Determine whether the statements below are part of the seven rights of medication administration

The research review targeted studies involving medication administration by nurses. This excluded several studies that assessed medication administration errors without differentiating whether the errors were associated with physicians, assistants, or nurses. None of these studies included interventions.

Thirteen studies explicitly reported types of MAEs associated with nurses. The incidence of MAEs was detected either formally through incident reports, chart reviews, or direct observation, or informally through anonymous surveys. Two studies conducted retrospective assessments, one using medical records43 and the other malpractice claims.80 Seven studies assessed self-reported MAEs from a nationally representative database44, 81–83 or self-reported errors using a nationally representative sample.84–86 None of these self-reported MAEs were verified. Eight studies assessed MAEs using direct observation of the medication administration process.24, 37, 78, 87–91

The incidence of MAEs varied widely with the different research designs and samples. Using chart reviews, Grasso and colleagues43 found that 4.7 percent of doses were administered incorrectly. Direct observation studies placed the estimate of total incorrect doses between 19 percent and 27 percent,87 and when an extra review was done to separate the errors into stages of the medication process, between 6 percent and 8 percent of doses were in error because of administration. The majority of types of MAEs reported were wrong dose, wrong rate, wrong time, and omission. All of the studies reviewed here reported wrong drug and dose, but varied across the other types of MAE categories (see Evidence Table 1); this was dependent upon the study methodology.

Five studies evaluated self-reported MAEs, involving incident reports and informal reports.38, 44, 81, 82 The most common types of reported errors were wrong dose, omission, and wrong time. Four of these studies38, 81–83 assessed a large secondary, nationally representative database containing MAEs reported to the MEDMARX database over five years.38, 81, 82, 44 found in the error reports submitted by nursing students that the majority of MAEs were associated with omission, wrong dose, wrong time, and extra dose. Of the reported contributing factors, 78 percent were due to the inexperience of the nurse. The Beyea and Hicks81, 82 studies looked at errors associated with the operating room, same-day surgery, and postanesthesia; they found the majority of errors attributable to administration but did not classify them by error type. The other study reviewed 88 incident reports from a long-term care facility submitted during a 21-month period. It found that the majority of MAEs were associated with errors involving interpreting or updating the medication administration record, delayed dose, wrong dose, or wrong drug.92 A separate component of this study surveyed administrative and clinical nurses and found that they believed the majority of medication errors occurred at either the administration or dispensing stage.

Two other studies assessed the type of MAEs reported by nurses in nationwide surveys.84, 85 While the majority (57 percent) of errors reported by critical care nurses involved MAEs, an additional 28 percent of reported errors involved near misses. Medication administration errors involving wrong time, omission, and wrong dose accounted for 77.3 percent of errors, while wrong drug and wrong patient accounted for 77.8 percent of near misses. The most frequent types of medication errors were wrong time (33.6 percent), wrong dose (24.1 percent), and wrong drug (17.2 percent), and the three most frequent types of near misses were wrong drug (29.3 percent), wrong dose (21.6 percent), and wrong patient (19.0 percent).85 Many of the reported MAEs in ICUs involved intravenous medications and fluids.84 In these surveys, the nurses who reported making errors described between two and five errors during a 14-day period.

At the more advanced stage of incident reports, one study reviewed 68 malpractice cases involving MAEs in Sweden.80 Among the cases reviewed, the majority of MAEs made by nurses involved wrong dose. When the nurses delegated the drug administration to subordinate staff, the majority of MAEs involved wrong drug or wrong concentration of a drug. Errors, which were reported to the immediate supervisor, were also reported to the physician in 65 percent of cases. The reported causes of MAEs were lack of administration protocols, failure to check orders, ineffective nurse supervision when delegating administration, and inadequate documentation.

One study assessed medication errors using 31 medical records of patients discharged from a psychiatric hospital and found a total of 2,194 errors.43 Of these, 997 were classified as MAEs (4.7 percent of all doses, and 66 percent of all errors). Of these, 61.9 percent were due to scheduled doses not documented as administered, 29.1 percent as drugs administered without an order, 8 percent as missed doses because of late transcription, and 3 percent resulting from orders not being correctly entered in the pharmacy computer.

