MEDICATION errors strike at the heart of being a nurse-the responsibility to do good and avoid harm. Medication errors have serious direct and indirect results, and are usually the consequence of breakdowns in a system of care. Direct results include patient harm as well as increased healthcare costs. Indirect results include harm to nurses in terms of professional and personal status, confidence, and practice.
Everyone concerned about patient safety equates medication errors with serious risks to patients. Medication errors also impact organizations and nurses. Ten to 18% of all reported hospital injuries have been attributed to medication errors. 1,2 Five percent of all medication errors reported to the US Food and Drug Administration (FDA) in 2001 were fatal. 3 United States data from 1993 indicate that 7391 patients died from medication errors, and patient stays associated with medication errors increased by 4.6 days, with a resulting cost increase of $4685 per patient. 1 In the state of California alone, over 700 patients die each year because of medication errors. 4
MEDICATION ERRORS
Medication errors are typically defined as deviations from a physician's order. Hospital medication error rates can be as high as 1.9 per patient per day. 5 Sources of errors include illegibly written orders, dispensing errors, calculation errors, monitoring errors, and administration errors (ie, giving the wrong medication to the patient). Physicians, pharmacists, unit clerks, and nurses can be involved in the occurrence of medication errors. 5 A single patient can receive up to 18 doses of medication per day, 6 and a nurse can administer as many as 50 medications per shift. 7 This places the nurse at the front line when it comes to drug administration accountability. 8,9
Medication errors negatively affect nurses. The psychological trauma caused by committing a medication error can be overwhelming to a nurse. First, nurses worry about the patient. Nurses may feel upset, guilty, and terrified about making a medication error. In addition, they can experience a loss of confidence in their clinical practice abilities. Finally, they can feel anger at themselves as well as the system. 10
No studies have demonstrated strong relationships between nurse characteristics (ie, age, years of practice, and education) and number of medication errors. 11,12 This would seem to indicate that any nurse is potentially at risk for making a medication error.
REPORTING MEDICATION ERRORS
Whether the nurse is the source of an error, a contributor, or an observer, organizations rely on nurses as front-line staff to recognize and report medication errors. Several studies have demonstrated underreporting among nurses. 10,12-15 Adding to the burden of reporting, more than 90% of the self-reports are paper-based in California. 16
Prevention of medication errors is linked to accurate reporting of medication errors. Reporting medication errors is dependent on individual nurse's decision making. 15 Underreporting or not reporting medication errors conceals flawed systems. 1
Currently, self-reported medication errors provide minimal information to organizations because discrepancies, in terms of reported-to-actual rates, are widespread. Medication errors are typically reported through institutional reporting systems such as incident reports. 15 Moore, however, estimated that organizations relying on incident reports to provide data miss up to 95% of the medication errors. 17 Reports are generated by the nurse who identifies the error and then are forwarded to management, quality departments, or risk management departments. Reporting systems are dependent on the nurse's (1) ability to recognize an error has occurred, (2) belief that the error warrants reporting, (3) belief that she/he has committed the error, and (4) willingness to overcome the embarrassment and fear of retaliation for having committed a medication administration error. 13
Nurses themselves believe only 25% of all medication errors are reported using incident reports. Only 3.5% of the nurses in the study of Osborne et al believed that all medication errors are ever reported. 12 Nurse managers and physicians also believe that medication errors are underreported by nurses. 13 Errors of both commission and omission go unreported. Failure to administer a medication is the most underreported error because nurses perceive that patients will not be harmed in this situation. Conversely, errors resulting in overmedication are the most frequently reported. 13
Nurses deliberately decide to not report some medication errors. 18 It is estimated that 95% of medication errors are not reported because staff fear punishment. 1 Disciplinary actions including job loss also affect reporting rates. 10,12,13,15Staff nurses also fear being revealed and labeled as someone who has made a medication error. Nurses and nurse managers report that they fear for the reputation of their service or unit. 13,19
Other reasons for not reporting medication errors include disagreement over the definition of an error and the need to report it as well as the degree of reporting effort, ie, time to complete reports. 15 Interestingly, Osborne et al found that 15.8% of the nurses in their study were unsure as to what situation constituted a medication error, and 14% were not sure when to report the error. 12
Most of these studies have limitations that include an inadequate number of sites (eg, one hospital) and units (eg, one medical and surgical unit). It is difficult to know to what degree the local culture influences nurse perceptions about medication errors in these limited site studies. Nonrandom sampling and small sample sizes raise additional methodological issues.
