140: Medication Safety and Adverse Drug Events

Patient safety is of great interest to many regulatory groups, such as The Joint Commission, hospital administrations, health care professionals, and the general public. The interest has continued to grow since the publications of two reports by the Institute of Medicine. The first report, in 1999, focused on all medical errors and introduced measures necessary to ensure a safer health care system.⁸⁵ The second report, in 2006, focused on reducing medication errors and adverse drug effects.⁸³ These reports and others reveal that medications errors represent up to 25% of all medical errors.⁴⁹ Many health care institutions address medication safety through their pharmacy and therapeutics (also commonly called drug and formulary), medication safety, patient safety, and quality improvement committees. Table 140–1 shows a timeline of some important developments in medication safety.

Studies of medical errors, including those involving medications, are of a highly variable quality and usefulness. This occurs in large part because they address diverse populations, the definitions utilized vary, and a variety of data collection techniques, including observational studies and voluntary reporting, are employed.

One study estimates that 180,000 people die each year of medical errors,⁶⁴ and 60% of these injuries are probably preventable.¹⁷ The Institute of Medicine estimates that an additional 44,000 to 98,000 people die each year from medical errors.⁸⁵ However, despite having methodologic differences, both studies establish that medical errors cause thousands of deaths each year. Studies like these illustrate the “Swiss Cheese Model of Error:”⁹² that is, multiple errors must align with an initial error to lead to harm.

More than 1.5 million preventable adverse drug events (ADEs) may occur each year.⁸³ An ADE is an untoward event or outcome associated with the use of a drug. The definition of ADEs includes medication errors as well as adverse reactions to a drug and drug interactions. A medication error is “any preventable event that may cause or lead to inappropriate medication use or patient harm while the medication is in the control of the health care professional, patient, or consumer.”⁸¹ An adverse drug reaction is a sign or symptom related to use of a medication that results in unpleasant effects when an error has not occurred. An adverse drug reaction is also an ADE.⁵ Those ADEs and medication errors with serious effects lead to 3.1% to 6.2% of all hospital admissions.⁶²

In 1997, the cost of a single ADE was estimated to be $2000 to $5000 at academic medical centers.^10,22,23 Similarly, in 2012, a retrospective multi-center study of community hospitals estimated an increased cost of $3000 to overall care and increased length of stay of 3 days above expected when an ADE occurs.⁴⁶ Other studies estimated that the annual cost of ADE morbidity and mortality was greater than $77 to 177 billion in the ambulatory care setting,^33,50$2 billion in hospitals,^10,23 and $4 billion in nursing homes.¹⁶ These costs exclude legal or other costs that accrue to the patient or their families.

Deaths from Medication Errors and ADEs

Although medication errors are the most common cause of iatrogenic patient injury, less than 2% result in injuries. Nevertheless, the incidence of ADEs in hospitalized patients is estimated to range from 2% to 20%²² resulting in 7000 deaths annually in the United States.⁸⁵ A retrospective review of hospital death certificates by ICD 9 and ICD 10 codes from 1983 to 2004 revealed an increase of 361% in fatal medication errors and a 33% increase in deaths from ADEs occurring in a patient’s home.⁹¹ The hospital death certificate review also revealed an increase in fatal medication errors 32 times higher when prescription medications were combined with alcohol and or illicit xenobiotics.⁹¹ According to voluntary MedWatch reports to the US Food and Drug Administration (FDA), 17% of reported ADEs were associated with death, 7% were associated with permanent disability, and many others had serious complications. Women and elderly patients were at the highest risk.⁷⁸

National Coordinating Council for Medication Error Reporting and Prevention Taxonomy

A useful medication error taxonomy, developed by the National Coordinating Council for Medication Error Reporting and Prevention (NCC MERP), classifies medication errors according to severity of outcome (Fig. 140–1).⁸¹ Importantly, the categories of least severity (A and B) describe circumstances or events in which the potential to cause error exists, or the error occurs but does not affect the patient. These “near misses” are so frequent that they serve as a critical source of information related to systems problems and education about medication error, but are typically underreported and underappreciated. In one study of 154,816 errors reported by hospitals and health systems to MEDMARX from 1999 to 2001, most of the errors were in category C (47%) and resulted in no patient injury, whereas there were 19 errors in category I, contributing to or resulting in patient death, comprising 0.01%.⁹⁶ See Fig. 140–2 for a comparison of errors in the inpatient versus the emergency department (ED) setting based on the NCC MERP classification.