The number of studies using direct observation of medication administration is increasing in response to the concern about the accuracy of other sources of data. Ten studies were found, only three of which were done in the United States. While we attempt to summarize across these studies, it is difficult to determine consistency across studies as each focused on different sets of errors (some only intravenous errors, some included gastrointestinal tube technique) and were conducted in different settings. In many of the non-U.S. studies, nurses dispensed drugs from ward stock and prepared many of the intravenous solutions for administration.

Three observational studies were conducted in pediatric units—one in France,78 one in Switzerland,25 and one in the United States.24 Buckley24 reported 52 of the 263 doses (19 percent) observed to be in error, but only 15 (6 percent) of those were in the administration stage. Those 15 were nearly evenly divided among wrong dose, wrong time, wrong technique, and extra dose categories. Prot78 reported nearly 50 percent more MAEs. Of the 1,719 observed doses, 467 (27 percent) were in error, including wrong time; excluding wrong-time errors, the error rate was 13 percent of doses. The categories with the most MAEs in Prot’s study were wrong time, wrong route (GI tube versus oral), wrong dose, unordered drug, wrong form, and omissions. Schneider and colleagues25 reported an overall 26.9 percent error rate with wrong-time errors, and an 18.2 percent rate excluding wrong-time errors. Common errors in addition to wrong time were wrong dose preparation and wrong administration technique.

The incidence of intravenous drug errors was observed in three studies, one in England,89 one in Germany,90 and one in both countries.37 About 50 percent of the doses were determined to contain at least one error. Compared to other studies, this rate is surprisingly high, and it included preparation technique errors (selection of diluent/solvent) as well as administration errors (rate of bolus injection and infusion rate). Part of the explanation may come from institutional (type of pharmacy support available) and professional training factors. (German nurses are not trained to do intravenous medications.)

Three studies focused on medication administration in ICUs in the United States,45 in France,91 and in the Netherlands.94 Kopp and colleagues45 looked at all medication errors and report that 27 percent of doses were in error; of these 32 percent could be attributed to the administration stage. Within the MAEs, most were omitted medications; the rest were evenly distributed among wrong dose, extra dose, and wrong technique. Few wrong-time errors were noted. Tissot91 and van den Bernt94 examined only administration stage errors and reported very different rates. Tissot reported 6.6 percent of the 2,009 observed doses were in error, most from wrong dose, wrong rate, and wrong preparation technique. Excluding wrong-time errors, van den Bernt reported a 33 percent error rate that included preparation errors with diluent/solvent issues, infusion-rate errors, and chemical incompatibility of intravenous drugs. It is likely that the differences in rates across these studies are due to the range of error types observed in each study as well as the varying responsibilities of nurses in the three countries.

The most extensive observation study, by Barker and colleagues,87 conducted observations of medication administration in 36 randomly selected health care facilities (acute and long-term care) in two States in the United States. Of the 3,216 doses observed, 605 (19 percent) contained at least one error. Nearly half of those errors were wrong-time errors. Other common types of errors included omission, wrong dose, and unauthorized (unordered) drug. In a much smaller study conducted in the Netherlands, Colen, Neef, and Schuring88 found an MAE rate of 27 percent, with most of these wrong-time errors. The rate of MAEs without wrong time was approximately 7 percent, and most of those were omissions.

Information from these research studies forms a consistent picture of the most common types of MAEs. These are wrong time, omissions, and wrong dose (including extra dose). Rates of error derived from direct observation studies ranged narrowly between 20 and 27 percent including wrong-time errors, and between 6 and 18 percent excluding wrong-time errors. The alarming exception to this was the nearly 50 percent error rate in observation of intravenous medication in ICUs in Europe.

Medication safety for patients is dependent upon systems, process, and human factors, which can vary significantly across health care settings. A review of the literature found 34 studies that investigated some aspect of working conditions in relation to medication safety.

Systems factors that can influence medication administration include staffing levels and RN skill mix (proportion of care given by RNs), shift length, patient acuity, and organizational climate. There were 13 articles presenting research findings and three literature reviews. The major systems/organizational factors included nurse staffing, workload, organizational climate/favorable working conditions, policies and procedures, and technologies enabling safety or contributing to MAEs.

