Article Text

Download PDFPDF

Medication errors in the neonatal intensive care unit: special patients, unique issues
  1. J E Gray2,
  2. D A Goldmann1
  1. 1Department of Medicine and Quality Improvement Program, Children’s Hospital, Boston, MA 02115, USA
  2. 2Department of Neonatology, Beth Israel Deaconess Medical Center, Boston, MA 02115, USA
  1. Correspondence to:
    Dr Gray
    Department of Neonatology, Beth Israel Deaconess Medical Center, Boston, MA 02115, USA;

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Medication errors are quite common in the neonatal intensive care unit

Medical errors are a common occurrence in the neonatal intensive care unit (NICU). Although this high risk, fragile patient population is prone to a wide array of errors, medication errors are particularly common. Medication errors were the most common error type submitted to the Vermont Oxford Network’s voluntary reporting system.1 Kaushal and colleagues2 identified errors in 5.5% of NICU medication orders. Of note, potential adverse drug events (errors that had the potential to harm the patient but were intercepted, or potentially harmful errors that reached the patient but fortuitously did not result in injury) occurred eight times more often in NICU patients than in adults in hospital. Neonates, especially very low birthweight babies, are particularly vulnerable to adverse sequelae of medication errors as they have a limited ability to “buffer” such mistakes.

Nursing practice has long recognised the need for extreme vigilance and a structured approach to preventing medication errors. The five “Rights” provide a framework for improving medication safety in nursing. These basic principles of standard operating procedure try to address all of the steps in the medication process: ordering, dispensing, administering, and monitoring drugs. Nurses attempt to ensure that the Right drug is given in the Right dose at the Right interval via the Right route to the Right patient.

Although nurses focus on providing error-free care, research into human factors teaches us that dedication, training, and vigilance are not enough to prevent errors in complex systems.3,4 Error prevention must be a multidisciplinary process, involving doctors, pharmacists, and nurses working as a team. The team must be backed up by robust healthcare delivery systems operating in a “culture of safety”, providing staff with a working environment that provides safeguards against human fallibility. Nowhere in the hospital is the challenge greater than in the NICU.

The repertoire of drugs prescribed in the NICU is relatively limited compared with adult and older paediatric populations, but the process of ordering, dispensing, and administering them is more complex in newborns. The process for ordering drugs in the NICU is uniquely complex; more than three quarters of medication errors occurred during this stage.2 As doses are calculated according to the infant’s weight, virtually all prescriptions require patient specific calculations and may need to be updated as the infant gains or loses weight. Weight and gestational age are not the only factors that need to be considered. For premature infants, doses must also be modified on the basis of the developmental maturity of specific metabolic and excretory pathways.

Drugs prescribed in the NICU are often used in an off label or unlicensed fashion.5 As a result, no comprehensive and authoritative standards for doses exist. Therefore clinicians are often confronted by a dizzying array of published reference standards for a single drug. Recommendations are surprisingly variable even for drugs that have been studied in neonates and approved for use by the Food and Drug Administration. For example, widely used references in the United States suggest total daily ampicillin doses that vary by a factor of 3–4 for the same 1 kg patient.6–8 Certainly, for a drug with a wide therapeutic index, this difference may not be clinically significant. However, the lack of a single dosing standard within a hospital can complicate the development of error reduction strategies in which doctors, nurses, and pharmacists verify doses.

NICU drug dispensing is also complex. Pharmacists often have to dilute stock solutions in order to provide doses that are extremely minute compared with adult standards.9 In this issue of the journal, Chappell and Newman10 document the potential for 10–100-fold dosing errors associated with the use of stock drug concentrations intended primarily for use in adults. Of particular concern is the fact that three of 10 drugs at risk of 10-fold dosing errors and all four at risk of 100-fold errors are high alert drugs as defined by the Institute for Safe Medication Practice.11 Even more alarming is the fact that these decimal point errors represent only a portion of the calculation errors that can complicate the ordering and preparation process.

