309) 0.2% (193) weight based 2,000mg weight based 4,000mg
H2 blockers 0.3% (297) 0.2% (226) 75 mg 300mg 150 mg 600mg
Ibuprofen 0.3% (297) 0.5% (566) weight based 800mg weight based 3,600mg
Insulin 0.3% (294) 0.3% 349) varied 100 Units varied 300Units
Enoxaparin 0.2% (255) 0.2% (166) 40 mg 180 mg weight based 300mg
Ipratropium 0.2% (253) 0.4% (387) 0.5 mg or 2 puff 1.5 mg or 4 puff 2 mg or 12 puff 6 mg or 24 puff
Eye/ear drops 0.1% (151) 0.2% (218) varied turned off varied turned off
Statins4 0.1% (138) 0.1% (119) drug based 80mg Drug based 80mg
Heparin 0.1% (120) 0% (6) weight based 30,000Units 40,000 Units 90,000 Units
Ferrous Sulfate 0.1% (103) 0.05% (47) varied 440mg varied 975 mg
Ondanestron 0.05% (55) 0.08% (80) 4 mg 32mg 12 mg 48mg
Morphine 0.03% (33) 0.2% (158) 30 mg PO 40 mg PO 30 MG PO or 180 mg IV 80 mg PO or 360 mg IV
Total 12% (12,461) 9% (9,072)

Dr. David Kaelber is a practicing internist and pediatrician and the Chief Medical Informatics Officer and Vice President of Health Informatics for the MetroHealth System in Cleveland, Ohio. He is board certified in clinical informatics. Dr. Kaelber is also currently a professor of internal medicine, pediatrics, epidemiology and biostatistics at Case Western Reserve University.

Embedded drug data and clinical screening modules to support appropriate medication prescribing
Organizations use Medi-Span® drug data and clinical screening modules to support appropriate medication decisions, and to help reduce prescribing errors by optimizing alert systems.