Medication administration is a key responsibility of nurses in many settings, and three studies assessed the relationship between nurse staffing, hours of nursing care in hospitals, RN skill mix, and medication errors. Two studies associated the total hours of care and the RN skill mix at a patient care unit to reported medication error rates in those units; one study used 42 units in a large Midwestern hospital95 and the other used 39 units in 11 small hospitals.96 Rates of MAEs, when the number of doses was the denominator, were highest in medical-surgical and obstetric units; when patient days were the denominator, the highest rate was in ICUs. In both studies the type of unit was controlled and the rate of reported medication errors declined as the RN skill mix increased up to an 87 percent mix. A third study of nurses in ICUs in 10 hospitals found an inverse relationship between rates of medication errors and staffing work hours per patient day in specific settings (e.g., cardiac ICUs and noncardiac intermediate care settings). A little over 30 percent of the variance in medication error rates resulted from the variance in staffing work hours per patient day.97

Other studies conducted prior to 1998 did not find a relationship between staffing and medication errors. Three literature reviews,30, 39, 98 concluded that the direct evidence for a relationship between staffing and MAE rates was inconsistent. Nurses’ perceptions of the impact of staffing or workload on medication errors, however, is quite consistent.

These findings are consistent with three studies and two literature reviews on the impact of heavy workloads, a component of nurse staffing, on errors. In one survey of nurses in 11 hospitals, both pediatric and adult nurses reported staffing ratios and the number of medications being administered as being the major reasons why medication errors occur.58 A second survey found that nurses from Taiwan also indicated that workload was a major factor in medication errors.93 Beyea, Hicks, and Becker81, 82 and Hicks and colleagues38 analyzed MEDMARX data for medication errors in the operating room, postanesthesia, and in same-day-surgery units. Most of these errors involved nurses (64–76 percent) and medication administration (59–68 percent). In all three sets of error reports, workload increases and insufficient staffing were noted to be causes of errors.

The effect of heavy workloads and inadequate numbers of nurses can also be manifested as long workdays, providing patient care beyond the point of effective performance. In a national survey by Rogers and colleagues,99 self-reported errors by nurses found that the likelihood of a medication error increased by three times once the nurse worked more than 12.5 hours providing direct patient care. Among nurses working more than 12.5 hours, the reported errors, 58 percent of actual errors and 56 percent of near misses were associated with medication administration.

Other findings support the importance of adequate nurse staffing and understanding the impact of shift work in decreasing medication errors. A review of incident reports found that the major contributing factors to errors were inexperienced staff, followed by insufficient staffing, agency/temporary staffing, lack of access to patient information, emergency situation, poor lighting, patient transfers, floating staff, no 24-hour pharmacy, and code situations.44 Certain aspects of shift work can also impact medication safety, as shown in a review of research conducted in the 1980s and early 1990s that indicated that there was a difference in the number of errors by shift, but no difference in the number of hours worked (8 versus 12 hours). However, there were more errors with nurses working rotating shifts.30

Other systems/organizational issues include the presence of favorable working conditions, effective systems, policies and procedures, and technologies that enable safety or contribute to MAEs. An assessment of medication administration behaviors of 176 nurses in rural Australia, using structural equation modeling to test the association between organizational climate and the administration behaviors of nurses, found that the variable “violations” was the only variable with a direct contribution to MAEs, but there was no direct linkage to actual errors. While it was not possible to determine the effect of organizational climate on violations, distress was positively associated with violations, while quality of working life, morale, and organizational climate had a negative association. The organizational climate was found to be linked with safety behavior.100 Hofmann and Mark101 did find that the safety climate on patient care units was linked to the rate of harm-producing medication errors in a study using data collected from 82 units in 41 hospitals. Higher overall safety climate was related to lower rates of medication errors and urinary tract infections.