Errors in the route of administration of drugs and enteral nutrition are also common, complicated 13.3% of potentially harmful medication errors seen in two NICUs in the United States.12 In another report, Suresh and colleagues1 noted potentially very serious administration errors, such as infants fed expressed breast milk intravenously. Unlike adult care units, enteral feeding tubes and intravenous lines are often of the same calibre and appearance and have hubs of similar size. This type of error could be prevented by adopting administration systems with “forcing functions” that prevent feeding pumps and syringes from being attached to intravenous lines. Regrettably, these systems are not in widespread use in NICUs, in part because of incompatibilities with existing equipment and work flow processes.

Finally, patient misidentification occurs commonly in the NICU. One quarter of the serious medication errors reported in this issue by Simpson et al13 involved patient misidentification. Similarly, Suresh et al1 found that 11% of NICU errors involved misidentification. The increasing incidence of multiple gestations with premature births is in part responsible for these errors, but suboptimal systems for identifying babies contribute to the problem. Analyses by the Center for Patient Safety in NICU care suggest that as many as one half of infants in the NICU are at risk of misidentification on any given day (unpublished work).

The patient safety movement has highlighted numerous approaches to preventing medication errors, but which interventions have the potential to have the greatest impact? Fortescue and colleagues12 have identified three interventions with the largest potential to decrease NICU medication errors: ward based clinical pharmacists, computerised physician order entry (CPOE), and improved communication among NICU clinicians.

The involvement of clinical pharmacists in intensive care unit rounds significantly reduces dosing and other types of error in adult care.14 In this issue, Simpson et al13 conclude that similar improvements can be achieved through the input of an NICU based clinical pharmacist. Although their data are encouraging, confidence in their conclusion must be tempered by several methodological concerns. Multiple interventions were applied during the study, and the exact timing and interaction of these interventions are unclear. Some discussion of the background and expertise of the pharmacists participating in the intervention would have been valuable as neonatal expertise and experience are almost certainly important. Unfortunately, the authors expressed the major outcome measure as the absolute number of medication errors, rather than error rates per number of patient days or per number of orders written. We hope that these important denominators remained relatively stable during the study period. In addition, it is unclear to what extent the ascertainment methods used, which relied on voluntary reporting by clinicians, were accurate and unbiased. Voluntary reporting, although valuable on many levels, cannot be relied on to provide accurate incidence data. Finally, the authors provide no statistical measures of differences between the periods before and after intervention.

Implementation of CPOE in the NICU presents special challenges. Systems designed for use in older patients may not adequately address the unique aspects of NICU medication ordering. Unfortunately, development of systems appropriate for use in paediatric and neonatal patients has lagged. Industry must be challenged to provide software applications that are appropriate for NICUs. CPOE almost certainly will have to be integrated with other hospital clinical information systems to have maximum impact on error prevention. Adequate, built in decision support, using population specific knowledge bases, is essential for detecting drug interactions, out of range doses, and other prescribing problems. The LeapFrog Group,15 a consortium of Fortune 500 companies, has urged hospitals in the United States to adopt CPOE. Given Leapfrog’s leverage and influence, recognition of the unique needs of NICUs would be welcome.

Where CPOE is not available, attention to good prescribing practices and accurate communication are essential.5,16 This is true not only for written orders, but verbal ones as well. The process for verbal orders should include a system of “read back” verification to ensure accuracy. Lacking CPOE, clinicians (doctors, nurses, and pharmacists) must implement unambiguous guidelines on appropriate dosing for NICU patients. Good communication and teamwork requires a blame free environment and a culture that places a high value on reporting and discussing patient safety concerns and systems problems.

Finally, NICU clinicians must remain aware of the advances in patient safety made in other industries. Crew Resource Management, which has been pivotal to improving the safety record of the aviation industry, may be particularly useful in helping teams communicate effectively and safely.17 Translation of technologies from the retail sector, such as bar coding and radio frequency identification, may be helpful in preventing patient misidentification. When feasible, engineering approaches using affordances and reminders, forcing functions, and constraints may help staff to avoid errors due to human factors. Of course, these novel approaches to creating a safe care environment will have to be tailored to the very special and challenging environment of the NICU.

Medication errors are quite common in the neonatal intensive care unit


View Abstract


  • Supported in part by a grant from the Agency for Healthcare Research and Quality (funded by AHRQ P20 HS 11583).

Linked Articles

  • Fantoms
    Ben Stenson