OBJECTIVES. This study was designed to evaluate (1) the incidence and common types of adverse drug reactions among hospitalized children, (2) the frequency of adverse drug reaction reporting by health care providers, and (3) the follow-up processes resulting from adverse drug reactions.
METHODS. A retrospective cohort study of pediatric patients who experienced an adverse drug reaction between January 1, 1995, and December 31, 2004, was conducted at a community-based, tertiary care, children's teaching hospital.
RESULTS. A total of 1087 adverse drug reactions were reported; the overall incidence was 1.6%. The severity of most adverse drug reactions was low (levels 1–3: 89%; high levels 4–6: 11%). Adverse drug reactions with low severity were significantly more common in both the general pediatric unit and the NICU. Adverse reactions resulting from use of antibiotics (particularly penicillins, cephalosporins, and vancomycin) were usually mild. In contrast, adverse drug reactions rated high in severity were significantly more common among reactions that led to hospital admission or occurred during surgery and among certain drug classes, including anticonvulsants and antineoplastic agents. Adverse drug reactions were reported by pharmacists (89%), nurses (10%), and physicians (<1%). Although documentation of physician notification occurred for 93% of adverse drug reactions, only 29% of cases were documented in the patient's medical chart, 13% included follow-up education for individuals involved, and 10% were updated in the allergy profile of the hospital computer system.
CONCLUSION. Measures to improve detection and reporting of adverse drug reactions by all health care professionals should be undertaken, to enhance our understanding of the nature and impact of these reactions in children.
Adverse drug reactions (ADRs) are described as toxic responses to a medication and can be physical or psychological.1 Monitoring and documentation of ADRs are crucial to encourage and to ensure safe use of pharmacologic agents. A serious ADR leading to a new black box warning or withdrawal from the market was detected for 1 of 5 medicines during postmarketing surveillance in the past 25 years.2 In the United States, most ADRs are reported to the Food and Drug Administration by drug manufacturers. Hospitals are required to monitor routinely for adverse drug experiences, including preventable and nonpreventable ADRs, and to report all ADRs that result in a sentinel event, on the basis of regulations established by the Centers for Medicare and Medicaid Services and the Joint Commission on Accreditation of Healthcare Organizations.3,4 In this setting, the documentation of ADRs relies heavily on spontaneous reporting by health care professionals.
ADRs can lead to significant morbidity and death among children.5–12 ADRs not only may result in hospital admission or prolonged hospitalization but also may lead to permanent disability or even death. Notably, in a meta-analysis by Lazarou et al,5 fatal ADRs among both adults and children ranked as the fourth to sixth leading cause of death in the United States. Another study demonstrated that ADRs were associated with an average of 243 reported deaths among young children, from newborn to 2 years of age, each year.10 Although it has not yet been studied among children, the annual direct cost to manage ADRs among hospitalized adults was estimated at $1.6 to 4.2 billion.13
The incidence of ADRs among hospitalized children in the United States has not been well studied. On the basis of a meta-analysis of 17 prospective studies conducted in the United States and Europe, the incidence of ADRs among hospitalized children was 9.5%, with severe reactions accounting for 12% of the total.6 The incidence of ADRs leading to admission to a pediatric hospital was 2%. Studies evaluating ADRs in the pediatric population explicitly in the United States were insufficient. On the basis of 3 studies conducted in the United States, the incidence of ADRs among hospitalized children was reported to vary from 0.85% to 16.8%.7–9 In addition to the uncertainty regarding the exact incidence of ADRs, no study to date has evaluated the difference in the severity of ADRs among children.
Because clinical trials involving neonates, infants, children, and adolescents are limited, the safety and tolerability of many pharmacologic agents are not well established. Often the pharmacologic actions of drugs in neonates, infants, and children are not similar to those identified for adults; therefore, information obtained from research with adults cannot be applied directly.14 The Pediatric Rule for Labeling, which was first issued in 1994 by the Food and Drug Administration, and its 1998 enactment requiring manufacturers of certain new and marketed drugs to conduct studies for pediatric labeling have increased the information available regarding drug safety for children.15 However, pediatric data remain inadequate, as indicated by the inclusion of a disclaimer regarding a lack of safety and efficacy information for use among children for many currently marketed drugs. As a result of these contributing factors, children may have greater potential to experience ADRs.