Lack of appropriate policies, procedures, and protocols can impact medication safety, as seen in a few small studies. In a study of malpractice cases, medication errors were associated with lack of administration protocols and ineffective nurse supervision in delegating administration.80 However, even when policies are in place, they may not necessarily improve safety. For example, a review of two studies in the literature found that medication errors did not necessarily decrease with two nurses administering medications (e.g., double-checking).30 In addition, appropriate policies may not be followed. Double-checking policies are commonly used as a strategy to ensure medication safety. When errors occurred under such policies, failure to double-check doses by both pediatric and adult nurses 58 and nurses in a Veterans Affairs (VA) hospital102 were reported. However, research presented in two literature reviews offers somewhat conflicting information. In the first review of three studies, following double-checking policies did not necessarily prevent errors.39 Yet in the other review, failure to adhere to policies and procedures was associated with errors.30

Process factors that influence medication administration include latent failures that can instigate events resulting in errors, such as administrative processes, technological processes, clinical processes, and factors such as interruptions and distractions. These factors reflect the nature of the work, including “competing tasks and interruptions, individual vs. teamwork, physical/cognitive requirements, treatment complexity, workflow.”103 A review of the literature found 18 studies and 2 literature reviews that contained process factors and their association to medication errors by nurses.

Factors such as distractions and interruptions, during the process of delivering care can have a significant impact on medication safety. Nine studies, four with nationwide samples, and two literature reviews present information on the association between MAEs and distractions and interruptions. One survey of nurses in three hospitals in Taiwan found that they perceived distractions and interruptions as causes of errors.93 In three other surveys in the United States, nurses ranked distractions as major causes for the majority of medication errors.58, 61, 102 In a small, five-site observational study of medication administration among 39 RNs, licensed practical nurses (LPNs) and certified medical technicians/assistants (CMT/As), Scott-Cawiezell and colleagues104 found an increase in medication errors attributable in part to interruptions, and when wrong-time errors were excluded, the error rate actually increased during medication administration.

These finding are furthered by research concerning self-reported errors from a nationwide sample of nurses.84 The nurses believed the cause of their reported medication errors and near errors were interruptions and distractions. In a secondary analysis of the MEDMARX® data base, distractions and interruptions were prominent contributing factors to medication errors.81–83 Furthermore, these findings are supported by three reviews of the literature: one found that distractions and interruptions interfered with preparing and administering medication, potentially causing errors;30 interruptions were perceived as causing medication errors in the second review;98 and the third indicated that rapid turnover and changes as well as distractions and interruptions contributed to errors.39

One small study investigated nurse adherence to a hospital policy to document medications administered and their effects on patients. From a sample of 12 nurses in one hospital, one-third of progress notes were found to contain information about administered medications, yet only 30 percent of those progress notes included medication name, dose, and time of administration, and only 10 percent documented information about desired or adverse effects of medications. Medication education, outcomes of administered medication, and assessment prior to administering were not documented in any progress note. Only half of withheld medications were documented.105 In a review of records to detect medication errors, Grasso and colleagues43 found that 62 percent did not document doses as administered.

Five studies and one literature review assessed the relationship between communication failures and medication errors. A small observational study of 12 nurses found that they communicated with other nurses about information resources on medications, how to troubleshoot equipment problems, clarification in medication orders, changes in medication regimens, and patient assessment parameters when handing over patients.106 Nurses communicated with physicians informally to exchange information, about the absence of other physicians, and in both unstructured and structured ward rounds. Nurses also communicated with pharmacists about information on medication administration and organizing medications for patient discharge. Another direct observational study of medication administration found opportunities for errors associated with incomplete or illegible prescriptions.91 This finding was supported by two related literature reviews that indicated that illegible and poorly written drug prescriptions and breakdowns in communication led to errors.30, 39 Another survey found that nurses ranked difficult/illegible physician handwriting as a cause of the majority of medication errors, but did not consider withholding a dose because a lab report was late or omitting a medication while the patient was sleeping as something that should have been communicated to physicians or others.61

A small survey of 39 nurses in three hospitals in Nova Scotia about communication failures during patient transfers found that more than two-thirds of nurses reported difficulty in obtaining an accurate medication history from patients when they were admitted; 82 percent reported patients were unable to provide accurate medication histories. When patients were transferred from across units, 85 percent of nurses reported that medication orders were rewritten at transfer, 92 percent that medication orders were checked against electronic medical records, 62 percent that it was time consuming to clarify medication orders, 66 percent that the reasons for medication changes were made at transfer, and 20 percent that blanket orders are often written as transfer orders.107

