Studies of all Types of Errors

Few studies have examined medical errors of all types in outpatient pediatrics. None of these have high sensitivity to detect the majority of these events. The Learning from Errors in Ambulatory Pediatrics study evaluated the scope, range, potential causes, and possible solutions to medical errors in pediatric ambulatory care. Among 14 participating pediatric practices, 147 medical errors were reported. The largest group of errors was related to medical treatment (37%), but errors were also associated with patient identification (22%), preventive care including immunizations (15%), diagnostic testing (13%), patient communication (8%), and other causes. Of the medical treatment errors, 85% were related to medications. Among the medication errors, 55% were related to ordering, 30% to failure to order, 11% to administration, 2% to transcribing, and 2% to dispensing. No denominator data were collected, so the actual error rates could not be determined.

Phillips and colleagues studied medical errors reported by clinicians, staff, and patients in 10 family medicine clinics of the American Academy of Family Physicians National Research Network. Participants were asked to report routinely during 10 weeks and to report every error on 5 specific days. Four hundred one clinicians and staff anonymously reported 935 errors within 717 events, 37% of which came from the 5 intensive reporting days and 61% from routine reports. Staff made 384 (53%) reports, primarily related to patient flow and communication, and clinicians made 342 (47%) reports, mainly medication and laboratory errors. Most (96%) errors reported were related to processes, not to knowledge or skill. Patients reported 18 errors. Among patients affected by errors, there was no breakdown by age, although children were included.

Neuspiel and colleagues reported a voluntary, anonymous, nonpunitive error-reporting project in an outpatient pediatric department in New York City with approximately 36,000 visits per year (2.2 errors reported per 1000 visits per year). In the first year of the project, 80 errors were reported, compared with only 5 errors reported during the prior year via a traditional incident reporting system. Most reports originated from physicians (45%) and nurses (41%). Errors were classified as involving office administration (34%), medications and other treatment (24%), laboratory and diagnostic testing (19%), and communications (18%). The most frequent office administration errors included wrong demographic information or date of visit (9%), misfiled papers in chart (9%), and delays in processing patients due to misplaced registration forms (9%). The most frequent errors attributed to medication and other treatment included errors in vaccine and medication orders (6%), wrong vaccine administered (5%), wrong outside prescription dispensed (4%), incomplete prescription returned by pharmacy (4%), and patient revaccinated too early (4%). Among errors attributed to laboratory and diagnostic testing, the most common were missed specimen pickup (10%) and mislabeled specimens (4%). The most frequent communication errors were patients leaving the practice before ordered vaccines were administered (8%).

In a pediatric outpatient department in Charlotte, North Carolina, with approximately 26,000 visits per year, employing a similar voluntary, anonymous reporting system, Neuspiel and colleagues reported 216 medical errors over 30 months (3.3 errors per 1000 visits per year). Most reports (77%) originated from nurses, physicians, and midlevel providers. The most frequently reported errors were misfiled or erroneously entered patient information (31%), laboratory test delayed or not performed (13%), error in medication prescribing or dispensing (11%), vaccine error (10%), patient not given requested appointment or referral (7%), and delay in office care (7%).

Studies of Malpractice Claims and Diagnostic Errors

One way of looking at the impact of medical errors is via paid malpractice claims. Bishop and colleagues studied data from the National Practitioner Data Bank from 2005 to 2009. Among outpatient claims, 8.4% (95% confidence interval [CI] 7.6–9.2) were for children 0 to 19 years. The most frequent basis for outpatient claims were errors in diagnosis (45.9%, 95% CI 44.4–47.4), followed by errors in treatment or medication (29.5%, 95% CI 28.2–30.9).

Singh and colleagues evaluated diagnostic errors among pediatricians at 3 tertiary care institutions and 109 affiliated clinics via anonymous survey. Among 1362 pediatricians, 726 (53%) responded, and 54% reported making a diagnostic error at least once or twice per month, with higher rates (77%) among trainees. Diagnostic errors causing harm to patients and occurring at least once or twice per year were reported by 45% of respondents. The most frequently reported process source of diagnostic error was failure to gather information through patient history, physical examination, or chart review. The most commonly reported system error was inadequate care coordination and teamwork. Most frequent specific errors were viral illnesses diagnosed as bacterial, misdiagnosis of medication adverse effects, psychiatric disorders, and appendicitis.

