Las Vegas—Informatics pharmacists shared their strategies for using health care technology to reduce drug alert fatigue, set up remote pharmacies for rural locations, make smart pumps safer, and much more, during a session at the American Society of Health-System Pharmacists 2012 Midyear Clinical Meeting sponsored by ASHP’s Section of Pharmacy Informatics and Technology.
Cleveland Clinic Makes Its Smart Pumps Smarter
By standardizing smart pump settings and better managing alerts, pharmacists at the Cleveland Clinic Health System were able to improve the large-volume pump system that goes across all 10 of its hospitals and its 19 family health and ambulatory surgery centers, according to Silvana Balliu, PharmD, RPh, a smart pump pharmacist at the health system.
Noting that alert alarm management is a key factor in reducing health technology hazards, Dr. Balliu described how Cleveland Clinic pharmacists focused on “decreasing the number of alerts received, to make them clinically significant to our caregivers.”
After reviewing continuous quality improvement (CQI) alarm report data and documenting the number of alerts received, a team at Cleveland Clinic tailored the audio alarm settings for alerts based on care areas and practice settings. The analysis revealed that many alerts received during infusion programming of pressors were caused by caregivers, who, when titrating doses downward ended up administering dosages lower than the lowest soft limits required by the smart pump’s settings. The health system significantly decreased alerts in this area by removing alerts for the lowest soft limits settings.
To reduce the risk for infusion pump programming errors related to high-risk medications, Dr. Balliu said they programmed high-risk medications in the pump to be given as primary, not secondary, infusions. Also, they prohibited secondary medications from being programmed with high-risk medications.
Cleveland Clinic also standardized settings between the smart pump and the electronic medical record (EMR) to provide clinicians with consistent information regarding dosage, duration and concentration, Dr. Balliu noted.
For heparin infusions, the CQI analysis revealed that some alarm alerts were caused by caregivers attempting to enter the volume per hour versus the dosage. The team eliminated this practice by implementing the lowest hard dosage limit for heparin.
Pharmacy works closely with nursing and anesthesia representatives during implementation and maintenance to “provide education and to make the library files transparent to them,” Dr. Balliu added. By making these types of improvements to the smart pumps, the number of alarm alerts decreased 50% from the fourth quarter of 2010 to the fourth quarter of 2011 (figure). Today, Dr. Balliu added, Cleveland Clinic is working toward its goal of having a library compliance rate of 98%.
CPOE Helps With Timing Of Therapeutic Drug Levels
Before CPOE was implemented at the 173-bed Lawrence Memorial Hospital (LMH), in Lawrence, Kansas, pharmacists there discovered that many tests for drug levels were being timed inappropriately, contributing to multiple redraws and inappropriate adjustments of therapeutic drug dosages, and “increasing the cost of care,” said Gregory P. Burger, PharmD, a clinical pharmacist at LMH.
In March 2010, when LMH went live with CPOE, management made the decision that pharmacy would time all therapeutic drug levels for doctors.
To evaluate the effects of CPOE on timing of drug levels, pharmacists at LMH looked at phenytoin level timing for the year before and after CPOE implementation. Dr. Burger said they found that before CPOE, 73% of their orders were timed incorrectly compared with 17% after CPOE implementation.
After analyzing the post-CPOE data, Dr. Burger said they found that most of the errors occurred because pharmacists “did not have all the information they needed to determine whether the doctor wanted a routine level or a stat level. We also had some confusion with laboratory not drawing the level at the correct time.”
In May 2012, LMH modified the CPOE system, putting in two different drug level orders, a “now” order (draw immediately) and “routine” order (draw based on best pharmacokinetic practices). The “now” order goes directly to the lab, but a communication also is sent to pharmacists so they can monitor the lab result. The routine order goes to pharmacy, which sends an order to the laboratory to get an appropriate level. Pharmacy also sends a communication to the clinical pharmacist, so he or she can monitor the result, and sends a “dummy drug” order (pharmacy communication) to the EMR, so the nurses can coordinate with the lab to make sure the level is drawn on time relative to administration of the dose.