Three studies investigated the impact of complexity on medication safety. In a small, five site observational study of medication administration of 39 RNs, LPNs and CMT/As in long-term care settings, Scott-Cawiezell and colleagues104 found that even though RNs administered fewer medications they had more MAEs, compared to LPNs and CMT/As. The suggested explanation was that the mediations RN must administer in long-term care are those with more complexity. Another survey of 284 RNs in 11 hospitals found that pediatric and adult nurses reported numbers of medications being administered as a major reason on why medication errors occur.58 Also, another survey of nurses found that they perceived that complicated doctor-initiated orders (24 percent) and complicated prescription were the major causes of MAEs related to the medication administration process.93

Three studies found that systems and process factors can interfere with medication administration when equipment used in administration does not perform properly, exposing the nurse and patient to safety risks. In two ICU studies, infusion pump problems were involved in 6.7 percent of 58 MAEs in one study24 and 12 percent of the 42 MAEs in the other sutdy.45 Another investigation of smart pumps with integrated decision-support software found that half of the ADEs were considered preventable (2.12 of 100 patient-pump days), and 72 percent of preventable ADEs were serious or life-threatening.108 Given the number of ADEs, the fact that the drug library was bypassed in 24 percent of the infusions, and the frequency of overriding alerts, the investigators concluded that use of the smart pumps did not reduce the rate of serious medication errors—but possibly could if certain process factors could be modified, such as not allowing overrides.

An essential component of the medication process related to the administration of medications is monitoring and assessing the patient by the nurse. Only two studies provided information in this area, offering scant evidence. In the first, based on a small sample of nurses in one unit in one hospital, a qualitative analysis of observed medication administration found that participants monitored patients before, during, and after medication administration.109 Nurses assessed vital signs, lab values, ability to swallow, and patients’ self- report of health. They also felt responsible for timing medication administration and providing as-needed (e.g., PRN) medications. In the second study, where ICU nurses were surveyed, no administration errors were found to be associated with inadequate monitoring or lack of patient information.24

There are a wide range of system-related human factors that can impact medication administration. These factors include characteristics of individual providers (e.g., training, fatigue levels), the nature of the clinical work (e.g., need for attention to detail, time pressures), equipment and technology interfaces (e.g., confusing or straight-forward to operate), the design of the physical environment (e.g., designing rooms to reduce spread of infection and patient falls), and even macro-level factors external to the institution (e.g., evidence base for safe practices, public awareness of patient safety concerns).103 There were 10 studies that assessed the association of human factors with MAEs. Four major themes emerged in the review: fatigue, cognitive abilities, experience, and skills.

Five studies assessed the association between fatigue and sleep loss with MAE errors. The first specifically investigated the effects of fatigue and sleep loss on errors using a national sample of nurses over a 2-week period. In this study, the rate of errors increased after working 12.5 hours.99 A subpopulation of critical care nurses reported forgetfulness, heavy workload, distractions, and high patient acuity as causes for their medication errors or near errors.84 Fatigue and sleep loss was also a factor in a subpopulation of ICU nurses, who reported errors with high-alert medications (e.g., morphine, chemotherapeautic agents).85 The other two studies assessed fatigue along with other variables associated with medication errors. In one of these, a survey of 57 nurses, respondents reported that the majority of medication errors were attributable to fatigue.70 The other study, a survey of 25 nurses in one hospital, found that one of the most frequently perceived causes of medication errors for nurses was being tired and exhausted (33.3 percent).102

The thought processes of nurses during medication administration was assessed in two studies. A semistructured, qualitative interview of 40 hospital nurses prior to implementation of a bar-coding system explored the thinking processes of nurses associated with medication administration.110 Their thought processes involved analyzing situations and seeking validation or a solution when communicating about patients; using knowledge, experience, and understanding of patients’ responses to anticipate problems; integrating their knowledge of lab values and patterns of pathophysiological responses to determine possible need to change dosage or administration timing; checking orders for validity and correctness; assessing patients’ responses for possible side effects and effectiveness of the drug; using cues from patients or family members about need for explanations about drugs; bypassing protocols or procedures, some taking a risk, to get drugs to patients or use time more efficiently; anticipating needs for future problem solving; and applying professional knowledge during drug administration. The other study of nurses, using direct observation in a medical and surgical unit in Australia, found that participants used hypotheticodeductive reasoning to manage patient problems.111 Graduate nurses used pattern recognition of patient characteristics and medications during decisionmaking. Intuition and tacit knowledge was used in relation to changes in patients’ vital signs and to objectively monitor patients.