Medications Involved

FDA reports suggest that opioid analgesics and immune modulators are the most common medications in which errors resulted in death.⁷⁸ The medications most frequently involved in errors and ADEs reported to MEDMARX were insulin, anticoagulants, morphine, and potassium chloride.⁹⁶ These medications are considered high alert drugs according to the sentinel event alert system of The Joint Commission. Recently, The Joint Commission has focused more on all opioids (not just morphine) in order to promote safe use in the hospital setting.¹⁰⁵ Identifying characteristics of high-risk drugs include low therapeutic index, pharmacokinetic interactions, inherent undesirable effects, newly approved or “off-label” use, and direct-to-consumer promotion.¹³Table 140–2 lists medications commonly reported and their classes of errors.

The Medication Process

The Medication Process comprises the five stages in the sequence of ordering a medication to its delivery to the patient. The stages are prescribing, transcribing, dispensing, administering, and monitoring.⁶ For patients discharged from the ED or hospital, discharge, and follow-up is the sixth stage.²⁸ Although the medicating process has only six stages, there are multiple steps within each stage. The potential for error is high, increasing proportionately as the number of steps and their complexity increase. Figure 140–3 describes typical errors that occur at each stage in the process, along with prevention strategies.

The greatest number of errors resulting in preventable ADEs (medication errors) occurs at the first prescribing stage.⁷ In a study of serious medication errors, 39% were found to occur at the prescribing stage, 12% at transcription, 11% at dispensing, and 38% at administration.⁶⁵ In one study of 17,808 inpatient and ED medication orders, 6.2% of orders written involved a prescribing error and 30% of these were likely to harm the patient if they were not discovered.¹⁵ Another recent prospective study suggests that up to 43.8% of medications orders on hospital wards had a prescribing error discovered by ward-based pharmacists, despite the level of prescriber experience.⁹⁸ In the ED, most errors occurred in either the prescribing or administration phase.⁸⁹

Transcribing errors usually involve poor communication due to illegible handwriting, the use of trailing zeroes, or inappropriate abbreviations. Poor handwriting can also lead to confusion, particularly with regard to look-alike and sound-alike medications.⁹⁰ All information, whether printed, spoken, or otherwise communicated, must be transmitted in a clear, unambiguous, and timely fashion with avoidance of abbreviations. In its 2004 National Patient Safety Goals, The Joint Commission developed a minimum list of five sets of dangerous abbreviations, acronyms, and symbols that should not be used, and proposed preferred terms.¹⁰⁴ The Institute for Safe Medication Practices (ISMP) also published a comprehensive list.⁴⁸ By not using these abbreviations, the hope is that communication will be improved and that there will be no misinterpretation of poor physician handwriting.

Dispensing errors are most commonly due to substitution and labeling errors.⁹⁹ Errors at the stage of administering medications include incorrect drug, incorrect dose, incorrect route, and a drug given to the wrong patient. Computerized provider order entry (CPOE) was introduced to improve the medication process from prescribing to administering but has introduced new problems. See the Information Technology section for further discussion.

Monitoring and discharging with follow up are associated with fewer errors. This phase involves attention to liver and kidney function, checking xenobiotic concentrations, and attention to and evaluation of drug interactions and pharmacokinetic interactions.²⁸ Monitoring must be an ongoing process that begins when the patient receives the medication, regardless of where the patient is in the health care system, and continues as long as the individual continues to have the medication prescribed. In the acute stages of treatment, such as in the hospital or ED, the process is especially important and requires optimal three-way linkage between the pharmacy, laboratory, and physician.⁹⁷ This process must be maintained throughout the individual’s continued relationship with the health care system. In the outpatient setting, providers must be aware of signs and symptoms related to adverse drug effects while they are monitoring and following up patients. In a study of 661 patients in four primary care practices, patient reports of medication side effects to their physicians led to changed therapies in 76% of cases. A failure to identify medication-related symptoms and change therapy resulted in 21% ameliorable and 2% preventable ADEs. Ameliorable ADEs were defined as those in which “the severity or duration could have been reduced substantially had different actions been taken.”¹¹²

