Drug Information Center

FAQ

Medication errors in US children’s hospitals

Traditional methods to detect and quantify medication errors in a hospital setting include voluntary reporting, chart review, and direct observation. Problems exist with these methods including an unwillingness to report errors and the potential to miss errors during a chart review. It is widely believed that medication error rates are higher than those seen with traditional methods. Based on this potential discrepancy, Takata and colleagues developed a tool to identify medication errors in pediatric patients and tested the tool in 12 children’s hospitals in the US.

This tool uses an approach known as the “trigger method”, which uses specific items to help identify medication errors. The trigger approach involves a focused review of the medical record for information associated with medication errors and has been shown to be useful in previous studies for identifying medication errors in adults and neonates in the neonatal intensive care unit (NICU). The tool was developed in a phase I portion of the study from a chart review of 931 patients from the 12 children’s hospitals. The phase I chart review used an adult trigger tool and focused on comparing the tool to traditional methods and getting input to adapt the tool for pediatric patients. Examples of the triggers (in the final tool) used by Takata and colleagues for phase II of the study and potential errors associated with them are summarized in the table.

Table. Examples of triggers and potential errors identified by the trigger.

Phase II of the study was a retrospective chart review at 12 children’s hospitals of randomly selected charts of children who were hospitalized for at least 2 days. Chart review was limited to the first 30 days of an admission; newborns and patients not meeting criteria during the specified time frame were excluded from review. Healthcare providers at the individual sites conducted the reviews and were extensively trained on the use of the tool and provided with an instruction manual. For the purposes of the trial, an adverse drug event (ADE) was defined as injury caused by a drug or nonuse of a drug. The preventability of the ADE was also determined at the site by 2 reviewers. Reviewers also recorded ADEs not associated with a specific trigger.

The outcome measures were the rate of ADEs, number of triggers per patient, positive predictive value of the trigger, severity of the ADE, percentage of preventable ADEs, stage of the medication use process where the medication error occurred, diagnosis on discharge, and the hospital unit where the ADE occurred.

A total of 960 charts were reviewed, which represented 6806 patient days. Analysis of demographic data revealed an average length of stay of about 7 days, an average patient age of 5.9 years, 14.3 medications on average per patient with 90.5 doses. Overall 2388 triggers were identified, mean rate of 2.49 per patient (95% confidence interval (CI): 2.39 to 2.59). One hundred seven ADEs were found for a mean rate of 11.1 per 100 patients (95% CI: 9.13 to 13.5), 15.7 per 1000 patient days (95% CI: 12.9 to 19), and 1.23 per 1000 doses (95% CI: 1.01 to 1.49).

The positive predictive value of the trigger tool was 3.7%. The majority of ADEs, 104 (97.2%) of 107 were classified as resulting in temporary harm and requiring intervention. The remaining 3 (2.8%) were classified as resulting in temporary harm and requiring hospitalization (initial admission or extended length of stay). None of the events led to permanent harm, need for a life-saving intervention, or death. A total of 22% were classified as preventable ADEs, 17.8% could have been detected sooner, and 16.8% could have been resolved more effectively. This study also demonstrated the shortcomings of relying on error reports in the hospital setting, as only 4 (3.7%) of 107 were reported via the hospital’s occurrence report system. When the medication use process was examined, the monitoring phase (62.5%) was most commonly associated with preventable ADEs such as not assessing the regimen for continued need and not using clinical and laboratory monitoring when indicated. The second most common phase was prescribing or ordering (50%, errors could overlap in the process).

Analgesics/opioids was the class of drugs most frequently associated with ADEs (51%), and the events were most common in hematology/oncology units (18 per 100 patients). Pruritis was the most common ADE (17.8%). The authors concluded that the trigger tool was effective in identifying ADEs in hospitalized pediatric patients, and that event rates were higher than described in previous literature. The majority of events caused temporary harm to the patients.

The Joint Commission Sentinel Alert
Following the release of Takata’s study and in light of recent high profile medication errors in children, The Joint Commission issued a sentinel event alert on April 11, 2008 on this topic. The alert points out that children are at higher risk for medication errors and subsequent harm as compared to adults for several reasons: • Commercially available dosage forms often need dilution or adjustment when given to pediatric patients. • Facilities lack adequately trained clinicians, notably in the emergency department. • Young children, as well as those who are small and/or ill may have less developed renal and hepatic function and may not tolerate an error to the same degree an adult would. • Young children cannot communicate problems they are experiencing.

The alert contains data from the United States Pharmacopeial Convention’s (USP) MEDMARK database for 2006 and 2007. The data reveal that incorrect dosing is the most common reason for errors (37.5%) followed by omission errors (19.9%). Errors in pediatric patients were most often linked to performance deficit (43%) and knowledge deficit (29.9%). Additional causes were calculation errors, computer entry errors, insufficient monitoring, and not using pumps correctly.

The Joint Commission’s alert contains several strategies to reduce medication errors in this population (http://www.jointcommission.org/SentinelEvents/SentinelEventAlert/sea_39.htm). Some key recommendations include using oral syringes to administer oral medications to prevent inadvertent intravenous administration, orientation of all pharmacy staff to pediatric and neonatal pharmacy services, development of preprinted order forms to standardize care, and appropriate use of technology.

The Joint Commission also provides a list of 9 suggested actions to prevent medication errors in pediatric patients. The emphasis lies in accurate dosing and appropriate formulations. Initially an accurate weight in kg for pediatric patients should be obtained at the time of admission or within 4 hours of an emergency situation. Furthermore, prescriptions should contain the dose and the dose per weight to allow for a double-check prior to dispensing. Pediatric strength formulations should be stocked in the institution to the extent possible. Adult formulations should be clearly distinguished from pediatric formulations with warning labels.

Pediatric patients are more at risk for medication errors compared to adults for a variety of reasons such as immature organ function, lack of specific drug formulations, and lack of specialty practitioners available for consultation. Recent high profile medications errors involving the incorrect strength of heparin being used for neonates has increased the attention on errors in pediatric patients. Takata and colleagues’ trigger tool represents a way for institutions to identify common ADEs allowing for analysis of root causes. Implementation of the strategies offered by The Joint Commission may further reduce the potential for errors in pediatric patients.

References

Takata GS, Mason W, Taketomo C, Logsdon T, Sharek PJ. Development, testing, and findings of a pediatric-focused trigger tool to identify medication-related harm in US children’s hospitals. Pediatrics. 2008;121(4):e927-e935.

The Joint Commission. Sentinel event alert: preventing pediatric medication errors. http://www.jointcommission.org/SentinelEvents/SentinelEventAlert/sea_39.htm. Accessed April 21, 2008.