Technology to the Rescue: Preventing Chemotherapy Errors

April 14, 2008

Chemotherapy errors have been attributed to many causes, including illegible handwriting, transcription errors, and human error.

It is no surprise that chemotherapy agents are high-alert medications. Errors involving high-alert medications have a heightened risk of causing harm to patients and in some cases may be lethal. Although the overall incidence of chemotherapy errors is unknown, errors were noted in 3% of chemotherapy orders written for adults and 1% of chemotherapy orders written for children in a 2005 study. However, unlike chemotherapy errors in adults in which pharmacists and nurses intercepted nearly half of the errors before they reached the patient and none of the errors caused patient harm, the majority of pediatric chemotherapy errors (85%) reached the patient, and 15.6% of these errors caused signifi cant harm.

Chemotherapy errors have been attributed to many causes, including illegible handwriting, transcription errors, “look-alike, sound-alike” chemotherapy drug names, use of acronyms and abbreviations, dose miscalculation, unclear decimal points, and human error. Factors that contribute to human error include time pressures, distractions, fatigue, understaffing, inexperience, and inadequate resources and available support. Although technology can be used to prevent many types of chemotherapy errors, the potential for human error still exists. Therefore, technology must be viewed as a helpful tool and not a guaranteed solution for eliminating chemotherapy errors.

Computerized prescriber order entry

The majority of published studies have examined the eff ect of technology on medication errors in general. One of the most studied technologies is computerized prescriber order entry (CPOE). A review of 12 studies of CPOE found that prescribing errors decreased 66% when handwritten orders were replaced by CPOE systems. But only about 9% of hospitals in the United States have CPOE, and no “industry standard” CPOE system exists. Some hospitals use commercially available systems, some create their own, and some adapt the available systems. Although the overall incidence of prescribing errors dropped in the 12 studies, the rate of one type of error—prescribing the wrong drug—did not decrease. “Wrong drug” errors were attributed to some of the safety hazards associated with CPOE, such as systems that list medications by generic name only or prescribers who inadvertently click the incorrect medication from drop-down menus, which are usually arranged alphabetically. Also, in five of the 12 studies, the number of adverse events from drug errors did not decrease.

Because CPOE systems can be manually manipulated, electronic alerts may be overridden by CPOE users. In a multicenter study of 3,481 CPOE alerts, prescribers overrode 91.2% of drug allergy and 89.4% of high-severity drug interaction alerts. This high rate of overriding electronic alerts increases patients’ risk of adverse reactions. Little is known about how often or under what circumstances chemotherapy orders are overridden manually. One published case report explained that—despite the use of a CPOE system—a patient received a cisplatin dose of 760mg instead of 190mg because of a flaw in the CPOE software. The overdose caused severe pancytopenia and renal failure that required the patient to undergo hemodialysis.

Double-checking dose calculations is an important step in reducing chemotherapy dose errors. Usually a handheld calculator—often a small tool that is vulnerable to users pressing an incorrect button—is used. In a few cases, doses are calculated manually using paper and pencil or calculated in the prescriber’s head. In a pediatric study, a CPOE system with an integrated calculator automatically used the patient’s age and current weight to adjust chemotherapy doses. The number of orders that contained errors decreased by 83%. Lead investigator Christopher Lehmann, MD, noted that the integrated calculator “stops ordering errors before they can even reach the pharmacy, let alone the patient.” Some CPOE systems have clinical decision support (CDS) features, which include dosing guidance, formulary decision support, and electronic system checks for allergies, duplicate therapies, and potential drug—drug interactions. Advanced clinical support features include dosing guidance for geriatric patients and patients with renal impairment, recommendations for medication-related laboratory testing, drug-pregnancy checking, and presence of drug—disease contraindications.

Bar coding

Bar-coding systems range from those used only in the pharmacy to point-of-care systems, in which nurses scan bar codes on their employee badges, patients’ wristbands, and the medications to be administered. Each medication has a bar code that encodes its National Drug Code, name of the medication and dose, and packaging. Portable devices at the bedside send encoded information to a computer server where algorithms verify the “fi ve rights” of medication administration. Most of the bar-coding literature consists of implementation considerations (eg, cost, equipment reliability, user resistance). A few studies have reported significant reductions in the number of medication errors that occur during the administration phase of the medication use process, as there are fewer opportunities to intercept errors than in the prescribing, transcribing, and dispensing phases. However, like CPOE, only about 5% of the nation’s hospitals have adopted this technology; bar-coding capital equipment costs range from $1 to $5 million, depending on institutional needs, and secondary costs—such as infrastructure and training costs—are unknown but thought to be significant.

The University HealthSystem Consortium recommends that institutions considering bar-coding system adoption fi rst develop a comprehensive strategic institutional plan for improving medication safety and that bar coding should be considered as one component of this overall plan. Bar coding can prevent errors but also can cause or contribute to medication errors. In an analysis of error reports submitted to MEDMARX, a national voluntary reporting database, 70 medication near-errors or “close calls” that occurred in 2000-2005 were thought to be prevented by barcode technology. Of these, 51 were errors in which bar coding detected that the wrong medication had been dispensed by the pharmacy. Th e MEDMARX database also contains 445 error reports in which bar coding was in place and the error was either caused by the technology or indirectly associated with the technology. Of them, 64% were either opportunities for error or actual errors that were intercepted by hospital staff before reaching the patient. For the 299 of the 445 reviewed reports that were actual errors, half originated in the dispensing phase of the medication use process, a third originated in the administering phase, and the remaining errors originated in the transcribing and prescribing phases. Errors that were directly or indirectly associated with bar-code technology included mislabeling errors (eg, affi xing a bar code with the correct medication but wrong dose; affixing a bar code with the incorrect medication but correct dose) and errors caused by inability to scan the bar code, failure to scan a bar code, manual overrides, technology failure, and “workarounds.” Workarounds refer to the use of a technology in ways that circumvent its safety advantages. For example, in the MEDMARX database, the most frequently reported nursing work-around involved scanning the patient’s identifi cation from the patient’s medical record rather than the patient’s wristband at the bedside.