In December, using the new process resulted in only two errors up to that point, Dr. Burger noted. “We have improved our processing to 92%, just an 8% error rate. Currently, all narrow therapeutic drug levels are being processed this way.”
Virtual Ambulatory Consult Is a Step Toward Collaborative Practice
To bridge the gap between pharmacy care for inpatients and outpatients in an affordable way, Roanoke, Va.-based Carilion Clinic, which owns and operates eight hospitals and 160 ambulatory practice sites in the western part of Virginia, is using its integrated EMR system to deliver post-discharge pharmacy consultations remotely.
Carilion started its remote program “very simply, with one pharmacist and one ambulatory practice site comprised of six health care providers,” said Kathy Yount, BS Pharm, RPh, a clinical pharmacist analyst at Carilion, in her pearl presentation.
Physicians work with an office referral coordinator to send an electronic consult order through the hospital’s EMR to a remote pharmacist. Once the pharmacist receives the referral message, he or she uses the EMR to review the patient’s entire medical record, including such data as the patient’s medical history, current medications, pertinent laboratory information, insurance coverage, notes and so on.
On average, the pharmacist consults remotely with about two outpatients a week, mostly providing medication therapy recommendations. Treatment recommendations are sent directly to providers via the EMR.
The program, which so far has been expanded to a second hospital and a second ambulatory site, has the capability “to put standards in place that can be used with multiple patients, multiple locations and multiple pharmacists,” Ms. Yount said.
Carilion also is working on developing an EMR medication management follow-up program to identify high-risk outpatients and flag their charts for follow-up consultations. Through that project, according to Ms. Yount, electronic health records will be shared among multiple hospitals, primary care practices and community pharmacies.
Better Timing of Vancomycin Levels Improves Safety
Following the 2009 introduction of recommendations for vancomycin trough levels in the American Journal of Health-System Pharmacy, Gainesville-based Northeast Georgia Medical Center (NGMC), a 557-bed community hospital that typically has an average of 300 patients a month receiving vancomycin, changed its vancomycin dosing and monitoring guidelines.
To help prevent the development of resistance, under the new guidelines, a 10-mcg/mL trough level became the minimum rather than the maximum level. The new maximum trough level was set at 20 mcg/mL, a level low enough to reduce the risk for vancomycin-induced nephropathy.
In late 2009, after NGMC made the changes, Melissa C. Frank, PharmD, a clinical coordinator at NGMC, said the center’s Medication Safety Committee noticed a trend in medication errors—specifically, nurses were administering vancomycin to patients whose trough levels were already greater than 20 mcg/mL.
A multidisciplinary team worked to make several initial changes to patient care. The lab initially adjusted the critical lab alert level, the point at which the lab would be required to directly notify the patient’s nurse or physician, from 40 to 30 mcg/mL.
NGMC also educated nurses about the risks associated with pushing trough levels higher than 20 mcg/mL and educated all pharmacists on the new guidelines, encouraging them to be “extra vigilant.”
But as the hospital moved into 2010, they noticed another four-month trend of the exact same type of errors. To address the problems, they built a “dummy drug” tool in their medication management software that allowed them to enter a vancomycin trough level together with the time it was due for nursing to see and chart against in the electronic medication administration record.
“So the time due is the actual time that the trough level will be drawn by the lab or by nursing in the critical care units,” Dr. Frank said. “The nurse is required to manually document the dose level as it is given. If their document for a vancomycin trough level is greater than 20 [mcg/mL], an alert fires in the system.”
The first alert is a soft stop that tells the nurse a trough level is higher than the upper limit. If the nurse tries to override, a second alert fires, which again reminds him or her of the risk level. The nurse also sees a note telling him or her to withhold the dose if the level is at or above 20 mcg/mL.
Following the implementation of the dummy drug tool and the training, Dr. Frank said, there have been “no cases in which patients with trough levels above 20 mcg/mL were administered a dose of vancomycin.”
