OBJECTIVES. The purpose of this work was to characterize medication errors and adverse drug events intercepted by a system of pediatric clinical pharmacists and to determine whether the addition of a computerized physician order entry system would improve medication safety.
METHODS. The study included 16938 medication orders for 678 admissions to the pediatric units of a large academic community hospital. Pediatric clinical pharmacists reviewed medication orders and monitored subsequent medication use. Medication errors and adverse drug events were identified by daily review of documentation, voluntary reporting, and solicitation. Each potentially harmful medication error was judged whether or not it was intercepted and, if not, whether it would have been captured by a computerized physician order entry system.
RESULTS. Overall, 865 medication errors occurred, corresponding with a rate of 5.2 per 100 medication orders. A near-miss rate of 0.96% and a preventable adverse drug event rate of 0.09% were observed. Overall, 78% of potentially harmful prescribing errors were intercepted; however, none of the potentially harmful errors occurring at administration was intercepted and accounted for 50% of preventable adverse drug events. A computerized physician order entry system could capture additional potentially harmful prescribing and transcription errors (54%–73%) but not administration errors (0% vs 6%).
CONCLUSIONS. A system of pediatric clinical pharmacists effectively intercepted inpatient prescribing errors but did not capture potentially harmful medication administration errors. The addition of a computerized physician order entry system to pharmacists is unlikely to prevent administration errors, which pose the highest risk of patient injury.
- medication errors
- adverse drug events
- hospital pharmacy services
- computerized physician order entry
Errors that occur during the care of hospitalized patients are an important cause of iatrogenic morbidity and mortality.1,2 A report by the Institute of Medicine, To Err Is Human,3 and other public health research findings have directed public attention to medical errors and their impact on patients within our nation's health care system. From this literature, it is apparent that a significant percentage of medical errors are related to medication use and that many adverse drug events (ADEs) are preventable.
Most studies of medication errors and ADEs are limited to adult patients. Less is known about the epidemiology of medication errors and associated injury in pediatric patients. Early publications suggested that hospitalized infants and children are at higher risk for error-related ADEs because of a broad range of patient ages and sizes, calculation mistakes in drug dosing, limited physiologic reserve, and an inability to communicate with caregivers.4,5 More recent literature has substantiated these assumptions with larger epidemiological studies.6,7
Many believe that the use of information technology can capture medication errors and prevent subsequent ADEs. In particular, computerized physician order entry (CPOE) with clinical decision support has been shown to significantly reduce serious medication errors in the adult hospital setting.8,9 However, the few studies demonstrating the benefit of CPOE in the pediatric hospital setting are limited in methodology and scope.10–13
In many pediatric hospitals, clinical pharmacists with specialized training in pediatrics intercept errors that occur during the medication use process, especially potentially harmful pediatric prescribing errors.14 However, the medication errors that are not intercepted in this setting have not been studied. For hospitals with such a system of pediatric clinical pharmacists, the nonintercepted errors represent the potential risk for patient harm and would ideally be targeted by the institution of additional medication safety systems (eg, CPOE, bar-coding, and automated dispensing devices). In a recent study of pediatric medication errors and ADEs, it was estimated that pediatric ward-based pharmacists could have intercepted 94% of near misses, but this assumption has not yet been adequately tested.6 Therefore, the goal of this study was to quantify the harmful medication errors and ADEs that occur in a pediatric inpatient environment with a system of clinical pharmacists and to estimate the incremental safety benefit that a CPOE system may provide.
The clinical and medication records of consecutive pediatric admissions at Cedars-Sinai Medical Center, an urban, tertiary care academic community hospital, were prospectively reviewed for a 3-month period from February through April 2002, before any implementation of CPOE. During this period, the pediatric units contained a total of 84 beds, including a 31-bed pediatric ward, a 45-bed NICU, and an 8-bed PICU.
The definitions for medication errors and adverse events used in this study were adapted from previously published studies.6,9,15 In our study, a “medication error” was defined as an error in the process of medication delivery, including those occurring during prescribing, transcribing, dispensing, administering, or subsequent monitoring. We subclassified medication errors based on their potential for patient injury as either nonharmful or potentially harmful. “Nonharmful” described those medication errors with negligible potential to cause patient injury or adverse effects. Although such errors may seem “harmless,” they represent system faults that, under different circumstances, might result in patient injury. For example, an order for acetaminophen 3 times the recommended limit would not likely cause harm, but an aminoglycoside order prescribed at 3 times the maximum dose likely could. So, the example of a wrongly prescribed aminoglycoside order would be considered a potentially harmful medication error, depending on the outcome. In some cases, a potentially harmful error could have but did not result in patient injury because either the error was intercepted or the error was nonintercepted and reached the patient but fortuitously did not result in observable harm. We defined this type of medication error as a “near miss.” A near miss has often been referred to as a potential ADE in earlier literature.6,9,16 On the other hand, if this same aminoglycoside overdose was administered to the patient and resulted in renal injury, it would then be considered an ADE, defined as an unfavorable reaction or clinically significant toxicity resulting from medication use.17 In some cases, it can be a lack of prescribed drug use that results in patient injury, for example, an antibiotic that was ordered for neonatal sepsis but was mistakenly not administered and resulted in hypotension and renal failure. This definition of an ADE is distinct from the term “adverse drug reaction,” which has been defined by the World Health Organization as a “response to a medicine that is noxious and unintended, and that occurs at doses normally used in humans.”18 Therefore, ADE is a broader term that encompasses injury from dosing errors, as well as medication errors of both commission and omission.15,19–21 An ADE was considered preventable when it was attributable to ≥1 error; whereas a nonpreventable ADE was not error associated and, thus, unavoidable (eg, hives in a patient given an antibiotic without a previous known allergic reaction). In summary, both near misses and preventable ADEs were considered potentially harmful medication errors.22 A rule violation occurred when a medication order or associated documentation was technically erroneous but routinely interpreted correctly by clinical workers (eg, an order for diaper rash ointment twice a day that did not include a route). Rule violations had negligible impact on medication safety and were, thus, excluded from the analysis.