Important medication safety issues exist for children and older adults, a problem that is exacerbated by their underrepresentation or exclusion from clinical trials. It is estimated, for example, that only a third of the medications used to treat children have been adequately tested in this population.²⁵ Similar concerns apply to medications used in older adults and those individuals with specific underlying medical conditions that would have excluded them from trials.⁴⁷

Pediatric Considerations

Pediatric adverse events involving medications may occur from the prenatal period and throughout childhood from the neonatal period through maturation. Errors may occur in all settings: the home, in ambulatory care, the ED, hospital floor, pediatric intensive care unit (PICU), and neonatal intensive care unit (NICU), and by all caregivers. Approximately 1 in 6.4 pediatric orders results in an error that reaches the child.⁷² Because of the greater need for dose calculations to allow for weight-based dosing, dose-related errors are more likely to occur in children.⁵² Errors also occur in the home environment with nonprescription medications, but the true incidence is unknown. In one survey, the errors associated with home antipyretic use were estimated at almost 50%.⁷³ These errors were typically associated with underdosing, which is generally of little immediate harm.⁹⁵ In the ambulatory care setting, the incidence is unknown but one study identified “numerous” errors in prescription writing in a pediatric clinic.¹¹⁰ In the ED setting, the error rate was estimated at 10% of all pediatric charts.⁵⁷ There may be an increased risk of errors in children cared for in nonacademic or rural EDs compared with academic or pediatric EDs.⁵⁵ A retrospective chart review of 177 pediatric charts in four rural EDs identified 84 different medication errors in 69 of the 135 patients who received medications. The outcomes of these errors were in NCC MERP categories A to D (Fig. 140–1).⁷¹ A study of 18 pediatric ED voluntary event/incident reporting data showed that 19% of all incidents were related to medication events. A total of 94% were medication errors and 6% were ADRs. Errors included using incorrect weights, duplicate dosing, and miscalculations. Human factors contributed to 84% of the medication errors in this study.¹⁰⁰ See the Factors Affecting Human ­Performance section for further discussion.

Hospitalized children experience up to three times the rate of medication errors and potential ADEs as do adults.⁵³ The incidence of medication errors of hospitalized children is estimated at about 6% for all orders written; the majority (74%) of these occur at the prescribing stage and approximately 20% are classified as potentially harmful based on a 4-point Likert scale developed by the researchers instead of the NCC MERP classification previously described.³⁸ Pharmacists based on pediatric hospital wards, discovered that 5.9% of orders contained a prescribing error and were able to intervene through the order entry system.³⁰ Generally, children in the intensive care unit appear to be at higher risk for errors compared to adults and hospitalized children not in the ICU, presumably reflecting the increased complexity of disease and the medications used.⁵⁵ Incorrect dosing, especially with the intravenous route, is the most commonly reported error. Dosing of antimicrobials and intravenous fluids are the most common medications involved.^29,36,53 Errors and discrepancies found in hospital discharge instructions, with lack of complete medication reconciliation, could lead to patient harm⁵¹ (Chap. 32).

Geriatric Considerations

Medication errors in older adults also occur throughout the health care continuum: the home, ambulatory care, in nursing homes, in the assisted-living setting, and in the hospital (Chap. 33). Adults older than 65 years of age have a relative risk of 2.37 for drug complications and 4.12 for medication errors compared to adult patients younger than 65 years of age.¹⁷ The incidence of medication error at home with nonprescription medications is unknown but, for reasons outlined below, would be expected to exceed that of the younger adult population. Using a variety of methodologies, ADEs were evaluated for a 12 month period in a multispecialty ambulatory care practice in a cohort of 27,617 Medicare enrollees, equivalent to more than 30,000 person-years of observation. Extrapolating their findings, to the estimated 38 million Medicare enrollees (those ≥ 65 years), would predict nearly 2 million ADEs annually, of which more than 25% would be considered preventable and about 180,000 fatal or life threatening.⁴³ A recent study estimates that there are about 265,802 ED visits in patients older than 65 years of age for ADEs and 99,628 of these patients require admission to the hospital. In the same study, patients older than 80 years of age constituted half of those admitted to the hospital. Medications or classes of medications involved in two-thirds of the hospitalizations were diabetic and antithrombotic medications.¹⁹

