Human factors engineering to improve technology

Improve safety while cutting costs and time

Implementing technology that considers human factors engineering in inpatient pharmacy operations processes of medication ordering, preparation, and dispensing can help pharmacies become safe and reliable while cutting costs and reducing turnaround times. That's the conclusion of Cleveland Clinic pharmacy informatics specialist Scott Kulper, PharmD, and colleagues in a commentary published in the American Journal of Health-System Pharmacy.

Kulper says the error-prone inpatient pharmacy component of the medication-use process relies on interactions among providers, patients, information, and technology, and the majority of medication errors are a direct result of the intrinsic complexity of those interactions. While successful technologies can reduce the potential for human error by automating tasks requiring high levels of accuracy and repetition, he says, diligence must be taken as automating a faulty process will fail to address problems and will provide new error sources.

"The ability to reason and learn from past experiences allows us to develop robust systems that support human work," he says. "It is important to understand human limitations and abilities when designing a system or when trying to understand why errors are occurring within a system."

Normal accident theory, he says, characterizes complex systems and explains why they are prone to errors. According to that theory, complex systems have interactive complexity (the presence of events in the system that are unplanned, unfamiliar, and unexpected); common mode connections between components (interaction between a unit and more than one subsystem); and unfamiliar or unattended feedback loops (presence of many control parameters with potential interactions, indirect or inferential information sources, and limited understanding). Successful technologies, according to Kulper, reduce the potential for human error by automating tasks that require high levels of accuracy and repetition.

The commentary looks at various aspects of medication ordering, preparation, and dispensing and the role they can play in medication errors.

Interpretation and transcription. The reading and entry of an order into an information system are responsible for some 12% of medication errors and likely result from poor handwriting, workplace complexity, and confirmation bias. Computerized prescriber order-entry systems (CPOE) target these errors by eliminating the process. Orders entered into a CPOE system are checked for accuracy and completeness and compared against clinical decision-support rules. While some studies have demonstrated the ability of CPOE systems to reduce medication errors, others have disputed the claim that they cut errors. Kulper says that COPE-related errors underscore the need for appropriate system and process design and high-quality hardware.

Decision-support systems. Although most current CPOE systems have fully integrated medical logic alerts for order management, many decision-making processes are performed in the pharmacy, according to the commentary, and use of decision-support systems in sites with CPOE is highly variable. Kulper says there is little question that alerting systems need to be overhauled to provide clinicians with accurate and meaningful alerts. Each alert should be assessed for evidence, relevance, patient-specific risk factors, and frequency. And less serious alerts should be presented in an unobtrusive manner.

Preparation of IV medications. Automatic compounding machines are described as closed systems that prevent external contamination caused by multiple manipulations. Some contemporary compounding machines also use bar-coding technology to ensure correct product selection. So far, deployment of the machines in hospitals has been limited because they cost a lot to buy and not all IV compounds can be made in them.

Unit-dose repackaging. American hospitals use unit-dose distribution because it has a significantly lower error rate than multiple-dose systems. However, the commentary says, unit-dose repackaging has shifted dose preparation from nursing to pharmacy, creating a new source for pharmacy errors. Unit-dose repackaging is largely a manual process and in addition, bar codes must be added in facilities that use them, which means there is another manual process that can be a source of errors.

Kulper says bar-coded verification is included in next-generation high-volume repackaging machines. Such devices repack bulk medications into unit-dose packages including either patient- or batch-specific information. Bar-code verification is said to prevent improper loading of high-speed packagers and allow batch-specific information such as expiration dates to be tracked. "Bar-coding technology, which has an error rate of about 1 in 10 million, has great potential for reducing medication errors," he says.

Dispensing. Selection of a drug product for dispensing is repetitive and complicated by upstream processes and the interruptions of daily pharmacy work, the commentary authors say. The dispensing error rate has been determined to be between 1.7% and 2.1%. While hospitals often use unit-based cabinets for dispensing drugs, filling those cabinets is prone to error, accounting for 7.1% of the errors reported to Medmarx in 2002, more than five times the number of errors reported due to unapproved abbreviations.

Kulper says that bar-coding technology has been suggested as a potential remedy to cabinet stocking errors. In such a system, each item is labeled with a bar code and the code is scanned before the item is loaded into the cabinet. He notes, however, that institutions that have implemented bar-coded checks for loading the cabinets tend to check only one of the items to be loaded into a slot and not each individual item. Issues also arise when medications are accidentally mixed after stocking due to overflow or returns. Selection of incorrect medications and medication expiration are problems that can occur with open-shelf storage of drugs. Although many pharmacies have taken steps to reduce selection errors, confusion over similar drug names still accounts for 10-15% of all Medmarx reports.

Distractions. A national study of medication dispensing showed an error rate around 1.7% due to distractions such as answering telephones, fielding inquiries, and managing inventory. But a study of highly-automated mail-service pharmacies found a significantly lower error rate at 0.075%. Kulper says the difference was attributed to implementation of technology and segregation of duties within the pharmacy. He says hospital pharmacies should consider segregating duties as assigning staff dedicated to individual tasks will likely reduce medication errors.

Pharmacies and pharmacists are key stakeholders in the medication use process, Kulper concludes, and it is important for members of pharmacy departments to take leading roles in reducing medication errors and examining the effects of technology and process change on medication errors.