Considering the impact of ADRs on morbidity and mortality rates and the potential vulnerability of children, especially if hospitalized, to experience ADRs, studies to evaluate the incidence and nature of ADRs in this population are warranted. Our study was designed to evaluate (1) the incidence and common types of ADRs, with respect to severity level, among hospitalized children, (2) the frequency of ADR reporting by health care providers, and (3) the follow-up processes after ADRs.
This retrospective cohort study was conducted at Miller Children's Hospital, a community-based, tertiary care, teaching hospital with 197 beds (including 69 NICU, 20 PICU, 94 general pediatrics unit, and 14 hematology/oncology unit beds). Although ADRs may be reported voluntarily by any health care professionals (eg, physicians, pharmacists, or nurses) with a standardized, handwritten form, most ADRs are reported by pharmacists. The pharmacists are trained to record ADRs through the policy and procedure manual for reporting of medication errors and ADRs, approved by the Pharmacy and Therapeutics Committee. The manual contains both the definition of an ADR and the process for documentation. To ensure consistency in classifying severity and causality, a pharmacist reviews all ADRs.
ADRs are identified through several methods, including direct observation, prospective evaluation of drugs with strong potential to cause ADRs (eg, vancomycin), investigation of new orders for certain medications that can be used as antidotes (eg, diphenhydramine and naloxone), participation in medical rounds, and notification by physicians and nurses. In our study, ADR report forms that were completed for pediatric patients (including neonates, infants, children, and adolescents) during their hospital stays between January 1, 1995, and December 31, 2004, were reviewed. The institutional review board at Miller Children's Hospital approved the study, to ensure protection of health information.
Data for each ADR were collected on a standardized data collection form that included the following information: patient demographic features (age, gender, drug allergy, admitting diagnosis, and location of care), ADR description (suspected drug name and class, dates on which drug administration was started and discontinued, and date and description of reaction), and ADR analysis/health outcome (severity level, causality, length of stay, person reporting, and follow-up processes). The nomenclature used for drug class was based on the American Hospital Formulary Service classification system.16 The annual number of drug prescriptions processed and hospital admission census data were obtained from pharmacy and information systems. Electronic medical records were reviewed for incomplete or inconsistent information on the ADR forms.
To ensure accuracy and consistency in the recording of the severity level, 1 qualified pharmacist reviewed all ADR reports and assigned severity levels, which ranged from 1 to 6 (Table 1). 17,18 ADRs with low severity were defined as those assigned levels 1 to 3, with high severity as levels 4 to 6. The causality of ADRs was classified as definite, probable, possible, or conditional (Table 2). ADRs also were categorized as either pharmacologic or allergic/idiosyncratic. Pharmacologic reactions result from a drug's pharmacologic characteristics and are usually dose-dependent and predictable and rarely life-threatening. In contrast, allergic/idiosyncratic reactions are unrelated to a drug's pharmacologic characteristics and are likely to be serious and life-threatening.19
According to the World Health Organization, an ADR is defined as any injurious, unintended, and undesired response to a drug administered at doses used normally among humans for prophylaxis, diagnosis, or therapy.20 This definition excludes drug misadventures, therapeutic failures, intentional or accidental poisoning, and drug abuse. This conservative definition was used in this study. Therefore, ADRs that resulted from drug errors, therapeutic failures, intentional or accidental poisoning, drug abuse, or noncompliance were excluded from this study.
To calculate the overall incidence of ADRs, the total number of patients with ADRs was divided by the total number of hospital admissions during the study period and the result was multiplied by 100. This simple calculation was also used to determine the incidence of ADRs according to year. The rate of ADRs per 1000 medication orders was determined by dividing the number of ADRs by the number of medication orders filled for the respective year and multiplying the result by 1000.
Descriptive statistics were used to analyze demographic data. The Pearson χ2 test (2-sided) was used to evaluate differences in categorical variables (eg, admitting diagnosis, location during reaction, drug class, suspected drug name, type of reaction, and causality) between ADRs with low versus high severity. Fisher's exact test (2-sided) was used for categorical data with small size. The t test was used for continuous variables (eg, age, incidence of ADRs according to year, and length of stay). Differences were considered significant if the P value was <.05. All statistical analyses were performed with SPSS for Windows 13.0 (SPSS Inc, Chicago, IL).