Thought process can also be distorted by distractions and interruptions. One study employed direct observation of medication administration to determine the effects of human factors on MAEs.24 The investigators found that slips and memory lapses were associated with 46.7 percent of MAEs. During both the prescribing and administration of medications, the causes of errors were attributable to slips and memory lapses (23.1 percent during prescribing vs. 46.7 percent during administration), lack of drug knowledge (46.2 percent during prescribing vs. 13.3 percent during administration), and rule violations (30.8 percent during prescribing vs. 13.3 percent during administration). Another study using direct observation found causes associated with MAEs to include slips and memory lapses (40 percent), rule violations (26 percent), infusion pump problems (12 percent), and lack of drug knowledge (10 percent).45

Experience and skills also impact thought processes. In one study of 40 student nurses and 6 nurses using a computerized program to assess the impact of dyslexia found that the greater the tendency towards dyslexia, the poorer the potential cognitive ability to effectively provide the skills associated with effective drug administration.112 Similarly, in two reviews of the literature, a number of medications errors were found to be caused by poor mathematical skills,30 especially if mathematical skills were needed to properly administer drugs.39

Lack of medication knowledge is a constant problem, and there is a need to continually gain more knowledge about current and new medications.30 Nurses with more education and experience may have greater knowledge of medications.39 However, experience has not been found to mitigate the effect of poor mathematical skills nor frequency of MAEs.30 Those new to a unit or profession may be at risk for errors.39 In a survey of nurses working in three hospitals in Taiwan, nurses reported causes of MAEs as new staff (37.5 percent), unfamiliarity with medication (31.9 percent), unfamiliarity with patient’s condition (22.2 percent), and insufficient training (15.3 percent).93 Inexperience may also contribute to performance (human) deficit, willingness to follow a procedure/protocol, and knowledge deficit. Of these reported contributing factors, 78 percent were due to the inexperience of staff.44 Blegen, Vaughn, and Goode113 found that medication errors rates were inversely related to the proportion of nurses on a unit with greater experience, but were not related to the educational level of the staff on the unit.

Strategies to improve medication safety focused on acute care settings. Twenty-six studies and descriptions of quality improvement projects were identified. Strategies used included recommendations from a nationwide voluntary organization to improve safety, education of nurses and other providers in safe practices, and system change and technology.

Lucian Leape and colleagues116 reported on a 15-month Institute for Healthcare Improvement Breakthrough Series Collaborative intended to reduce ADEs. Eight types of strategies were successfully used, including documentation of allergies, nonpunitive reporting, and standardizing medication administration times. Effective leadership and appropriateness of intervention were associated with successful change implementation. The converse was associated with failure, as were unclear aims, poorly designed interventions, lack of focus on underlying system failures, unclear measures, too much focus on data collection, involvement from only some stakeholders, opposition from physicians and nurses, and conflicting time demands for team members. The findings were limited by the lack of an analysis of the relationship between established safety policies and practices and the success of implementing new strategies, as well as the relationship between the implementation and the occurrence of ADEs.

A survey of 148 hospitals about the characteristics and barriers associated with adoption of the National Quality Forums’ 30 safe practices was done by Rask and colleagues.117 These practices included unit dosing, adopting computerized physician order entry (CPOE), and having a culture of safety. Of the recommended practices, there was high adoption of standardized labeling and storage of medications (90.5 percent), identification of high-alert medications (81 percent), and use of unit doses (81 percent). For-profit hospitals were more likely than not-for-profit hospitals to have unit-dose medication distribution systems (93.1 percent vs. 78.2 percent) and policies on reading back verbal orders (83.1 percent vs. 58.4 percent). There were greater distractions affecting medication administration in large hospitals. Hospitals with 100–299 beds were more likely to report using pharmacists to review and approve nonemergency orders prior to dispensing; and, 69.4 percent of all hospitals used data analysis to drive patient safety quality improvement efforts.