Studies of Outpatient Medication Errors

To date, the most widely studied type of error in the pediatric office environment has related to prescribing, dispensing, and home administration of medications. McPhillips and colleagues randomly selected 1933 children receiving a new prescription using automated pharmacy data from 3 health maintenance organizations, 2 using paper prescriptions and 1 using an electronic prescription writer. They determined that 15% of children had potential dosing errors: 8% overdoses and 7% underdoses. In children less than 35 kg, 33% had dosing errors, and more than 1% were prescribed twice the recommended maximum dose. Twenty percent of children under 4 years of age experienced dosing errors, compared with 13% of children aged 4 to 12 years (odds ratio [OR] 1.7, 95% CI 1.1–2.6). Those receiving 5 or more prescriptions were more likely to have dosing errors than children with a single prescription (OR 3.3, 95% CI 1.4–7.7). The most frequent overdosed medications were analgesics (15% overdosed), and the most frequent underdosed medications were antiepileptics (20% underdosed). There were no reductions in error rates in the site using an electronic prescription writer.

A series of publications has emerged from a prospective cohort study at 6 office practices in the Boston area over 2 months. Kaushal and colleagues discovered 57 preventable adverse drug events (rate 3%, 95% CI 3%–4%) in the care of 1788 patients. None of the events was determined to be life threatening, but 8 (14%) were serious. Forty (70%) were related to parental drug administration. The authors determined that improved communication between providers and parents and between pharmacists and parents was the preventive strategy with the most potential benefit to prevent these errors. Children with multiple prescriptions were at increased risk of preventable adverse drug events (OR 1.46, 95% CI 1.01–2.11). Taking more than 1 medication (OR 1.68, 95% CI 1.15–2.46) and age under 5 years (OR 2.35, 95% CI 1.05–5.28) were associated with risk of medication administration errors. Kaushal and colleagues showed that 94% of medication errors with minimal potential for harm and 60% of near misses occurred at the prescribing stage, most frequently due to inappropriate Abbreviations, dosing errors, and illegibility.

Several studies have looked at prehospital antipyretic dosing errors in children seen in pediatric emergency departments. Li and colleagues found that 51% of surveyed caregivers gave inaccurate doses of acetaminophen (62% inaccurate) or ibuprofen (26% inaccurate), especially to infants younger than 1 year of age (relative risk [RR] 1.40, 95% CI 1.06–1.86 compared with older children). Another study determined that 53% of children received an improper antipyretic dose at home. Goldman noted that 47% of parents gave acetaminophen at recommended doses; 12% gave overdoses, and 41% underdosed their children. Among parents with febrile children, 54% would not have come to the emergency department had the fever subsided at home. Parents speaking English as a primary language were more likely to give recommended doses than non-English speaking parents.

Medication errors by pediatric emergency department staff have also been reported. Losek examined errors in the application of standing orders for acetaminophen in a pediatric emergency department (ED). In a 1-week period, there were 122 (78%) correct doses, 15 (10%) underdoses, and 19 (12%) overdoses. The rectal route of administration was associated with significantly greater overdoses (35%, 95% CI 14%–48%) than orally administered acetaminophen (8%).

In the pediatric emergency department of the Hospital for Sick Children in Toronto, Canada, among 1532 children, 10.1% had medication prescribing errors. Error risk was increased when medications were ordered by trainees (OR 1.64, 95% CI 1.06–2.52) and in seriously ill patients (OR 1.55, 95% CI 1.06–2.26). Shaughnessy and Nickel reviewed prescriptions written by 20 family medicine residents. Among 1814 prescriptions, 21% (n = 373) contained at least 1 prescription-writing error, including omissions (6%), unfulfilled legal requirements (1%), incomplete directions (1%), dose or direction errors (3%), unclear quantity to be dispensed (3%), or prescriptions written for nonprescription products (5%). The proportion of pediatric prescriptions was not indicated. Condren and colleagues found errors among 9.7% of prescriptions written by pediatric residents in an outpatient clinic. The most frequent type of error was incomplete information entered (42%), followed by medication dosing errors (34%).

Several investigators have studied ambulatory medication errors in children with specific health conditions. Rinke and colleagues analyzed both inpatient and outpatient error reports involving antidepressants in children under 18 years from the United States Pharmacopeia MEDMARX database from 2003 to 2006. Among 451 error reports, 95% reached the patient; 6.4% required increased monitoring and/or treatment, and 77% involved off-label medication use. The cause of the error was cited as administration in 33% of cases, dispensing in 30% of cases, transcribing in 28% of cases, and prescribing in 7.9% of cases. The most frequent medications cited were sertraline (20%), bupropion (19%), fluoxetine (15%), and trazodone (11%). Outpatient errors involved more dispensing errors and errors due to inaccurate or omitted transcription (both P <.001) compared with inpatient errors. No denominator information was available to establish error rates.