VA Tackles Electronic Prescriptions for Controlled Substances
In a presentation on creating electronic prescriptions for controlled substances, Robert Silverman, PharmD, the program manager of pharmacy benefits management for the Department of Veterans Affairs (VA), outlined steps that hospitals can take to facilitate implementation of ePrescribing for controlled substances (EPCS) via their computerized prescriber order entry (CPOE) systems.
Through its CPOE system, the VA has been electronically prescribing lower-level schedules of controlled substances for more than 15 years. Electronic prescriptions of controlled substances for physicians require a two-factor authentication “that you can’t get around,” Dr. Silverman explained. “You need any two of these: something you have (e.g., a smart card used as a personal identification verification card), something you know (a password) or something you are (a biometric ID of a fingerprint or an eye).” The computerized system that an EPCS prescriber uses must present them with an “attestation statement,” stating that what they are about to do is the equivalent of their legal signature. Additionally, the prescriber application for EPCS needs to be able to capture the ePrescriber’s explicit “intent to sign” for each individual prescription, and both ePrescribing and pharmacy applications need a third-party audit to verify that they are fully compliant with Drug Enforcement Administration (DEA) rules.
At press time, the VA was still finalizing the interview and reporting process for security and was anticipating that its CPOE EPCS system would be DEA-certified in the first quarter of this year.
Dr. Silverman said that hospitals pursuing certification for controlled substance ePrescribing should “check out the EPCS certification status of your vendors via the DEA’s approved website [http://www.deadiversion.usdoj.gov/ecomm/e_rx/thirdparty.htm#approved]; evaluate the certification practices as shown there; evaluate practices for receiving fax or paper copies after electronic prescribing; and, most importantly, make sure your CPOE EPCS providers have their two-factor authentication systems in place.”
Software Snafu Points To Need for Pharmacist Input in IT Systems
In November 2011, Lifespan, a Providence, R.I.-based health system that includes Rhode Island Hospital, discovered that some 2,000 patients may have received prescriptions or continuity-of-care documents with the wrong form of medications because “of a software issue that affected patient discharges,” said Donald McKaig, RPh, CDOE, a clinical pharmacist specialist in medication safety at the private, not-for-profit 719-bed hospital. The changes occurred after final electronic approval of the order by the physician when the order was in post-processing, Mr. McKaig noted. He explained that when the hospital’s CPOE master drug database’s software reached out to a national drug database to perform clinical checking, information from that database’s fields for a different form of the prescribed drug was pulled into Lifespan’s CPOE database, “overriding some of the data in our systems, leading to changes in the drug sentence,” Mr. McKaig said. The replaced fields included drug descriptions, dose forms and strengths that were originally present in the inpatient medication orders. For example, an inpatient order for “morphine SR 30 mg 2XD” was changed to morphine 30 mg 2XD.”
A pharmacy-led review of the CPOE master drug database’s 13,500 drug entries and 357,000 data fields containing drug descriptions, dose forms and dosing strengths from Lifespan’s four hospitals was undertaken and the data were compared with output from the national drug database. The review detected 3,132 potentially problematic drug entries. Lifespan also discovered “errors during manual entry of data into our CPOE database that led to incorrect, invalid descriptions of medication dosages and dose forms in our system,” Mr. McKaig said. They also discovered that in some cases, unlicensed personnel were performing data entry processes manually.
Now with a new process in place, Lifespan is performing enhanced testing and verification of all data fields in its CPOE master drug database, and is verifying all data fields populated during data entry as well as outputs for reporting purposes. They also are developing automated scripting tools to eliminate manual entries of data.
The lesson learned from the software problem, according to Mr. McKaig, is that “pharmacists need to be involved in the testing, implementation, design and ongoing monitoring of [medication] management technology. We need to be out there with
our IT teams knowing exactly what the source of truth is for all the documentations used in our electronic systems. Standardized tools must be developed to ensure accuracy of data entry processes. Manual data entry by unlicensed personnel must not be allowed.”