Educational strategies aimed to improve medication safety and avert unnecessary medication errors. One randomized controlled study used an interactive CD-ROM education program to improve the use of safe medication practices and decrease the rate of MAEs.118 Direct observation of medication administration was used to assess the impact. After the training, nurses’ use of safe administration practices increased, but preparation errors did not decrease. There were too few actual medication errors to analyze pre-post differences. Another approach used an 11 module Web-based educational strategy to improve drug safety with a small sample of nurses.119 Direct observation of medication administration was used to determine the outcome. After using these modules, rates of nonintravenous MAEs decreased from 6.1 percent to 4.1 percent. Rates of errors in intravenous drug administration did not decline as expected. Dennison120 reported the results of a medication safety training program for nurses. Knowledge scores improved in this pre-post test study, but there was no significant change in safety climate scores, labeling of intravenous infusion setups, or the number of self-reported errors.

Attempts to improve basic and continuing education in medication safety have been reported, but they have not assessed the impact on actual error rates. In a small pilot study, a problem-based learning approach was found to enable students to use findings from topic-specific research to develop and apply solutions for clinical problems. Papastrat and Wallace121 proposed using problem-based learning and a systems approach to teach students how to prevent medication errors and suggested content, but their approach was not compared to other teaching methods. Another proposed educational strategy for practicing nurses was to use simulation of medication administration and errors in a controlled setting to improve medication safety, “duplicate the complexity of the nurse-patient interaction and related cognitive thought”122 (p. 249). Simulations could be used to prepare nurses to recognize and manage medication errors when and if they occur.

Several attempts to change the system have been tested. Some of the strategies addressed the thoroughness of error reporting, some the processes and events surrounding medication administration, and some focused directly on reducing errors. Using a hospitalwide performance improvement project that emphasized system factors, not individual blame, error reporting increased from a rate of 14.3 percent to 72.5 percent.123 To address intravenous infusion problems, a medication safety education program and medication calculation worksheets were introduced, followed by ongoing Plan-Do-Study-Act cycles.124 Multiple system changes were also used to improve safety of intravenous drug infusion. These included removing 90 to 95 percent of potassium chloride ampoules from the bedside; developing preprinted labels for five common drug infusions; removing four-channel infusion pumps the unit and replacing them with double-channel infusion pumps with a simple interface design; standardizing administration of drugs given by bolus dose using a syringe pump; decreasing missed doses of immunosupression drugs for transplant patients from 25 percent to 9 percent by incorporating them into the main drug chart; implementing standardized prefilter and heparin-lock central venous catheters and heparin infusions into ICU protocol; redesigning drug infusion administration practices throughout the hospital; eliminating burettes for IV drug infusion; preparing standardized drug infusions for 36 drugs; and providing Intranet-based up-to-date drug information.

A time study and focus groups were used to compare nurse efficiency during medication administration using either medication carts with unit doses or a locked wall-mounted cupboard in each patient room.125 After 12 weeks, the wall-mounted units were found to have decreased medication administration time for nurses an average 23 minutes per 12-hour shift. Time saved by not having to search for missing medications saved 0.38 full-time equivalent (FTE) annually. Pharmacists spent an additional 0.05 FTE in stocking room cupboards. Nurses reported more contact time with patients when using room cupboards and fewer interruptions by colleagues during medication preparation and administration. Two small experimental studies attempted to reduce distractions that frequently interrupt nurses during medication administration and thereby introduce the potential for error.126, 127 In both studies a standardized protocol for safe administration of medications was introduced to the nursing staff in the experimental group and signage was used to remind others (physicians, patients, other staff) to not interrupt. The signage in the first study was a vest that the nurse administering medication wore; in the second it was a sign above the preparation area. Direct observation of the number and types of distractions provided the outcome measures in the first study; a questionnaire completed by each nurse administering medications provided the measure of distractions for the second. In both studies, the number of distractions was significantly reduced. Medication error rates were not captured.

