CDS to Avoid Excessive Erythropoietin Stimulating Agent Administration
High Value Practice Academic Alliance Clinical Decision Support Innovation
Shobha W Stack MD PhD1, Elena Meeker PharmD2, Andrew A White MD1
1 University of Washington School of Medicine, Department of Medicine, Seattle, WA
2 University of Washington School of Pharmacy, Seattle, WA
Corresponding Author: Andrew A White, MD, University of Washington School of Medicine, Seattle, WA; andwhite@uw.edu
Patient Case
A patient with stage IV chronic kidney disease (CKD) developed a gradual decrease in hemoglobin to 8.7 g/dL. His nephrologist diagnosed anemia of CKD and ordered erythropoietin-stimulating agent (ESA) therapy. An ESA order included instructions to suspend therapy and contact the physician for a hemoglobin greater than 11 g/dL. The night before each weekly visit, a hospital pharmacist reviewed a paper copy of the order, entered it into the electronic medication administration record, and dispensed a dose to the infusion center. Neither the pharmacist nor the infusion center nurse noted that the patient’s hemoglobin level before the third ESA dose had risen to 13.2 g/dL because the lab value was in a different component of the chart and this value did not go to the pharmacist’s inbox. This was also missed when the patient presented for his fourth and fifth doses. At clinic follow-up after the five darbepoetin doses, the nephrologist ordered a repeat hemoglobin; the laboratory called with a critically elevated value of 17 g/dL.This required urgent phlebotomy, although even more costly harm, such as stroke or myocardial infarction, was avoided.
Plan
Recurring orders for ESAs require monitoring to prevent iatrogenic erythrocytosis and waste of these high cost drugs. However, free text comments to the nurse and pharmacist about holding unnecessary doses can be missed. Additionally, information in free-text fields cannot be used by clinical decision support (CDS) mechanisms that require discrete data. A root cause analysis of a case similar to this one highlighted the need for both enhanced clinical decision support (CDS) and standardization of care. For example, the infusion room lacked a standard threshold for suspending therapy, allowing each oncologist, cardiologist, and nephrologist to potentially provide a different hemoglobin level. Although a fixed hemoglobin limit for ESA dispensing would prevent ESA waste in CKD, it interferes with other valid uses which target higher hemoglobin levels such as cyanotic heart disease or myelodysplastic syndrome. Thus, standardization to a common threshold with exception criteria is needed for viable CDS logic.
Unlike many medication CDS alerts which target the physician at the time of order entry, excessive ESA use may only become apparent to the nurse of pharmacists after weeks of medication use. To avoid ESA overuse in the framework of the “5 rights” of CDS, the alert must be presented to the right health care provider at the right time. Although physicians should review lab results, they may not access the chart between the time of lab reporting and ESA administration. Additionally, some EHR interfaces for pharmacists do not display hemoglobin in the order approval window, necessitating navigation to another screen to review results; this seemingly small detail increases the chances of wasteful medication use. By sampling the EHR for signs of unnecessary ESA dosing, we found evidence of a range of 6 to 44 excess doses per year, but were unaware of catastrophic events, such as strokes, resulting from this overtreatment.
Do
Care standardization and an alert to the administering nurse were developed in collaboration between medication safety leaders, nephrologists, cardiologists, oncologists, and nursing leaders.
- Inclusion criteria: All patients in an outpatient infusion area receiving an ESA with a hemoglobin level of >12g/dL OR no hemoglobin measurement within the last 30 days.
- Exceptions are permitted to accommodate higher hemoglobin goals for cyanotic heart disease and myelodysplastic syndrome
- The alert fires to pharmacist at the step of preparing the dose, and also to the nurse prior to administering the dose (at the time of bar code medication administration).
ESA Monitoring Alert for PHARMACISTS if no hemoglobin within 30 days is available:
Copyrighted Material Used with Permission of Cerner
ESA Monitoring Alert for PHARMACISTS if Hgb > 12 :
Copyrighted Material Used with Permission of Cerner
ESA Administration Alert for nurses if no hemoglobin within 30 days is available:
Copyrighted Material Used with Permission of Cerner
ESA Administration Alert for nurses if Hgb > 12:
Copyrighted Material Used with Permission of Cerner
Study
Over four months of monitoring after system improvements, there were 210 ESA orders in the combined inpatient and outpatient settings, which triggered five alerts for elevated hemoglobin values. Two of the alerts resulted in cancellation of the order after the pharmacist or nurse contacted the provider, sparing those patients unnecessary ESA therapy. The other three alerts resulted in administration of the ESA after verifying the presence of an appropriate exclusion criteria for cyanotic heart disease.
Building and testing this CDS required modest IT resources. Programmers and pharmacists dedicated 8 total hours of labor to this project, at an internal cost of $752. However, this does not include the effort of the patient safety leaders who performed a root cause analysis or the multispecialty team that met to reach consensus about target hemoglobin levels. We estimated pharmaceutical savings from this CDS intervention by assuming it averted 6 to 44 unnecessary doses per year and by applying an average purchase cost of $589 for the mix of ESAs used at our institution. This suggests annual direct medication cost savings between $3534 and $25,916. Avoided charges to payers would be greater and this estimate does not include savings related to preventing harm to patients, ranging from unnecessary travel to serious thrombotic events. Even with the most conservative estimates of medication savings, this CDS recoups investment in 4 months.
Lessons Learned (Act):
These alerts fire infrequently, but appropriately. Combined with clear standards, this has meaningfully reduced the possibility of inappropriate ESA administration, and the related patient harm. This approach upholds several important principles in the creation of safe clinical systems: use electronic systems thoughtfully to avoid overreliance on human memory, present the right information at the right time, reserve interruptions for critical information to avoid alert fatigue, and design computer interfaces around workflows to reduce cognitive burden on users.
This report has notable limitations. Due to concern for patient safety, the CDS system was implemented without a comprehensive needs assessment for potential cost savings. There are opportunities to further improve the function of this CDS system in subsequent PDSA cycles. For example, the ESA orders could require the user to select an indication from a structured list, allowing the CDS logic to avoid unnecessary alerting for patients with an appropriate indication for a higher hemoglobin target.