System of Clinical Pharmacists
In our institution, pediatric-trained clinical pharmacists are actively involved in the medication use process on the pediatric units. During and before the study period, the pharmacists worked from a satellite pharmacy area within the pediatric section of the hospital from 7:00 am to 11:30 pm 7 days a week. They reviewed medication orders for errors (ie, potential drug-drug and drug-allergy interactions, ambiguity, etc) before entry into the pharmacy information system. This pharmacy information system did not offer any automated drug-drug, drug-allergy, or dose-range decision support. After 11:30 pm, new medication orders were checked by the night pharmacy staff (not necessarily pediatric pharmacists) and dispensed. A pediatric pharmacist promptly double checked these night orders in the morning. In the ICUs, the clinical pharmacists participated in daily patient rounds with the physicians and nurses. They served as a resource for prescribing physicians, were involved in the education of the nursing staff regarding proper medication administration, and assisted in reconciling dispensed medications against the paper medication administration record.
Before data collection, the investigators formally introduced the study objectives and encouraged pediatric attending staff, house officers, medical students, nurses, and pharmacists on each unit to report any medication errors or suspected ADEs. Voluntary reports were made via an existing hospital-wide computerized reporting system (MIDAS+, Affiliated Computer Services, Dallas, TX), by confidential written forms deposited in an envelope posted on each unit, and by daily verbal solicitation of the nurse in charge of each unit on weekdays throughout the study period.
In addition to gathering reports of medication-related incidents, 2 data collectors performed a daily chart review of each admission including orders, notes, flowcharts, laboratory data, and medication administration records on weekdays. Chemotherapeutic agents were excluded from the study because of the lack of standard dosing practices for children on many experimental cancer treatment protocols, the high rate of expected adverse reactions, and the difficulty in differentiating whether a toxic adverse effect was error related or just a predictable reaction to chemotherapy. Data were collected on the following Monday for patients remaining in the hospital over the weekend.
During the study, the pharmacists agreed to annotate a brief description of each error intercepted and the corresponding corrective action on a copy of the original orders. This documentation was collected each morning by the data collectors and was entered for review by the study investigators.
Data collectors entered incident information using a client-server database application (Oracle Corporation, Redwood Shores, CA) on laptop computers with a wireless network connection immediately after the error or ADE was detected. Blinded interrater reliability was tested between the 2 data collectors on a random sample of 10% of the collected data.
Data collectors entered information on all of the suspected medication errors and ADEs. Any suspected near misses or ADEs were marked for independent review by 2 study physicians (Drs Wang and Herzog) who confirmed or reclassified the events as nonharmful medication errors, near misses, ADEs, rule violations, or no error. For all of the errors, the stage of process where the event occurred (ordering, transcribing, dispensing, administering, or monitoring) was recorded. Nonharmful medication errors and near misses were subclassified as intercepted versus nonintercepted. ADEs were classified as preventable versus nonpreventable and also rated on severity using a 5-point Likert scale. All of the discrepancies in classification or ratings between the physician experts were resolved by consensus, and blinded interrater agreement was recorded. This methodology was similar to that used in previous studies.6,9 The data collectors and physician experts at the study site were trained in this regard by Dr Kaushal.
Two physician reviewers (Drs Wang and Herzog) rated independently whether or not potentially harmful errors that were not captured by the system of clinical pharmacists (ie, nonintercepted near misses and preventable ADEs) could have been intercepted by an idealized CPOE system with advanced clinical decision support and error checking customized for pediatric use. The reviewers assumed that the hypothetical pediatric CPOE system was 100% sensitive in the following areas of automated decision-support: (1) drug-allergy contraindications; (2) drug-drug interactions; (3) drug dose error checking based on patient weight, age, indication, renal function, and single or cumulative dosing; (4) suggested calculated dosing based on patient parameters; (5) medication order checking for completeness and valid route, dose unit, frequency, and “as-needed” indications; (6) drug laboratory alerting; (7) critical laboratory alerting; (8) corollary orders; (9) therapeutic duplicate checking; (10) laboratory route contraindications (eg, rectal administration in a neutropenic patient); and (11) drug-diagnosis contraindications. This model was based on a published decision support taxonomy for CPOE23 and medication decision support functionality offered by a commercially available drug information engine (First Databank, San Bruno, CA).