There are more than 1.5 million nursing home residents in the United States. The average such resident uses six different medications, and 20% use ten or more.¹⁴ Extrapolating the findings of a study of 18 community-based nursing homes in Massachusetts over a 1 year period predicts 350,000 ADEs annually, more than one-half of which would be preventable.⁴² Fatal or life-threatening ADEs would represent 20,000 of these predicted events of which 80% would be preventable. Approximately one million other seniors live in assisted-living facilities, and are vulnerable to medication errors for a variety of reasons, including inadequate physician support, inadequately trained staff, and staffing shortages. ADEs cause 10.5% of hospital admissions for geriatric patients and are the most common type of adverse event occurring in hospitalized elderly patients.^66,114 These drugs result in nearly 50% of ADE-related visits to the ED but are only prescribed during 9.4% of outpatient visits.²⁰

Addressing these issues is becoming ever more important as the US Census Bureau predicts a rise of 62 million in the number of Americans 65 years of age or older by the year 2025 and a 68% increase in the 85 years of age or older population that may be at an even higher risk.⁴ Advancing age brings with it several important considerations from the point of view of medication safety. Medical comorbidity increases with age, and therefore an increasing likelihood of receiving multiple medications. With more medications, the number of potential errors and interactions increases. Frailty and cognitive decline in older adults may result in errors following self-administration. Alterations in medication absorption, metabolism, distribution, and elimination may all affect the efficacy of the medication (Chaps. 9 and 33).

Criteria were developed in both the United States (Beers)^11,12,35 and Canada⁷⁴ to determine appropriateness of medication prescribing for nursing home residents. Forty eight medications or classes of medications to avoid, and medications to avoid in the presence of 20 diseases/conditions were identified.³⁵ The prevalence of inappropriate medication use in older adults is estimated to be in the 12% to 40% range.¹⁰² One in five prescriptions given to elderly individuals are considered inappropriate based on Beers criteria, with diphenhydramine and amitriptyline being the most common medications prescribed inappropriately.⁸⁷ Older adult patients medicated with benzodiazepines, for example, have a four-fold increase in falls (Chap. 33).

One of the leading causes of medication errors is human performance deficit. Almost invariably a human action will precede the ADE, and this temporal contiguity of action and consequence typically generates a tendency to blame someone. In most cases the blame will fall on the last person to have had contact with the patient. In recent years, however, a consensus has emerged that blaming people for errors is counterproductive. The number of ADEs that result from egregious behavior is very small, and more often than not an explanation for the fault will be found within the system that allowed the error to occur. Attempts to understand the nature of these faults, and to correct them, while being sensitive to the potential for unintended consequences, is the most appropriate response. Several processes can be used to respond to errors or system-problems. These include root cause analysis, clinical incident analysis, failure mode and effect analysis, and Lean Six Sigma (LSS).

Root cause analysis (RCA), a term originally used to investigate major industrial events, is a technique that provides a structured, process-oriented analysis of sentinel events. The Joint Commission in accredited hospitals mandated its use in 1997. It is a time-consuming process requiring multidisciplinary teams with specialized training, and is subject to bias and methodologic limitations.¹⁰⁸ Nevertheless, a judiciously conducted RCA may provide insights into systemic failures underlying the ADE, and identify areas that require change. An alternative approach, a clinical incident analysis protocol, was developed that more appropriately shifts the emphasis from the individual to the system.¹⁰⁷ The clinical incident analysis protocol utilizes seven factors as the basis for an investigation. Some organizations have developed a hybrid combining these two approaches. Both RCA and clinical incident analysis are conducted retrospectively, and therefore subject to retrospective bias

An alternative approach is failure mode and effect analysis (FMEA), which proactively attempts to identify potential errors, to initiate preventive measures. A multidisciplinary group is utilized in the FMEA approach to identify a process or subprocess that needs analysis, to identify the steps of the process and determine the risk/likelihood/severity of failure of each step. Once this phase is accomplished, the team prioritizes a high-risk step and conducts an RCA to make recommendations on redesigning the step. The establishment modifications are analyzed in their performance to determine change and decrease in risk. This process is designed as a quality control and assurance to protect the proceedings from legal investigation.³