One randomized controlled trial compared the use of a dedicated nurse for medication administration to nurses providing comprehensive care, including administering medications, to their patients in two hospitals.128 MAEs were then assessed using direct observation. The investigators found the error rates to be 15.7 percent at the intervention hospital and 14.9 percent in the control hospital. The rate of MAEs was not significantly different between control and experimental groups.

Involving patients in the administration of medications while in the hospital is another system strategy that has been assessed. With this intervention, hospitalized patients have the responsibility for administering their own medication under the supervision of nursing staff. A literature review reported on 12 studies that described and evaluated a patient self-administration program.129 This review found that the patients’ knowledge about their medications and the prescribed dosing increased, but knowledge about the potential side effects of their medications did not. Given the body of the reviewed literature, it appeared as though patients and families make as many or more MAEs than do health care providers.

Another rapid-cycle implementation project over 6 months used continuous quality improvement data before and after implementing a modular, computerized, integrated infusion system.130 Most infusion error warnings occurred between 3 p.m. and 9 p.m., peaking at 6 p.m. Nurses responded to 12 percent of the infusion error warnings by altering the setting and averting errors. The nature of the 88 percent of warnings not responded to was not discussed. Risk scores associated with heparin infusion rates decreased almost fourfold. Almost all nurses used the new software correctly.

Two studies focused on documentation of medication administration. One study introduced a charting system with decision support and used a quasi-experimental design to determine the effects.131 Researchers collected medication charting data for 8 weeks in both the control and study units. Staff in the study unit received an educational intervention about error avoidance through real-time bedside charting, followed by 12 weeks of monitoring and performance feedback. After the 12 weeks, medication charting rates increased from 59 percent to 72 percent in the intervention group. The second study used a computer-based “unreported meds followup” to remind nurse staff about scheduled medications omitted or not documented.132 After charts were prospectively reviewed, a mandatory medication error prevention seminar was given to nurses, and a medication review report was created for nurses. Reported medication errors and documentation of medication administration were reviewed, medication administration policies were developed, and focus changed to the potential causes of errors. Documentation errors decreased over the 3 years of the study, and reported error rates increased by 0.5 percent each year.

Bar-coded medication administration (BCMA) is promoted as the most effective way to reduce administration errors and is being implemented widely. Conceptually this technology should catch nearly all errors, but rigorous evaluation of the impact of technology on error rates has lagged behind implementation. The biggest challenge to determining the effectiveness of BCMA or other interventions is the lack of valid measures of MAEs. Data from voluntary self-reported medication errors are known to capture only a small portion (5 percent to 50 percent) of actual errors, and the BCMA system itself greatly alters nurses’ awareness of errors, thereby systematically affecting reported error rates. Many studies reporting analysis of the impact of BCMA have used data collected by the system only after implementation.133–136 From these we learn the types of errors intercepted by the system. Three other studies of the impact of BCMA on administration errors reported very large reductions: 59–70 percent decrease,137 71 percent and 79 percent drops.138 However, the sources of the data for determining these decreases are not known.

Direct observation of medication administration, a resource- and time-intensive approach to data collection, is the only way to gather unbiased data to evaluate the impact of BCMA on medication administration errors. Three studies have used direct observation; however, each evaluated the implementation of a different set of technology. Franklin and colleagues139 reported a decline in MAE rates from 8.6 percent to 4.4 percent when a new system was implemented in a teaching hospital in England. The system included BCMA, computerized order entry, automated dispensing, and electronic medication administration record. Prescription errors also declined from 3.8 to 2 percent. It is noteworthy that the rate of both administration and prescribing errors by direct observation was much lower than other direct observation studies have reported. Paoletti and colleagues140 used direct observation to determine the impact of BCMA and an electronic medication record in a hospital in the United States. They reported that the rate of MAEs declined from 13.5 percent to 3 percent. Finally, the implementation of only the electronic medication administration record led to a decline in MAEs from 10.5 percent to 6.1 percent using direct observation.141 Health-related technology designed to increase medication safety has great promise, but more study using valid outcome measures and controlled interventions needs to be done to demonstrate the potential benefits.