Of the 1235 ADR report forms reviewed, 148 were excluded, for the following reasons: 95 resulted from drug misadventures, 49 occurred in an outpatient setting and the patient was not hospitalized, 2 were accidental ingestions, and 2 were suicide attempts. The remaining 1087 ADRs were included in our study.
The overall incidence of ADRs per hospital admission during the 10-year study period was 1.6%. The annual incidence of ADRs ranged from 0.4% to 2.3% (Fig 1). A significant increase in the reporting of ADRs was observed between 1998 and 1999 (0.9% vs 2.0%; P < .001). On the basis of the number of medication orders, the annual rate of ADRs was 1.2 to 1.8 ADRs per 1000 medication orders for years 2001 to 2004.
The severity of most ADRs was considered low (levels 1–3: 89%; levels 4–6: 11%). Two thirds of ADRs resulted in either discontinuation of administration of the suspected drug or changes in drug dosing or frequency, with possible use of an antidote or other treatment (categorized as severity level 3). The characteristics of patients with ADRs rated low versus high in severity were similar (Table 3). The mean age was 7.0 ± 6.2 years. The most common admitting diagnoses were infectious diseases or fever (20%), hematologic malignancies (16%), prematurity (11%), and solid tumors (9%). Approximately 2.5% of ADRs that occurred in outpatient settings led to hospital admission. Most patients (46%) had no known drug allergies.
The distributions of ADRs according to patient location during reaction were similar for the general pediatrics unit, hematology/oncology unit, and NICU/PICU, each of which accounted for one third of total ADRs (Table 3). In comparisons according to severity level, ADRs with low severity were significantly more common for both the general pediatric floor and the NICU. In contrast, ADRs with high severity were significantly more common for patients who experienced ADRs during surgery or who experienced ADRs that led to hospital admission. Anaphylaxis (including Stevens-Johnson syndrome) and cardiopulmonary depression (including asystole and apnea) were most frequent among ADRs leading to hospital admission or occurring during surgery. Although anticonvulsants (valproic acid, phenobarbital, and phenytoin) were the drugs associated most commonly with ADRs that occurred before hospital admission, cefazolin and general anesthetics (sevoflurane, desflurane, and propofol) were associated most commonly with ADRs that occurred during surgery. The median length of hospitalization for all ADRs was 12.0 days (range: 1–361 days). No statistically significant differences were detected with respect to age, gender, history of drug allergy, or length of hospitalization for ADRs classified as low versus high severity.
Drug Classes and Specific Drugs
Antibiotics (33%), narcotic analgesics (12%), anticonvulsants (11%), and anxiolytic agents (10%) were the drug classes associated most frequently with ADRs. ADRs might involve multiple suspected drugs. Although reactions with antibiotics, especially penicillins, cephalosporins, and vancomycin, were usually mild, anticonvulsants and antineoplastic agents were associated more commonly with reactions rated high in severity (Table 4). Among specific drugs, vancomycin (12%), morphine sulfate (5%), and fentanyl, metoclopramide, lorazepam, asparaginase, and valproic acid (∼3% each) were associated most commonly with ADRs. Red-man syndrome was the primary reaction observed with the use of vancomycin. No statistical differences were detected between the groups for most drugs except vancomyin and asparaginase (Table 5).
Causality, Type, and Time of Reactions and Deaths
The causality of most ADRs was probable (44%) or possible (44%). Only 8% of ADRs had definite causality. Compared with ADRs rated low in severity, definite causality was significantly more common with ADRs rated high in severity (low: 7%; high: 15%; P = .003).
The types of reaction were distributed equally among pharmacologic (n = 536; 49%) and allergic/idiosyncratic (n = 551; 51%). The organ systems affected most commonly were dermatologic (37%), cardiovascular (23%), and neurologic (16%). On the basis of severity level, rash, flushing, and pruritus were associated more commonly with ADRs rated low in severity (Table 6). In contrast, high-severity ADRs were more common with arrhythmias, respiratory depression, elevated liver function test results or jaundice, and difficulty breathing or chest wall rigidity.