Other processes are being introduced into the health care industry to improve quality of care and safety and reduce costs. Two such approaches are Lean and Six Sigma, sometimes utilized together as LSS. Both were initially described in the Toyota Production System and have been adapted to health care. Define, measure, analyze, improve, and control are all steps related to LSS. It involves defining the process that needs evaluation, determining how to measure the process, analyzing the data collected based on the determined measures, forming an improvement plan based on the analysis, and putting a new process in place with the goals of developing a lasting culture regarding the newly improved process with elimination of waste.^1,116 Each of these steps has principles to guide the use of LSS. The Agency for Healthcare Research and Quality has made specific forms to aid health care facilities when adopting these approaches to safety and improvement.⁶³

To reduce medication errors, each hospital should simplify, standardize, and stratify processes and communication. The medication process should be carefully automated with computer order entry and bar coding as extensively as the system will permit. While these information technology solutions will aid the medication process, nothing is perfect and each change introduces new problems. See the Information Technology section later. Limitations of attention and vigilance should be understood and the reporting of errors in a nonpunitive environment should be encouraged.⁶⁵ Improving information access, error proofing, reducing reliance on memory, enhanced training, and the use of buffers or redundancy in an attempt to prevent errors should also be encouraged.⁶⁴ Each time a medication is given, the focus should be “right drug, right dose, right route, right patient, at the right time.”

In 2003, the American Academy of Pediatrics Committees on Drugs and on Hospital Care developed a position statement regarding the reduction of medication errors in the pediatric population. The recommendations can be adapted to all hospital settings and to patients of all ages. These recommendations include appropriate staffing, utilization of the resources in the pharmacy, standardization of hospital equipment and protocols, and the development of a nonpunitive, barrier-free system to report and easily track errors.¹⁰³ Other more recent groups also encourage reduction of medication errors and participation in medication safety. The American College of Medical Toxicology encourages medical toxicologists to use their expertise and training to aid health care systems in reduction of errors and other pharmacy and therapeutics issues.²

For older adults, prescribers should review medication indications and avoid age bias. Polypharmacy should be limited, with medications prescribed only as necessary. Dosing should be adjusted, as needed based on renal and hepatic function. Medication review by pharmacists can aid in decreasing the use of potentially inappropriate medications and in decreasing the number of medications prescribed (polypharmacy).⁷⁶ In the hospital, outside of the emergency department and emergent situations, pharmacists review all medication orders, when available, and verify the medication as appropriate prior to it being dispensed. Pharmacists can also aid in medication reconciliation at transitions of care and upon hospital discharge to prevent complications of polypharmacy.^40,59

Human factors lead to many medication errors. In fact, it is not surprising that human performance deficits are the primary causes of medication errors in such an extremely complex medical environment. A performance deficit means that the individual making the error had the prerequisite knowledge to avoid the error but failed to do so. There are numerous variables that contribute to performance deficit, including many ergonomic issues such as workload, distractions, resource limitations, and staff shortages. This environment is both burdened and enriched by many inherited properties. Human factors and ergonomics theory draws on a variety of disciplines, including industrial engineering, industrial psychology, cognitive psychology, and information technology. Much can be done to optimize the interface between humans and the work environment and to ensure that systems operate more efficiently. As a general principle, it would be preferable if the dominant purpose in designing medical devices and processes was that they fit human users, and not the converse. Table 140–3 lists some of the more common human performance deficits. Such errors often manifest as simple slips of action, or execution failures, arising from distraction by something other than the task at hand.⁹² Vigilance is better maintained in individuals who are well rested and working without interruption or distraction in a well-designed environment. Fewer medication errors occur in optimally designed environments.⁶⁵

Human performance deficits such as diminished memory, sleep deprivation, depression, and distractions contribute to errors. Resident physician mood and lack of sleep are also factors in performance deficits. Depressed residents made 6.2 times as many medication errors as nondepressed residents.³⁴ Sleep deprivation and improper supervision were highlighted initially after the Libby Zion case, in which a patient on phenelzine developed fatal serotonin toxicity following the administration of meperidine by a sleep-deprived and inadequately supervised intern. Interns working a ­traditional schedule of call every third night with extended work shifts (36 hours) made 35.9% more serious medical errors than those with the current reduced work schedule.⁶⁰ In particular, there were 20.8% more serious medication errors during the older work schedules compared with the current reduced hour schedule.⁶⁰ In 2008, an Institute of Medicine study addressed resident work hours and patient safety. This report recommended residents be provided with designated sleep time during each day and rest periods each week in order to decrease the risk of fatigue-related medical errors.⁸⁴

