This study is both the largest analysis of NICU medication errors to date, and to our knowledge, the first to show that use of ISMP High-Alert Medications is a risk factor for harm in the NICU. Additional risk factors for harm also emerge in this study and include errors that begin in the prescribing phase of medication use and failures in equipment and delivery devices. Collectively, these factors have begun to lay the groundwork for a risk profile for harmful medication errors in the NICU.
In addition, this study reveals that although the majority of reported medication errors do not result in harm to the NICU patient, medication errors are common in this clinical setting. In this analysis, the bulk of medication errors occur in the administering phase of medication processing, and human factors are the most common cause of error. Furthermore, electrolytes, caloric, and water balance agents and anti-infective agents constitute more than half of the medication categories identified in the error reports, and gentamicin and ampicillin are the most commonly identified medications. TPN, fat emulsions, and fentanyl, which are also ISMP High-Alert Medications, are on the list of the top five most commonly named medications.
The findings of this study are consistent with those conveyed by Suresh et al. Both studies highlight the large number of administering errors that occur in the NICU and how human factors surface as the most frequent cause and contributory factor to medication errors. Harm scores, however, differ between the two studies. Suresh et al. found that approximately 27% of reported errors resulted in harm (categorized as minor harm, serious harm, or fatality). Our analysis, however, finds that 4% of reported medication errors result in harm. This discrepancy likely reflects the use of different harm definitions and categories between the studies, as well as possible differences in the propensity to report 'near misses.' In addition, our study examines reports that are exclusive to the NICU and focuses purely on medication error reports. It does not include other medical errors or involve other locations that provide care for neonates such as nurseries, step-down units, delivery rooms, or emergency rooms as was the case in the Vermont Oxford Network study. Furthermore, our report uses a modified NCC MERP classification system as opposed to having errors categorized solely by experts in the field of safety.
Our findings are also consistent with earlier studies that have highlighted anti-infective, analgesic/sedative, and electrolytic/fluid agents as the most common drugs involved in pediatric, and now specifically, NICU medication errors.[8,15,18,27] Our results likely reflect the frequent use of these medications in the NICU. One definition of risk indicates that risk is directly proportional to frequency of occurrence and to harm. Accordingly, attention should be paid to those medications that are used regularly and that have narrow therapeutic windows. Interestingly, four of the top five medications identified in this report have narrow therapeutic windows relative to other drugs. For instance, TPN and fat emulsions have been shown to predispose NICU patients to harm from electrolyte imbalances and hypertriglyceridemia.[28–30] Given the frequent use and harm potential, TPN and fat emulsions have been targets for intervention in the NICU. On the contrary, efforts to prevent harm from aminoglycosides such as gentamicin and opiates such as fentanyl have not received as much attention in the neonatal population. Opiates run a 6 to 8% risk for respiratory depression in the neonate, and aminoglycosides have serious side effects such as nephrotoxicity and ototoxicity. Overdoses of these medications or improper monitoring protocols have the potential for devastating outcomes including death. Despite improvements in the delivery of continuous intravenous infusion medications, more work is needed to establish guidelines for their use.
Surprisingly, age is not a significant risk factor for harm in this study. These results suggest that harmful errors may be a result of system failures as opposed to the underlying medical illness or changing physiology of the NICU patient as depicted by chronological age alone. Other clinical data required to interpret patient age in the NICU such as gestation and birth weight is needed to help clarify this issue. Last but not least, our data suggest that there may be a protective effect of transcribing and documenting errors. One explanation for this finding is that these errors are not as likely to reach the patient because of various checks-and-balances in the medication processing system.
Interestingly, stress and workload is the least cited error cause in this study suggesting that underreporting of these types of errors may be present in MEDMARX. In addition, very few error reports cited that errors were reported to caregivers and patients, thus underscoring that a culture of non-disclosure of errors may be present in the health-care system. Alternatively, as majority of reporters are often nursing staff, the nursing staff may have been unaware of disclosure conversations between the physician staff and the patients and families. However, more information on facility and staffing characteristics is needed to interpret these findings.
MEDMARX has limitations.[35–40] First, errors are voluntarily reported and hence not representative of all medication errors. More complete reporting of all categories of errors, such as Category A and B medication errors, could shift the distribution of the reported errors. In addition, because total numbers of doses dispensed are not available, it is not possible to calculate error rates. Nevertheless, occurrence data remain a good way to identify trends. Second, the majority of reports were made by nurses. This characteristic potentially skews the data toward administering errors. Even so, the nursing perspective provides information regarding medication errors undetectable by other methodologies. Third, patient descriptor variables and facility characteristics are limited in MEDMARX. For example, had more patient characteristics such as gestational age and birth weight been collected, more risk factors for harmful medication errors could have been investigated and insight into the lack of significance of patient age for harm been gained. Moreover, information such as nursery level, number of annual deliveries, and number of NICU beds would have been helpful in characterizing the settings in which the various errors occurred. Indeed, the majority of NICUs reporting errors in this study were part of community and general hospitals, reducing the generalizability to other health-care settings such as academic NICUs. We recommend that these variables be added to reporting systems. And finally, although a good inter-rater agreement in MEDMARX has been established, the definitions of the classification system are still subject to interpretation by the reporter, which can lead to misclassification of error data.
According to Leape, successful reporting systems are responsive and provide expert analysis and timely feedback. Toward this aim, our study has identified several targets for intervention in preventing harmful medication errors in the NICU. First, our study calls for more work on the delivery and monitoring of ISMP High-Alert Medications as well as on the most frequently cited medications in NICU error reports such as aminoglycosides and opiates. Second, our study indicates that the current push for installing computerized physician order entry systems nationwide[42–45] would be beneficial in preventing harmful medication errors in the NICU. However, because over half of the errors occurred at the administering stage of medication processing, other interventions such as the smart pump would be useful to reduce these errors. However, caution must be exercised when introducing these devices in the NICU. As shown in this study, faulty equipment and delivery devices for medication administration were associated with harm. Strategies to help reduce this potential for harm need to be explored and can include techniques such as direct observational analyses and simulations.[47,48] By targeting these risk factors for harm in the NICU, our data suggest that a significant number of errors that occur in this clinical setting could be reduced.
This work was conducted while the first author was a Neonatology Fellow at Johns Hopkins University.
Conflict of interest
The following authors have no conflict of interest with publication of this paper to report: Theodora A Stavroudis, MD and David Bundy, MD. Andrew D Shore, PhD receives 20% salary support from a United States Pharmacopeia (USP) contract and also has a separate consulting agreement to perform additional analyses as requested by USP. Marlene R Miller, MD, MSc and Laura Morlock, PhD have a research contract with USP to analyze data and have full authority to publish findings of their work without requiring pre-approval from USP. Rodney W Hicks, PhD, ARNP was formerly the Manager of Patient Safety Research at USP and is currently the UMC Health System Endowed Chair for Patient Safety at the Texas Tech University Health Sciences Center School of Nursing.
We thank Maureen Fahey, MLA for her help with data coding; Panagiotis K Ordoulidis for his assistance in the preparation of this manuscript. This work was supported by the Agency for Healthcare Research and Quality Grant 1R03HS016774-01A1.
J Perinatol. 2010;30(7):459-468. © 2010 Nature Publishing Group
Cite this: NICU Medication Errors: Identifying a Risk Profile for Medication Errors in the Neonatal Intensive Care Unit - Medscape - Jul 01, 2010.