No matter what reporting mechanisms, policies, or procedures are in place, reporting medication errors remains dependent on the nurse's ability to detect medication errors and individual nurse's decision making to report medication errors. The Joint Commission on Accreditation of Healthcare Organizations (JCAHO) states that root cause analysis, while analyzing underlying systems, must include an assessment of the human and other factors. 20 For various reasons, possibly including nurse inability to detect medication errors, perceptions that certain medication errors will not harm patients, or fear of consequences associated with reporting, nurses do not report all detected medication errors. This study was undertaken to understand more about nurse perceptions of medication errors and reporting.
METHOD
This study examined the perceptions of medication errors among a large number of randomly selected nurses in multiple settings. More specifically, it was designed to investigate what nurses believe constitutes a medication error, what is reportable, and what barriers to reporting exist. A self-report survey method was used to collect data for this descriptive, correlational study. The current study replicated a study conducted by Osborne et al by using a larger sample of nurses working in more diverse settings. 12
Population and sample
United Nurses Association of California/Union of Health Care Professionals (UNAC/UHCP) union-represented registered nurses (RNs) practicing in 16 Southern California acute care hospitals represented the study target population of 9000 acute care nurses. Work settings consisted of private, government, military, and health maintenance organization hospitals.
A random sample of 5000 RNs was selected from the above population. According to Roscoe 21 a final sample size of 1000 RNs would be optimal for this type of study. In 2 previous studies that used this study's instrument, response rates ranged from 61.9% 12 to 79%. 10 Historic response rates for surveys mailed to UNAC RNs ranged from 25% to 40%. So, taking a conservative approach, 5000 UNAC RNs were mailed surveys in an attempt to obtain a final sample size of approximately 1000 participants.
Instrument
The Modified Gladstone 12 was chosen to collect data for this study. This instrument measured (1) nurse perceived causes of medication errors (10 items); (2) percentage of drug errors reported to nurse managers (1 item); (3) types of incidents that would be classified as (a) medication errors, (b) reportable to physicians, or (c) reportable using an incident report (6 items); and (4) nurse views about reporting medication errors (6 items). For the purpose of this study, one additional item was added to "types of incidents" that reflected a therapeutic drug level medication scenario. The last portion of the instrument captured nurse demographic data (11 items). Instrument content validity was determined acceptable by previous investigators. 10,12 In addition, Osborne et al established reliability using the test-retest method (0.78) in their sample. 12
RESULTS
Nine hundred eighty-three RNs responded to the survey, representing a 20% return rate. Similar to nurses across the country, the RN mean age was 44.6 years (range = 23-74 years; SD = 9.07). Nurses were primarily female (95%), had been practicing for an average of 18.7 years (range = 1-45 years; SD = 9.94), worked full time (62.7%), and were in benefited positions (88.2%). Similar to the state of California RNs, the ethnic backgrounds of the study participants were varied (49% white, 34% Pacific Islander, 8% Hispanic, 4% African American, and 4% other), as was their highest level of education (11% diploma, 40% associate degree, 44% bachelor's degree, 3% master's degree, and 3% other). Nurses represented all working shifts (42% day, 18% evening, 17% night, 12% 7 AM-7 PM, 8% 7 PM-7 AM, and 4% other). Medical/surgical (M/S), critical care, and maternal child health (MCH) practice settings were represented (Table 1). Overall, the RNs responding to this survey were representative of nurses working in Southern California.