A particularly important goal for human performance deficit is the reduction of cognitive load and distractions. Many medication errors originate from cognitive failings because of interruptions, distractions, inexperience, or simple overloading—referred to as performance deficit. In the ED, increased cognitive load such as related to increased number of boarding patients, increased number of medication orders, and increased number of medications to be administered all lead to increased medication errors.⁸⁹ Further efforts must be directed at strategies to reduce cognitive failure. The adoption of some very simple strategies based on human factors engineering principles will reduce error, such as simplification of the number of steps involved, reducing reliance on memory, applying cognitive forcing strategies,²⁷ and using cognitive aids. Research into reducing distractions during medication administration has focused on adapting principles from the airline industry, “the sterile cockpit principle.”³⁷ This principle limits interruptions and distractions that could interfere with proper performance of a critical task or tasks. In the airline industry, the use of this principle occurs during takeoff, taxi, and landing. In mediation administration, this principle involves the use of “do not disturb” signs or vests worn by the nurse administering medications, no conversations to disrupt the medication nurse during this important task, and the other nurses on the unit answering phone calls or patient/family questions. This approach reduces medication administration errors by improving human performance.^37,93 One particularly useful aid to reduce cognitive load is the color-coded Broselow-Luten system, for pediatric medication dosing.⁶⁹ This approach has the potential for further development to improve the safety of nonprescription medications and other potentially dangerous products used in the home.

Another factor that contributes to errors is the assignment of a new role to a provider such as asking physicians to dispense or administer medications. Pharmacists are the only professional group formally trained and experienced in dispensing medication, and not surprisingly, their presence is associated with a lower medication-dispensing error rate.^58,67 Nurses administer medications because they receive such training and the administration should be restricted to them except during specific circumstances such as procedural sedation.²⁸

The two most common factors contributing to prescribing errors are related to knowledge deficits. These deficits include lack of knowledge about the drug and about the patient.^65,68 Knowledge deficits about medications are likely due to the number new medications, dosage formulations, and indications. With the large number of medications prescribed and the continuing and frequent release of new medications, it is difficult to know all of the relevant clinical pharmacology of each medication, including interactions, side effects, and contraindications. Decision support, most commonly in electronic format, should be contemporaneously available to health care professionals to assist in decreasing prescribing errors.⁹⁰ However, electronic pharmacopeias, as a form of decision support, can have errors and omit warnings, such as recently discussed with extended-release or long-acting opioids.⁶¹

Knowledge deficits may also affect dosing calculations and drug ordering. This may be related in part to the lack of formal education regarding the medication process during medical school training. Only 10% of medical students answered dose calculation questions correctly.¹¹³ However, students in their final year of training performed much better than students in the lower years.¹¹³ It is unclear where, when, and how the students obtained the knowledge in the higher years. Specific training is needed in both of these areas of calculation and order writing. Improvements have been demonstrated following a short educational intervention,⁸⁶ but it is unclear how long this knowledge is retained. Generally, there appears to be a tacit assumption that these skills will be acquired during clinical training, but this study suggested that this knowledge might not be acquired independently and that specific training is indicated.

The prevailing emphasis in physician training is on knowledge acquisition. Less time is spent inculcating critical thinking skills and/or teaching reasoning, the assumption being made that these are passively acquired during the process of training. Although this is partially true, it does not exclude opportunities to improve these cognitive faculties by direct intervention.²⁶

Improved team communication should result in fewer medication errors. Orders should be written clearly, or CPOE (discussed in detail in the Information Technology section) should be used. Orders should optimally include the indications for the medication. Both generic and trade names should be included in orders so that confusion with look-alike/sound-alike drugs may be avoided. Abbreviations should have limited use and trailing zeros should be eliminated. Verbal orders should be used sparingly due to the high risk of sound-alike medications and confusion with dosing.⁹⁰ If a verbal order is necessary, then it should be restated immediately to the ordering clinician to ensure that it is correct and restated from the physician to the nurse in a “closed loop” communication process.