Taylor and colleagues studied outpatient oral chemotherapy medication errors in children with acute lymphoblastic leukemia. During a 2-month period at a single center, 1 or more errors occurred in 17 of 172 medications (9.9%). Of the 17 errors, 12 were related to administration and 5 to prescribing; there were no pharmacy dispensing errors. All errors were related to incorrect dosing or failing to administer an indicated medication. Among the 69 study patients, 13 (18.8%) experienced at least 1 medication error.

In a study of medication errors in the homes of children with sickle cell disease and seizure disorders, Walsh and colleagues directly observed medication administration, reviewed medications, and checked prescription doses. In 52 home visits, 280 medications were reviewed, and 61 medication errors were detected (95% CI 46–123), including 31 with potential for injury and 9 with actual injury to the child. Frequent sources of error were when parents failed to fill prescriptions or to change doses due to communication problems. These errors led to further testing or continued pain, inflammation, seizures, vitamin deficiencies, or other injuries. In some cases, communication failures between 2 parents resulted in administration errors and difficulty preparing medications for administration. Among parents using support tools (eg, alarms, reminders) for home medication use, error rates were 44%, compared with 95% among parents not using such tools ( P = .0002). Patients’ physicians were unaware of 80% of the detected errors.

Studies of Immunization Errors

Several investigators have reported studies on immunization errors noting that fragmentation of care leads to unnecessary overimmunization. Feikema and colleagues, using data from the US 1997 National Immunization Survey, found that 21% of children were overimmunized, and 31% were underimmunized for at least 1 vaccine. The largest contributors to overimmunization were having more than 1 immunization provider (OR 2.8, 95% CI 2.4–3.2) and having multiple types of providers (OR 2.0, 95% CI 1.6–2.4). The 1997 costs associated with extra vaccination were estimated conservatively at $26.5 million. Butte and colleagues, in a large urban clinic, determined that 35.5% of pediatric patients had at least 1 invalid vaccine dose. Darden and colleagues, using the 1999 to 2003 National Immunization Survey, determined that among 19- to 35-month-old children, 21.9% were underimmunized, and 9.4% received extra immunizations. Overimmunization was associated with multiple providers (adjusted odds ratio [AOR] 2.3, 95% CI 2.13–2.47) and multiple facility types (AOR 4.67, 95% CI 4.23–5.15).

Measurement Devices and Home Medication Errors

Yin and colleagues studied parental liquid medication administration errors related to dosing instrument type and health literacy. In a single center in New York City, 302 parents were observed for accuracy of a 5 mL dose using several standardized instruments: dosing cup with printed calibration markings, dosing cup with etched markings, dropper, dosing spoon, and oral syringes with and without a bottle adapters. Parents dosed least accurately (within 20% of recommended dose) using the cup with printed (30.5%) and with etched markings (50.2%). Over 85% of parents dosed accurately with the other instruments. After adjustment for confounders, cups were associated with increased odds of making a dosing error compared with oral syringe (printed cup: OR 26.7, 95% CI 16.8–42.4; etched cup: OR 11.0, 95% CI 7.2–16.8). Cups were also more likely than oral syringes to be associated with making large dosing errors (printed cup: OR 7.3, 95% CI 4.1–13.2; etched cup: OR 6.3, 95% CI 3.5–11.2). Dosing errors were also more frequent with limited health literacy (OR 1.7, 95% CI 1.1–2.8). Sobhani and colleagues also found that adults were able to measure liquid acetaminophen more accurately with an oral syringe than a cup, but only two-thirds of participants measured acceptable doses with the syringe.

Yin and colleagues determined the frequency of inconsistent dosing directions and measurement devices among pediatric over-the-counter (OTC) medications at the time (November 2009) the US Food and Drug Administration (FDA) released new recommendations for greater consistency and clarity in dosing directions and measuring devices. They studied 200 top-selling pediatric oral liquid OTC medications during the year preceding the new FDA recommendations for inclusion of a measuring device, consistency between labeled dosing directions and markings on the measuring device, use of nonstandard units and Abbreviations, and presence of Abbreviation definitions. The investigators found that measuring devices were included with 148 of 200 products (74%). There were inconsistencies between medication dosing directions and measurement device markings in 146 of 148 products (98.6%), including missing (N = 36, 24.3%) and superfluous markings (N = 120, 81.1%). Of all products, 11 (5.5%) used atypical measurement units for listed doses. Non-standard Abbreviations for milliliter were used in 97 products. Among products using Abbreviations, 163 did not define at least 1 Abbreviation. The authors conclude that at the time of the new FDA recommendations, top-selling pediatric OTC medications were using highly variable and inconsistent dosing directions and measuring devices.