The median time from initiation of drug administration to detection of reaction was longer for high-severity reactions (3 days; range: 1–497 days), compared with low-severity reactions (1 day; range: 1–716 days; P = .073). Most ADRs, independent of severity, were detected within 1 day after initiation of administration of the suspected drug. Among ADRs with high severity, 40% resulted from chemotherapeutic agents (eg, endocrine effects resulting from asparaginases) and 24% were hypersensitivity reactions (including Stevens-Johnson syndrome) caused primarily by tricyclic anticonvulsants (eg, phenytoin and carbamazepine).
Two deaths associated with ADRs are described in Table 7. The 13-year-old patient who was transferred from another hospital was receiving anesthesia for MRI of her thoracic scoliosis. The patient experienced full cardiac and respiratory arrest after receiving midazolam, fentanyl, and propofol. After 3 attempts at resuscitation, the patient was revived and then transported to our hospital for continued care. The patient was pronounced brain dead on the basis of examination and isoelectric electroencephalographic results. The patient did not have any significant medical history except for developmental delay.
Persons Reporting and Follow-up Processes
The ADRs were reported by pharmacists (89%), nurses (10%), and physicians (<1%). On the basis of the information provided on the ADR forms, physician notification of ADRs occurred frequently. However, other follow-up processes to address the ADRs did not occur for most ADRs (Table 8). All follow-up processes were performed for ∼0.8% of ADRs.
Most studies evaluating pediatric ADRs were conducted in Europe. The need for more studies evaluating ADRs in children is evident in the lack of published clinical studies specifically in the United States and our limited knowledge of the safety of many pharmacologic agents that are currently on the market. Our study included the largest sample of pediatric patients with ADRs reported to date; the largest reported previously consisted of 565 patients.7 Our study concludes that the incidence of ADRs among hospitalized children is ∼1.6%. In 3 published studies conducted in the United States, the incidence of ADRs in children varied from 0.85% to 16.8%.7–9 The most recent study reviewed ADRs up to 1999.7 The 2 other studies were conducted before 1980; therefore, the validity of extrapolating the results from those studies to ascertain the current incidence and impact of ADRs is limited.
Several factors may contribute to the difference in the incidence of ADRs in our study, compared with other studies. First, the low incidence in our study may reflect underreporting of ADRs. At our hospital, ADR documentation by health care professionals relies on voluntary reporting, which, according to 2 studies, may identify only 4% to 6% of adverse drug events.21,22 Second, certain ADRs (eg, infusion-related effects with intravenously administered immunoglobulins and red-man syndrome with vancomycin) occur quite frequently in our study population and are not reported consistently. In addition, our study demonstrated that most ADRs were reported by pharmacists. Clinical pharmacy services, with participation in patient-care rounds in the NICU, PICU, and hematology/oncology unit, resulted in the high percentage of pharmacists reporting ADRs. Because physicians and nurses have direct patient contact, they have the most opportunities to detect and to report ADRs; however, <11% of the ADRs in our study were reported by physicians or nurses.
Increasing awareness of both the importance of documenting ADRs and the structure of the reporting system within the hospital is important for improving reporting by all health care professionals. In addition, simplifying the reporting process can encourage physicians and nurses to document ADRs themselves or to solicit follow-up reporting of ADRs by pharmacists. Strategies that have been successful in stimulating reporting include 24-hour reporting hotlines, ADR newsletters, in-service educational programs, reminder letters for reporting certain drugs and reactions, participation of a clinical pharmacist in daily rounds, and use of morning report as a forum for detection and documentation of ADRs.23–27 Formation of a multidisciplinary ADR committee also may improve reporting and may encourage collaborative efforts of all health care professionals.23 In a study conducted in a pediatric ward, a computer system that analyzed changes in laboratory data led to a twofold increase in detection of ADRs.28 Computer surveillance to detect potential adverse drug events seems more efficient and more promising than voluntary reporting and chart review.22,29 Furthermore, computer-based monitoring can reduce the incidence and severity of adverse drug events.30,31
The significant increase in ADR reporting after 1998 might have been affected by the enactment of the Pediatric Rule for Labeling, which advanced the understanding of the safety of drugs used for the pediatric population and thereby enhanced recognition of potential ADRs among the pharmacists at our hospital. Other factors that might have contributed to this increase include improved awareness of the need to report ADRs to meet standards established by the Centers for Medicare and Medicaid Services and the Joint Commission on Accreditation of Healthcare Organizations, improved reporting processes with simplification of the ADR reporting form, better accessibility of computerized references (eg, Micromedex and Poisondex), increased usage of drugs for pediatric patients with more-complex diseases, improved accessibility of electronic medical records containing patient information, and increased numbers of clinical pharmacists.