Smart cards

Radio frequency identifi cation (RFID) technologies store unique identifi cation codes on chips—usually in the form of a “smart card.” Th e smart card stays with the patient (some RFID chips are actually implanted in patients; www., whereas bar coding is imprinted on armbands that are placed on patients. Th e patient places the smart card on a medication dispensing unit that contains an RFID reader that sends the code via a wireless link to a database that contains the patient’s treatment information and confi rms that the medication should be dispensed at that time. Because this technology is so new, outcome data is not yet available. In addition to its use in the medication administration process, RFID technology is being studied as a way to prevent the wrong blood products from being transfused, as a tracking method to identify the location of surgical sponges and newborn babies, and as a way to prevent drug counterfeiting.

Spectroscopic medication verification

ValiMed is a specially developed optical device that uses spectroscopy to recognize highrisk intravenous (IV) medications, including chemotherapy. After an IV medication is mixed, it is dropped into the device and exposed to ultraviolet light, which gives it a fl uorescent glow. Every medication has a unique “glow” based on its chemistry. ValiMed then determines if the correct drug has been prepared. With many chemotherapy agents having similar names, this technology reduces the likelihood that “lookalike, sound alike” wrong drug errors could occur.

Automated dispensing cabinets

Automated dispensing cabinets (ADCs) use technology to dispense oral and injectable medications in unit dose form. According to a survey conducted by the American Society of Health-System Pharmacy (ASHP), 72% of US hospitals have adopted ADCs, and nearly 90% of these hospitals have the cabinets linked to patients’ medication profiles. Historically, ADCs often were manually overridden; however, the ASHP surveyors found that overrides declined from 22% in 2002 to 13% in 2005. In Pennsylvania, 15% of all medication errorsreported to the state’s patient safety reporting system cited ADCs as the source of the errors, and 23% of these reports involved high-alert medications. Errors included wrong drug errors, stocking and storage errors, and medications administered to patients with documented allergies. Contributing factors included excessive use of overrides, failure to screen medication orders, and failure to recognize look-alike medication names from the ADC’s alphabetic “pick list” or storage compartments, which led to the selection of the wrong drug from the ADC. Because these systems are automated, there is a tendency to over-rely on them and omit manual safety checks. For instance, nurses sometimes become conditioned to expect that a certain medication will be in a specifi c ADC compartment. When new medications are added to the ADC and compartments are rearranged, the wrong medication could be selected if there is no physical check of the product and label. And although the process of restocking medications into an ADC is primarily a pharmacy function, in some oncology settings, nurses perform this activity and may inadvertently place a medication in the wrong compartment.

Recommendations to decrease the risk of ADC-associated medication errors include purchasing or upgrading to ADCs with integrated bar-code technology to reduce restocking errors, ensuring that all orders are screened by a pharmacist, placing only unit dose medications in the ADC (eg, avoid stocking multidose vials or bulk quantities of the same medication), placing allergy reminders on the cabinet screens, and limiting the override feature to emergency situations.

Smart pumps

Several hundred cases of IV medication errors that have been reported to the FDA involved IV pumps. When researchers observed 426 medications infusing via an IV pump and compared the medication, dose, and infusion rate on the IV pump with the prescribed medication, doses, and rate in the medical record, 285 (66.9%) had one or more errors associated with their administration. Overall, 389 errors were documented; 37 were rate-deviation errors, three of which were judged to be due to a programming mistake. If an error occurs with a high-alert medication that is being infused via an infusion pump, such as chemotherapy, it is possible that the error might not be detected early, and in some cases, may go undetected. “Smart pumps,” or infusion pumps with integrated dose calculation software, may help prevent pump-programming errors. The software in these pumps essentially transforms a conventional IV pump into a computer that has been programmed with standard medication concentrations and upper and lower dose limits. The pump alerts the nurse if it has been programmed outside of prescribed limits and will prevent administration of doses that are outside of these limits. Smart pumps log data about all such alerts, including the time, date, drug, concentration, programmed rate, and volume infused. This data often is examined as a component of continuous quality improvement and institutional safety programs. In addition, as required by the Joint Commission, all smart pumps have free-fl ow protection, a safety feature that prevents unintentional overdoses of medication or fl uid when the pump is turned off but still connected to the patient.

Technology in its infancy

Existing medication safety technology is sure to be improved upon, and new technology will be introduced with the goal of preventing medication errors, including chemotherapy errors. Unfortunately, technology cannot completely eliminate human error nor eradicate the desire of some nurses to circumvent or work-around safety features. Technology is only as good as its users.

Lisa Schulmeister works as a nurse in an office oncology practice; teaches the oncology nursing elective at LSU New Orleans; and provides practice, education, and legal consulting.