Communication theory should receive more emphasis in health care training. Good communication skills both within and among disciplines and especially between practitioners and their patients will limit errors. As patients are admitted, transferred within, and discharged from health care facilities, it is essential that precise communication as to what medications, strengths, and doses the patient is receiving occur. This process is known as medication reconciliation. Any changes that have occurred must be accurately recorded. This effort was adopted by The Joint Commission in 2005 and continues to be an important National Patient Safety Goal. As with other aspects of patient safety, these issues should be formally introduced into the education curriculum. Studies have shown that pharmacists are an integral part of optimal medication reconciliation.^40,59

Information technology (IT) in health care has been ponderously slow to develop compared with its use in other organizations, but it is now gathering momentum and has obvious potential to improve patient safety.^8,82 On the other hand, as considerable gains may be made, new technology can also be expected to introduce new types of errors.^21,115 The US Pharmacopeia announced in 2004 that nearly 20% of 235,159 medication errors reported to MEDMARX involved computerization or automation.⁹⁶ One study evaluating CPOE in a tertiary care hospital found that the system actually facilitated a wide variety of medication error risks.⁴⁴ Common errors associated with order entry include wrong medication selection and input of incorrect data (patient weight, creatinine clearance).¹⁰⁶ Another study identified 24 different types of failures associated with CPOE but suggested that many could be easily corrected, in particular by concentrating on organizational factors.⁵⁶ However, more detailed insights have been offered into why process-supporting IT systems fail.¹¹¹ Much of the data involved in the medication process are relatively straightforward, amenable to rapid and efficient processing, including crosschecking with patient medication history and evaluating for drug interactions (an example of decision support). A study reviewing 28 trials of point-of-care decision support report showed a 4.2% improvement in patient care with a trend toward better improvement in care when a response was required from prescriber¹⁰¹

Past studies demostrate that computerized order entry with decision support reduces the incidence of serious medication errors by 50% to 55% once the transitional instability has passed.^9,70 This approach mainly reduces errors at the prescribing stage of the medication process and, while likely to prevent 80% of prescribing errors that led to no patient harm, the errors most likely to cause patient harm may not be as amenable to reduction as these errors may be related to dispensing, administration, or monitoring.¹⁵

In high-risk populations, CPOE reduces medication errors. In children, CPOE with substantive decision support reduced ADEs (including medication errors) and potential ADEs in an inpatient pediatrics ward.⁴⁵ It also decreased the rate of nonintercepted medication errors by 7% although there was no change in the rate of patient harm.^53,109 In the NICU, CPOE eliminated all calculation errors as decision support included an automatic dosage calculator.²⁴ However, these decisions support systems cannot improve human errors due to entering incorrect weights, incorrect units of weight (kilograms versus pounds), or medications entered on the wrong patient.¹⁰⁶ In geriatric medication errors, CPOE resulted in less potentially inappropriate medication prescribing through decision alerts, which led to a pharmacist call to the prescriber that offered alternative choices of medications if possible.⁷⁷ Decision support in CPOE can also lead to too many alerts when orders are placed, causing alert fatigue and resulting in providers entering an answer to proceed past (or “blow-by”) the alert without reading or comprehending the alert. A study reviewing an alert about critical drug–drug interactions with warfarin found that only 19% of providers and 9.7% of pharmacists responding appropriately to the alert and changed the medication order.⁷⁵

Use of barcoding for dispensing and administration of medication ensures that the correct drug is given to the intended patient at the dose intended. Barcoding significantly decreased the relative risk of targeted, preventable ADEs by 47% to 50% in an NICU.⁸⁰ The risk of opioid related ADEs were also reduced when barcode medication administration was implemented.⁷⁹ This technique decreases the amount of time nurses spend administering medications in the ICU, allowing them to reallocate that time to other direct patient care activities.³¹ As with all technology, new problems arise with the implementation of barcoding and can lead to errors. Such problems range from inability of the scanner to read the barcode on the medication or on the patient identification band, downtime processes, mismatch of barcodes due to new manufacturer barcode, or new or aging scanners, and many more.³²

