Intralipid infusions remain a critical part of ensuring adequate nutritional supplement and growth in premature and term infants. Managing intralipid therapy requires great care to prevent metabolic and physiological side effects. The authors sought to systematically study medication errors associated with intralipid administration in the neonatal intensive care unit (NICU). A descriptive quantitative and qualitative analysis incorporating secondary data was used. Medication error data were drawn from 54 institutions that voluntarily participated with MEDMARX, a national, Internet-accessible medication error reporting program owned and operated by the United States Pharmacopeia. These errors were associated with NICUs, and each medication error record identified nursing staff as making the initial error. A total of 257 errors were reviewed, with 3.9% resulting in harm. The mean age of the neonate was 7 days, and more errors occurred on Mondays than any other day of the week. Errors disproportionately occurred between 6 pm and midnight, with a significant difference between errors near 7 am and 7 pm (P = .002). Wrong dose errors occurred in 69% of the sample. Nearly one quarter of the errors resulted from misprogramming infusion devices (either pumps or syringes). Qualitative findings revealed that many of the errors were the result of the nurse's misinterpretation of the modes (ie, time, volume, or rate) on the infusion device or by not recognizing the decimal point on the device's display panel. Several errors involved switching the rate of infusion with total parenteral nutrition and that of intralipids. Voluntary medication error reporting offers valuable insights into intralipid errors occurring in NICUs. Secondary analysis is an ethical, economic means of studying the occurrence of such errors. MEDMARX data suggest that some of the serious errors are the result of complex care and equipment needed for these vulnerable infants.

Researchers from the University of Medicine and Dentistry of New Jersey (UMDNJ), Robert Wood Johnson Medical School (RWJMS) and medication safety experts from the United States Pharmacopeia (USP) identified key variables related to intralipid medication errors that occurred in NICU settings. This study obtained full exemption from subject review from the Institutional Review Board at UMDNJ.

Because MEDMARX is a dynamic database that accepts new reports daily, USP routinely creates "static" views for research to ensure that data do not change. A primary key (known as the MEDMARX record number or the facility ID number) links the database tables for appropriate data extraction as outlined in the sampling criteria. Data extraction via Crystal Reports yielded individual portable document format (PDF) reports for review by the researchers for qualitative analysis. Error descriptions reflecting similar content were grouped by similar themes. Crystal Report worksheets were quantitatively analyzed using statistical software (Statistica, Version 7, Tulsa, OK).

From the MEDMARX staff field, the researchers identified the level of staff involved in the error. Nursing service was defined by the titles: registered nurse, licensed practical/vocational nurse, nonspecific nursing personnel, nursing assistant, unit secretary/clerk, or unlicensed assistive personnel.

Quantitative Findings

Facilities Participating and Sample

Between January 1, 2000, and December 31, 2005, MEDMARX received 266 reports of NICU medication errors involving fat emulsion. Of these reports, 257 (96.6%) identified a member of nursing service as the person responsible for initiating the error. The 257 reports came from general community hospitals (n = 40) or teaching hospitals (n = 13). Data from 1 children's hospital were present in the sample. The mean age of the infants involved in the errors was 7 days (range, 1-189 days).

Error Severity

FIGURE 1. Day of Week and Medication Errors.

Time of Error

FIGURE 4. Shift Change and Intralipid Medication Errors.

Types of Error

TABLE 3. Types of Intralipid Errors

Cause of Error

The cause of error is another multiselect field containing 63 different selections. However, in this study, only 26 selections were associated with the intralipid errors. The cause of error was identified in 253 records (of 257), and several records made more than 1 selection given that there was a total of 402 selections. The

TABLE 4. Causes of Intralipid Medication Errors

Contributing Factors

TABLE 5. Factors Contributing to Intralipid Medication Errors

Qualitative Findings

Types and Causes of Error

Nearly one quarter of all errors was associated with the improper use of an infusion device and often resulted in an infant receiving the wrong amount (Improper dose/quantity) of intralipid infusion. In some cases, the patient received too little intralipid infusion because of erroneous pump settings or tubing not connected to the patient. In most of these cases, however, the neonate received too much of the drug because of an incorrect infusion rate. Several cases described that errors were not detected at change of shift, but rather at the time that the subsequent infusion was initiated or when the present infusion completed earlier than planned.