Consistent with the results of other studies,7,32 antibiotics, narcotic analgesics, and anticonvulsants were the drug classes associated most frequently with ADRs. ADRs that are common (eg, chills and rigor with the use of intravenously administered immunoglobulins) are not documented consistently at our hospital. This may contribute to the minor variability in the frequencies of drug classes.
Approximately 11% of ADRs were considered of high severity, defined as those requiring transfer to a higher level of care or leading to disability or death (Table 1). A meta-analysis of 39 prospective studies (only 2 of which were pediatric) from US hospitals estimated that the combined incidence of serious and fatal ADRs was 7%.5 Fatal ADRs accounted for 0.18% of all ADRs in our study, compared with 0.37% to 1.1% in other studies.5,7 The exclusion of preventable ADRs in our study might have contributed to the differences in the incidence of ADRs.
In our study, patients who experienced ADRs before hospital admission or who received medications during surgery were more likely to experience reactions rated high in severity. In addition, anticonvulsants and antineoplastic agents (eg, asparaginases) resulted commonly in serious or fatal ADRs. Furthermore, the time from initiation of drug administration to detection of ADR was longer for reactions with high severity. The duration of drug exposure, specifically to anticonvulsants or asparaginases, for the development of hypersensitivity or endocrine reactions has been shown to be delayed among children.33,34 On the basis of these findings, patients receiving these specific drug classes or receiving care in these settings require more-frequent and more-vigilant monitoring. These pharmacologic agents should be targeted for close evaluation, to detect ADRs earlier and to minimize drug exposure. Additional studies are necessary to examine the impact of ADRs in outpatient and surgical settings, because the affected sample was small.
ADRs with low severity, particularly those requiring discontinuation of administration of the suspected drug, were very common. Furthermore, antibiotics were associated frequently with mild reactions. These results corroborate the finding that ADRs experienced by patients with the admitting diagnosis of infectious diseases or fever were commonly low in severity. It is pertinent to continue to document these ADRs to prevent them in the future, either in the same patient or, with more research, in other patients.
Unique to this study is the assessment of the persons reporting the ADRs and the follow-up processes to address the ADRs. As discussed earlier, most ADRs were reported by pharmacists, and reporting by physicians and nurses should be targeted for improvement. With the exception of physician notification, most follow-up processes were not conducted or were not documented on the ADR forms. Independent of severity level, documentation in the patient's medical record (both the written medical chart and an allergy update in the hospital computer system) should be targeted for improvement. Follow-up education for individuals involved is not pertinent in situations in which the reaction is known and is not preventable (ie, pharmacologic reactions). However, it is still essential to educate the affected patients and parents, to allow them to recognize and, if possible, to avoid future use of the suspected drug.
Although it was not the focus of our study, the 8% incidence of preventable ADRs was lower than values reported in other studies, which ranged from 19% to 21% among children to 28% among adults.7,32,35 Since 1996, our hospital has used computerized order entry for all medications, and a clinical pharmacist participates in daily rounds in the hematology/oncology unit and NICU/PICU. All patients admitted to the NICU or PICU are monitored closely by clinical pharmacists, on a daily basis. In addition, our hospital has a drug information service and electronic resources (Poisondex, Micromedex, and Up-to-date) that are available for quick, point-of-care access by physicians, pharmacists, and nurses. These factors may reduce medication error rates, leading to the low incidence of preventable ADRs.36–39
There were several limitations to our study. First, the retrospective nature of the study and the long study period limited our ability to retrieve complete data for all ADRs. Some data were not retrievable from either the ADR forms or the electronic medical record. This resulted in different totals for some variables. Second, the changes in the format and content of the ADR reporting form in 1999 (to integrate preventability and follow-up processes) and again in 2001 (to integrate severity levels) might have contributed to the incompleteness of data collection. Furthermore, implementation of form changes requires users to understand the content and to adapt to the new format. This might have influenced the overall incidence of reporting. Lastly, because this study was based on experience from a single center, the results cannot be extrapolated to hospitals with different patient populations and ADR-reporting systems.