As part of the demographic survey, RNs were also asked the number of medication errors they could remember making over the course of their career. The mean number of errors recalled was 4.9 per nurse (range = 0-100; SD = 5.67). However, most nurses (68.3%) recalled making 2 to 5 errors over their career.
Analysis of the data focused on addressing the study aims that were to describe the following:
1. nurse perceived causes of medication errors,
2. nurse evaluation of medication scenarios,
3. nurse perceptions about reporting medication errors, and
4. relationships between nurse characteristics (demographics) and perceptions regarding medication errors.
Causes of medication errors
Table 2 portrays the ranked causes of medication errors as perceived by the participating RNs. Nurses ranked the listed causes from 1 to 10, with 1 indicating most frequent cause and 10 indicating least frequent cause. Mean scores were calculated for each item and are listed in the table. The top 3 ranked (out of 10) perceived causes of drug errors were the following: (1) MD handwriting is difficult to read or illegible, (2) nurses are distracted, and (3) nurses are tired and exhausted.
Medication scenario evaluation
Based on 6 quite different scenarios presented to the nurses, Table 3 represents how nurses classified each scenario as a medication error (yes or no responses) and if they would or would not report the situation to a physician or complete an incident report. Classifying and reporting medication errors differed between and within scenarios.
Some scenarios elicited common responses in terms of classifying medication scenarios. For example, most nurses (96.6%) responded that they would classify a fast running TPN (total parental nutrition) rate (200 mL/h for 3 hours instead of the correct 125 mL/h) as a drug error; 92.1% would notify the physician; and 93.3% would complete an incident report. On the other hand, most nurses (91.8%) would not classify as a medication error the withholding of a routine morning dose of digoxin because the digoxin blood level report was late. However, in this case 55.4% would notify the physician, but only 11.2% would complete an incident report.
For other scenarios, nurses had quite disparate responses. For example, nurses were split (55.5% versus 44.5%) in their classification of a scenario involving omission of a medication while the patient was sleeping. However, once again more nurses would notify the physician (62.7% versus 37.3%), yet were split (48.3% versus 51.7%) when it came to completing an incident report.
In 5 out of the 6 scenarios, more nurses would notify physicians than not notify them no matter how they first classified the scenarios. In 5 out of the 6 scenarios, more nurses would not complete an incident report; this decision mirrored their original classification of the scenario as either being a medication error or not being a medication error.
Reporting medication errors
In addition to evaluating scenarios, nurses were also asked, "In your estimation, what percent of all drug errors are reported to the nurse manager by the completion of an incident report?" The mean percentage was 45.6, indicating that less than half of the nurses believed that all drug errors are reported to a nurse manager using an incident report.
Table 4 presents additional nurse responses to statements about reporting medication errors. Most nurses indicated that they knew what constituted a medication error (92.6%) and when to report an error using an incident report (91.3%). Reasons for not reporting errors included "afraid of manager reaction" (76.9%), "afraid of coworkers' reactions" (61.4%), and "not thinking an error was serious enough" (52.9%). However, the majority of nurses (80.4%) do not seem to fear disciplinary action (losing one's job) because of committing an error.
Links between nurse responses and nurse characteristics
Very few nurse characteristics were associated with the survey responses including the number of errors nurses remembered making over their career. Overall, nurses working in an MCH versus an M/S setting reported they perceived a greater percentage of medication errors are reported. For example, nurses working in neonatal intensive care units indicated they perceived a greater percentage of medication errors (52.5%) are reported to nurse managers than did nurses working in M/S units (eg, 35.3% oncology).
Even though nurses working in a MCH unit versus M/S unit responded as above, the relationship between type of unit and percentage of perceived reported errors was weak (r = 0.21; P = .01). There also was a weak relationship between percentage of errors perceived reported and years of RN practice (r = 0.15; P < .001). This means that these 2 RN characteristics, type of unit and years of practice, explain very little about how nurses responded to this question. Overall, while there were a few significant relationships found between the nurse characteristics and items on the survey, these relationships were weak.