Several cases in this study noted problems in setting the correct duration of the infusion. Multiple cases describe infusions that were intended to be administered over several hours, but in actuality, the neonate received the entire infusion in several minutes. In 1 case, the order was to deliver the volume during a period of 20 hours; however, the infusion completed in 20 minutes. In a similar case, the intent described was to deliver the infusion in 12 hours, but the infusion was complete in 12 minutes. In a third case, the intralipids should have been infused in 24 hours, but the entire dose was delivered in 30 minutes. In yet another case, the reporter described an intended infusion that was to have lasted 24 hours. The actual infusion was complete in 1 hour. When investigating this particular error, it was determined that the pump used did not have sufficient settings to allow for an infusion of less than 1 mL per hour. One of the more serious cases, affecting a 4-day-old infant, involved an infusion with an intended duration of 24 hours that completely infused in less than 1 hour, which resulted in severe apnea in the infant.

Decimal point settings on infusion devices also led to errors. In 1 case involving an 8-day-old neonate, the infusion was ordered at 0.5 mL per hour, but the infusion device was programmed to deliver 5.0 mL per hour. As a result of this error, additional laboratory tests were necessitated to monitor the infant for more serious effects. In a similar case, an infusion was ordered at 0.2 mL per hour in a 2-day-old neonate, but the pump was set to deliver 2.0 mL per hour. Many of these 10-fold dose variances were the result of misinterpreting the decimal value on the pump's display panel.

Multiple parenteral infusions were implicated in some of the errors analyzed for this study. In 1 case, the neonate had hyperalimentation infusing at 3.5 mL per hour and intralipids at 0.5 mL per hour. The rates for the 2 infusions were transposed, resulting in the intralipids being infused in a period of 7 hours, rather than 20 hours as ordered. In a similar case involving a 14-day-old neonate, the hyperalimentation was intended to deliver at 15 mL per hour and the intralipids at 2 mL per hour, but the 2 were inadvertently switched between the IV pump rates. Another case involving a 30-day-old infant indicated that total parenteral nutrition was intended at 14.1 mL per hour with intralipids at 2.5 mL per hour, but these were switched.

Several cases in this study implicated infusion syringes resulting in medication errors. For example, some cases describe that a 20-mL volume was intended, but a 30-mL syringe was used. The rate was based on the 30-mL syringe. In one case, a 20-mL syringe was used and inserted incorrectly into the pump. The pump erroneously detected 30 mL and automatically adjusted the infusion rate.

Additional records analyzed described pump programming as responsible for the medication error. When programming the device, nurses enter the amount to be infused, the rate, the duration, and the total volume. In at least 2 cases, the pumps were programmed with volume rather than time (eg, 24 mL became 24 minutes).

Omission errors fell into broad categories. In some cases, the infusions were never initiated. The report's description indicated that either infusion devices were not turned on or orders for therapy were overlooked. In other error descriptions involving omission cases, tubing either became disconnected or was never connected to the patient.

Comment

The data in this study came from 54 institutions that voluntarily reported 257 NICU intralipid medication errors. All errors were associated with nursing service and reported to a large Internet-accessible medication error reporting program. While such data are sometimes criticized as not being truly representative of the extent of actual medication errors, it does represent one of the most efficient means to gather data that are reflective of the experiences of many different institutions. As with other external voluntary reporting programs, such as the Vermont Oxford Network, data clearly identify patterns by aggregating larger data samples than would normally be available in a single institution. Improving medication safety requires a broad understanding of the nature and causes of medication errors. Participation in event reporting programs addresses the need to learn from the mistakes of a few to prevent future errors by many.

This study points to 2 important conclusions. First, this study provides evidence that both nonharmful and harmful medication errors involving intralipid infusions may be more common than previously known. Between January 1, 2000, and December 31, 2005, there were 173 hospitals reporting NICU medication errors to MEDMARX. From this set, 54 hospitals (or about 30%) reported intralipid errors. More than two-thirds of the errors reported did not necessitate further workup and were not harmful, but nearly one-quarter required additional intervention to preclude harm (Category D). The typical initial responses included stopping the fat emulsion infusion, obtaining a triglyceride level, and continued cardiopulmonary monitoring. Subsequent actions depended on the triglyceride level (ie, >200 mg/dL) and the presence of clinical decline. Nearly 4% of the errors resulted in harm, with one case showing permanent harm that required substantial ventilatory support and nitrous oxide.