Unit dose dispensing systems, usually in association with CPOE and barcoding of medications, reduced monthly errors from five to none in the inpatient setting.³⁹ The package sent from pharmacy is prepared to administer to a specific patient at the appropriate dose, eliminating the need for the nurse to draw up medications resulting in fewer errors. Other dispensing systems, like automated dispensing cabinets (ADCs), have also been linked to CPOE, barcoding, and can allow for a pharmacist to verify correct drug, correct dose, patient allergies, and use of the medication prior to a drug being dispensed from the cabinet. Theses ADCs can decrease dispensing errors. However, errors can occur when using ADCs. One such problem, “drug override,” can occur, particularly in an emergency situation or as a work-around. This allows a drug to be dispensed prior to pharmacist verification. Another error, look-alike medication errors, can occur due to the small screens on the ADCs. Errors have also been known to occur when the ADCs are stocked such that an incorrect drug is placed in an incorrect bin of the ADC.⁴¹

Clinical pharmacists are now employed in high risk areas such as the ICUs, the pediatrics services, and EDs to identify and prevent medication errors. In one ICU study, the input of a clinical pharmacist during rounds saved an estimated $270,000 annually in costs of rehospitalization due to ADEs.⁶⁷ The involvement of a clinical pharmacist in work rounds of an adult ICU reduced preventable ADEs by 66%.⁶⁷ The introduction of clinical pharmacists in pediatric services is credited with a 94% reduction of potential ADEs and medication errors.⁵³ In particular, the addition of pharmacists in the PICUs reduced the serious medication error rate by 80%.⁵⁴ One study showed that pharmacists in three EDs identified 2200 interventions with an estimated savings of $488,000. This savings came from lower-cost medications, reduced length of stay, and fewer readmissions.⁹⁴ Another study showed that there were 16 errors per 100 medication orders in the control group (without a pharmacist), while the intervention group (pharmacist present in the ED) only had five errors per 100 medication orders, resulting in a 67% reduction.¹⁸ A recent study suggests that the availability of ED pharmacists has even more impact on reducing errors by their assistance for questions, discussion, and consultation rather than in the review of medication orders.⁸⁸ And, as previously discussed, hospital pharmacists have been able to improve medication reconciliation across the continuum of care for patients.⁵⁹

However, fiscal restrictions will inevitably mitigate against the expansion of service pharmacists as cost-benefit arguments will be applied. An unintended consequence of having a clinical pharmacist present may be that nurses, residents, and attending physicians will always defer to them and consequently spend less time developing and maintaining their own skills for off-hour periods when such help may not be available.

Clinical and medical toxicologists are in a unique position to investigate causes and help decrease the incidence of both medication errors and ADEs in the institutions where they work. With their specialized knowledge and training in pharmacology, poisoning, and clinical medicine, both physicians and pharmacists with advanced training in toxicology can also aid in the prediction, identification, and management of ADEs. Medical toxicologists should optimally be involved in Pharmacy and Therapeutics and Medication Safety committees (or related efforts) at their institutions. They should educate the providers at their hospitals as to means to reduce errors and encourage diligent reporting of medication errors to the hospital and the national databases such as MedWatch at the FDA. They should also encourage drug manufacturers to be diligent in alerting physicians with regard to identified problems.²

• Medications are the principal commerce of modern medicine, and medication safety is of paramount importance to health care systems. The delivery of medications safely to patients is a complex process and is particularly important in pediatrics and geriatrics.

• Health care professionals should be aware as to why errors occur and how they can be prevented in order to improve patient and medication safety.

• Information technology holds promise for significant improvement in medication safety. Focus should be on CPOE, barcoding, unit dose dispensing, avoidance of look-alike/sound-alike medications, and education.

• Use of clinical pharmacists and updated electronic-based literature should be encouraged.

• Communication and nonpunitive local and national reporting must be encouraged in order to learn from errors.

• Patient safety requires a collaborative effort particularly with the medication manufacturer, but also with federal regulation authorities, independent research organizations, error theorists, hospital administrators and managers, information technologists, nurses, cognitive psychologists, human factors ergonomists, and chronobiologists and all of the health care professionals involved in patient care.

• Clinical and medical toxicologists have a significant role in medication safety.

Acknowledgment

Patrick Crosskerry, MD, PhD, contributed to this chapter in previous editions.

References