Measures to improve detection and reporting of ADRs, classified as either low or high in severity, by all health care professionals should be undertaken to enhance our understanding of the nature and impact of these ADRs in pediatrics. With this information, our management of ADRs might be improved and, more importantly, ADRs might be prevented by implementing strategies to target specific drugs that are commonly suspected. Follow-up mechanisms should be incorporated for all reactions, to ensure awareness of the ADRs among health care providers and individual patients. Ultimately, this should promote the safe use of drugs in the pediatric population.
- Accepted March 24, 2006.
- Address correspondence to Jennifer Le, PharmD, Department of Pharmacy Practice, College of Pharmacy, Western University of Health Sciences, 309 E Second St, Pomona, CA 91766-1854. E-mail:
This work was presented at the annual meeting of the American College of Clinical Pharmacy; October 23–26, 2005; San Francisco, CA.
The authors have indicated they have no financial relationships relevant to this article to disclose.
- ↵World Health Organization. Lexicon of alcohol and drug terms published by the World Health Organization. Available at: www.who.int/substance_abuse/terminology/who_lexicon/en. Accessed July 25, 2005
- ↵Gray J. Bill would force drug makers to give consumers data on risks. New York Times. July 25,1996;A11
- Mitchell AA, Goldman P, Shapiro S, Slone D. Drug utilization and reported adverse reactions in hospitalized children. Am J Epidemiol.1979;110 :196– 204
- ↵Moore TJ, Weiss SR, Kaplan S, Blaisdell CJ. Reported adverse drug events in infants and children under 2 years of age. Pediatrics.2002;110 (5). Available at: www.pediatrics.org/cgi/content/full/110/5/e53
- Mitchell AA, Lacouture PG, Sheehan JE, Kauffman RE, Shapiro S. Adverse drug reactions in children leading to hospital admission. Pediatrics.1988;82 :24– 29
- ↵Gustafson SR, ed. The Pediatric Patient. Philadelphia, PA: JB Lippincott; 1971:112– 115
- ↵McEvoy GK, Snow EK, Kester L, Litvak K, eds. AHFS Drug Information, 2005. Bethesda, MD: American Society of Health-System Pharmacists; 2005
- ↵Hartwig SC, Denger SD, Schneider PJ. Severity-indexed, incident report-based medication error-reporting program. Am J Health Syst Pharm.1991;48 :2611– 2616
- ↵Rawlins MD. Clinical pharmacology: adverse reactions to drugs. BMJ.1981;282 :974– 976
- ↵World Health Organization. International Drug Monitoring: The Role of the Hospital. Geneva, Switzerland: World Health Organization; 1966. Technical Report Series 425
- ↵Jha AK, Kuperman GJ, Teich JM, et al. Identifying adverse drug events: development of a computer-based monitor and comparison with chart review and stimulated voluntary report. J Am Med Inform Assoc.1998;5 :305– 314
- Clarkson A, Ingleby E, Choonara I, Bryan P, Arlett P. A novel scheme for the reporting of adverse drug reactions. Arch Dis Child.2001;84 :337– 339
- ↵Weiss J, Krebs S, Hoffmann C, et al. Survey of adverse drug reactions on a pediatric ward: a strategy for early and detailed detection. Pediatrics.2002;110 :254– 257
- ↵Bessmertny O, Hatton RC, Gonzalez-Peralta RP. Antiepileptic hypersensitivity syndrome in children. Ann Pharmacother.2001;35 :533– 538
- ↵King WJ, Paice N, Rangrej J, Forestell GJ, Swartz R. The effect of computerized physician order entry on medication errors and adverse drug events in pediatric inpatients. Pediatrics.2003;112 :506– 509
- Copyright © 2006 by the American Academy of Pediatrics