Second, this study provides evidence that NICUs must appreciate the inherent risk surrounding nursing staff's utilization of infusion pumps, especially when lipids are being infused. Infusion pumps deliver intravenous medications and fulfill a vital role in the medication administration. Key elements in initiating a medication in NICU through an infusion device include:

* verification of the therapy order,

* verification of the product for infusion,

* confirmation of patient identification,

* programming of the device to ensure appropriate volume, rate of infusion, and

* preparation of the syringe, if necessary.

Nutritional therapy orders are quite complex and highly individualized. The reason for the number of errors occurring on Mondays and at later hours in the day is perplexing. Perhaps staff members who work weekends are off on Mondays, resulting in practitioners who may not be familiar with an infant's plan of care. Perhaps institutions utilize the same set of intralipid orders for Friday through Sunday, and on Monday morning, new orders or changes in orders originate after the attending staff members complete rounds, and change the orders to reflect the weekend's weight gain. Nursing support personnel transcribe the written orders and update the nursing care plan. After receiving the orders, the pharmacy department begins a review and then dispenses the therapy for delivery to the NICU for administration, which could be a reason for errors occurring later in the day. The medication use process must allow sufficient time for third-party suppliers to provide sterile compounded products if the hospital has outsourced this operation. Further research will determine if these findings are meaningful trends.

The impact of contributing factors warrants further investigation. Perhaps staff members are accustomed to current workflow processes and not able to identify those organizational, situational, or environmental factors that contribute to errors. Consideration of such factors is essential when examining and planning for systems-levels interventions to reduce future errors. Not recognizing and reporting contributing factors leaves a sizeable gap in understanding medication errors.

Limitations and Assumptions

Secondary data analysis has inherent limitations. Two leading criticisms of voluntary medication error reporting are the lack of representativeness of what may actually happen and the issue of under-reporting. Hospitals and health systems that voluntarily report medication errors, by definition, have self-selected their participation, negating the underlying tenets of random selection. The analyses presented in this report came from MEDMARX-participating hospitals and health systems. Because of this bias, the data are not risk-adjusted for opportunities for error, meaning there is no denominator data available.

An additional criticism is that such a database of errors is not representative of all hospitals. Despite the limitations associated with using existing data sets and self-reported data, the MEDMARX data provide the most cost-effective and time-efficient resource for examining the nature of intralipid-related medication errors from the NICU setting and the impact of such errors on patient safety.

Clinical Implications

This study documents the multifactorial causes of medication errors at the point of drug administration. The findings suggest that multiple strategies are necessary to reduce future occurrences of such errors. Even this small sample of NICU medication errors should stimulate systems changes aimed at eradicating errors with intralipids. As one means of further exploring such medication errors, NICUs should begin to examine the workflow plan associated with intralipid therapy to identify potential weaknesses.

Some of the errors analyzed occurred after the change-of-shift report. Perhaps change-of-shift reporting in NICU settings should include additional systems and safety checks, especially pump inspection and validation of infusion rates. At shift change, verification of orders and infusion device settings (inclusive of performing a complete trace of the infusion pump tubing) should occur. Practitioners must become overly scrupulous in including such safety checks in practice, and clinical documentation must reflect that these activities occurred.

To the extent possible, the NICU must begin to standardize infusion pumps. At a minimum, NICUs should attempt to obtain pumps from a single manufacturer because the use of multiple brands of devices requires additional, and often, different, equipment and supplies. Given that it is routine for infants to have more than one infusion device, clinicians often must position the various pumps in a manner that accommodates the care needs of the infant. As manufacturers incorporate more electronic display panels into devices, the opportunity to overlook the decimal point setting on the pump may increase, especially in the presence of the low light setting of the NICU or with multiple stacked devices. Display panels must be within reasonable line of sight of the practitioner caring for the infant.

Each NICU should have a multidisciplinary team that examines the full medication use process. The team should develop an appropriate FMEA plan and then develop interventions to address failure points. The team should be prepared to investigate serious medication errors through the root cause analysis (RCA) process. Findings from both should lead to constructive feedback and quality improvement activities.

Institutions must examine the physical characteristics of infusion devices. With infusion pump programming accomplished by repetitively touching keypads, the surface plate may become worn, distorting the keypad language and further increasing the opportunity for error. Organizations should remove from the practice setting infusion devices that have evidence of abnormal wear and replace them with newer equipment.

Future research should affirm that changes in practice and technology do not introduce new error. Ongoing voluntary reporting of medication errors, along with corresponding analysis and dissemination, must continue to ensure a safer environment for our neonatal population.

References