The readability of retail pharmacy-generated consumer medical information and the features of measuring devices for oral liquids were studied by Wallace and colleagues. Investigators filled similar prescriptions for prednisolone and amoxicillin at 20 different pharmacies in Colorado, Georgia, and Tennessee. Many materials were at ninth-11th grade readability level, too high for many parents, and 3 pharmacies provided no materials at all. One-third of pharmacies provided measuring devices that would require multiple measurements for the prescribed doses.

Health Literacy, Numeracy, Cultural, and Language Barriers

Leyva and colleagues conducted a study to examine the impact of language barriers on medication errors in a Spanish-speaking population in the Bronx, New York. They sought to determine how well Bronx Spanish-speaking Latino parents of children 5 years and younger understand written medication instructions. After being given instructions on administration of ferrous sulfate, only 22% of parents demonstrated correct medication administration (amount and frequency). Subjects reporting comfort speaking English were more likely to demonstrate correct medication amount to be administered (50% vs 21%, OR 3.8; 95% CI, 1.2–12.2) and correct frequency (93% vs 51%, OR 12.4; 95% CI, 1.5–99.1). Both education (OR 1.22, 95% CI, 1.03–1.45) and comfort speaking English (OR 3.81, 95% CI, 1.13–12.86) independently predicted correct medication dosing.

Sharif and Tse surveyed all pharmacies in the Bronx, New York, and visited selected pharmacies to assess computer software used to generate Spanish medication labels. Among 316 pharmacies, 286 (91%) participated. Of these, 209 (73%) provided medicine labels in Spanish. Pharmacies providing Spanish labels mostly frequently (86%) used computer programs to generate them. Lay staff members translated 11% of labels, and 3% used professional interpretation services. Fourteen different computer programs were used to generate Spanish labels, but 3 major programs were used by 70% of pharmacies. The investigators studied 76 medication labels generated by 13 different computer programs, and found that 32 Spanish labels (43%) included incomplete translations (mixture of English and Spanish), and 6 additional labels contained misspellings or grammar errors. The overall error rate was 50%. They concluded that while pharmacies were able to provide Spanish labels, the translation quality was inaccurate and hazardous.

Flores and Ngui also point to racial, ethnic, and language barriers as contributors to increased risk of pediatric errors. Yet in 2004, the majority of pediatricians reported using untrained interpreters to communicate with patients and their families with limited English proficiency.

Adolescents are appropriately given increased responsibility for self-care, with less parental supervision, yet this may increase their risk for home medication errors. Wilson and colleagues evaluated misconceptions and knowledge gaps among teens about OTC medications. They determined that 78% of the adolescents had used OTC medications in the previous month, most commonly ibuprofen and acetaminophen. Although 35% of respondents reported knowing about acetaminophen, 37% did not know it was the same as Tylenol, and many had limited knowledge about adverse effects and contraindications. Hispanic teens reported less use of acetaminophen and had lower knowledge scores than other ethnicities.

Lokker and colleagues note that parents commonly misunderstand labels on OTC pediatric cough and cold preparations. They recruited 182 caregivers from clinics at 3 institutions, and questioned them about the use of 4 common OTC medications. Mean education level of caregivers was 12.5 years, but only 17% had higher than ninth grade numeracy skills. Although all of the medications were labeled with advice to consult a physician for use in children younger than 2 years, over 50% of the time parents would give these products to a 13-month-old child.

Chang and colleagues surveyed parents of febrile children less than 6 years of age hospitalized in Taiwan. After being provided with written medication instructions for antipyretics, one-third of parents had more than 1 misunderstanding for medication timing, time interval of administration, and/or dosage, and almost two-thirds of parents misunderstood acetaminophen adverse effects. Poorer comprehension of instructions was associated with lower academic qualifications.

Costelloe and colleagues found limitations in the ability of parents to measure weights of preschool children at home for calculation of antipyretic dose. Only 40% of parents had home scales, most of which were analog. Research scale weights were heavier than those from home scales, with a mean difference of 0.41 kg (95% CI 0.24 –0.74 kg).