Limitations
The sample for this study was drawn from a healthcare union, and therefore, the findings may not be generalizable to nonrepresented RNs. Regarding the ranking of medication errors, there may be additional causes that were not identified on this study's survey. However, the top ranked causes from this study could be a starting point for organizations to address system issues. In addition, the scenarios were brief and did not provide situational details. We have no way to know if nurses read more into the scenarios than what was provided; however, lengthening surveys can decrease overall return rates. The decision was made to retain the brief scenarios in order to optimize the sample size.
NURSING IMPLICATIONS
Similar to other studies, no single or combination of nurse demographic characteristics were strongly associated with nurse perceptions of medication errors or the reporting of medication errors. Thus, all nurses in an organization may need help in identifying what is a medication error, when to report it, and to whom. What this means for nurses working on quality and patient safety programs is that, most likely, there are no specific groups of nurses to target for interventions or education. For example, inexperienced nurses are not reporting more or less errors than do experienced nurses. This once again emphasizes systemic problems at issue in regard to medication errors.
This study calls attention to the need to clarify with nursing staff what constitutes a medication error. Interestingly, nurses were "usually sure what constitutes an error" (92.6% yes, 7.4% no) yet were not in high agreement with one another when given actual medication scenarios (ie, 52.7% yes, 55.5% no). This study has identified a gap between the nurse's perceived knowledge and his or her actual knowledge. It is clear that nurses need specific information about what constitutes medication errors. The information gained from this study can be used in educational programs designed to promote the recognition of these errors.
Now that we know nurses differ in their perceptions as to what constitutes a medication error, do organizations have clear guidelines available as to what situations represent medication errors? Regardless of our personal opinions, traditions related to nursing's 5 rights of medication administration, or our unstated expectations, this study demonstrates that nurses are not "on the same page" as to what is a medication error and when to report to it.
Currently, organizations are having their patient safety programs examined in great detail by accrediting and licensing bodies. Paramount to any patient safety program is the medication error-reporting component. The purpose of having a comprehensive, accurate, and timely reporting program in place is to be able to identify and correct knowledge and system defects immediately.
Similar to studies by Gladstone 10 and Osborne et al, 12 this study identified differences in reporting medication errors as well as perceived barriers to reporting. However, strong barriers to reporting did not include fear of disciplinary action but were more in line with interpersonal reactions from managers and staff. Discussions among staff and nurse managers about the taboo subject of medication errors are desperately needed. Also, integrating human factors theory into managerial level educational programs may help managers focus on system redesigns to reduce or eliminate reporting barriers within their organizations.
Finally, systems-oriented and critical thinking should be promoted to enhance nursing judgment to decrease the notion that certain errors do not warrant reporting. All errors including near misses should be reported so that organizations have an opportunity to improve their patient safety programs.
Any practicing nurse knows that the causes of medication errors are both varied and complex. Because medication errors are such a concern to the public, healthcare organizations, and nurses themselves, this study was undertaken to ask nurses about what they believe constitutes a medication error, what is reportable, and what barriers to reporting exist. Additional dialogue and research with nurses are needed. Questions to raise with nurses include the following:
1. How do nurses define medication errors?
2. Is there a unique and different definition for reportable medication errors versus non-reportable errors?
3. Why is there a difference in nursing judgment between reporting medication errors to physicians and reporting medication errors using incident reports?
4. What can organizations do to promote the reporting of medication errors and near misses?
While this study has generated some important questions, it also has provided some insights into medication errors and reporting. The knowledge gained from this study can contribute to educational programs that promote the recognition of medication errors. The knowledge also can assist with system redesigns to reduce or eliminate barriers to reporting medication errors. Patient safety programs can be strengthened through timely, accurate, and comprehensive reporting, ultimately ensuring the highest quality patient care.
REFERENCES