The physician reviewers also assumed that, in the hypothetical CPOE system, all of the medication orders were coded, legible, and robustly interfaced to a pharmacy information system without requiring any nonautomated transmission. For this judgment, blinded interrater agreement was recorded for near misses and ADEs, with discrepancies resolved by consensus.
Using the observed number of orders, medication errors, and ADEs as criteria for agreement between data collectors, we measured the interrater reliability (κ statistic) to be 0.73 (95% confidence interval [CI]: 0.67–0.78) on a 10% random sample of patient days. In addition, the κ statistic between blinded physician reviewers was 0.89 (95% CI: 0.84–0.95) for incident classification, 0.88 (95% CI: 0.81–0.94) for incident severity, and 0.95 (95% CI: 0.91–0.99) for the possibility that the incidents could be intercepted by CPOE, corresponding with an excellent interrater agreement.
χ2 analysis and the Fisher's exact test for smaller samples were used to compare error interception rates by a system of pediatric clinical pharmacists with and without an idealized CPOE system across a number of medication error variables. SAS analytic software (SAS Institute Inc, Cary NC) was used for statistical analysis. The study protocol and analysis plan were approved by the institutional review board at Cedars-Sinai Medical Center (3672-02 and 3672-03).
A total of 16938 medication orders and actions related to their subsequent transcription, dispensing, administration, and monitoring were recorded. These medications were ordered across 5172 patient days, representing 678 admissions and 617 distinct patients. Thirty-two percent of medication orders were written for patients on the pediatric ward, whereas the NICU and PICU accounted for 46% and 22% of medication orders, respectively.
The rates of nonharmful medication errors, near misses, and ADEs are shown in Table 1. Across the entire study, a total of 865 errors were observed, with 687 (79%) classified as nonharmful medication errors, 162 (18%) as near misses, and 16 (1.8%) as preventable ADEs. We also observed 36 nonpreventable ADEs that, by definition, did not involve an error.
A majority of nonharmful medication errors (51%) and near misses (65%) originated from the ordering stage; even so, ordering errors accounted for a minority (31%) of preventable ADEs. Errors in transcription accounted for 15% of near misses, resulting in a single preventable ADE. Administration errors resulted in 15% of near misses, but these errors accounted for 50% of preventable ADEs. No near misses or preventable ADEs resulting from dispensing errors were observed (Table 2).
Nonharmful medication errors occurred most commonly with the use of antibiotics, intravenous fluids and electrolytes, nonopiate analgesics and antipyretics, and opiate agonists. These same classes of medications were also most frequently associated with near misses, although in a slightly different order. Medication errors were commonly associated with the intravenous route (47%), followed by medications ordered by the oral or enteral route (20%) and by inhalation (6.5%).
Physicians and related care providers (ie, medical students and nurse practitioners) were involved in the initiation or propagation of 46% of errors, whereas nurses were involved in 44%. Pharmacists and ward clerks were involved in a smaller percentage of medication errors (6.1% and 0.4%, respectively).
Medication Errors Intercepted With a System of Clinical Pharmacists
Forty-six percent of nonharmful medication errors and near misses were intercepted in this study with a system of clinical pharmacists in place (Fig 1). Near misses were intercepted more frequently than nonharmful medication errors (59% vs 44%; P < .001). Pediatric hospital pharmacists were responsible for 67% of interceptions, whereas nursing staff accounted for 24% of error interceptions. Physicians, respiratory therapists, and other members of the clinical health care team intercepted the remaining errors. The severity scores of intercepted and nonintercepted near misses were not significantly different (P = .65).
Potentially harmful errors, defined as both near misses and preventable ADEs, are shown in Table 3. The highest proportion (111 of 178 [62%]) of potentially harmful errors occurred at the ordering stage, but 78% of these were intercepted. As a result, 5 (31%) of 16 of ADEs were attributable to errors originating at the medication ordering stage. Conversely, potentially harmful errors originating at subsequent stages of medication use were significantly less likely to be intercepted (P < .001). Specifically, 18% (32 of 178) of the potentially harmful errors originated at the administration stage; however, none were intercepted and accounted for 50% of the preventable ADEs.
Different classes of medications were associated with significantly different rates of interception (P = .009). Although incidents involving diuretics, cathartics and laxatives, and benzodiazepines and barbiturates accounted for a minority (15%) of the potentially harmful errors, they were less likely to be intercepted than incidents involving other medication classes. Medication route was not associated with significantly different rates of interception (P = .95).
Preventable ADEs and near misses involving physicians were captured 76% of the time. Similarly, those involving pharmacists (67%) and other members of the health care team (56%) were intercepted a majority of the time. However, near misses and preventable ADEs involving nursing staff (12%) were less likely to be intercepted before reaching the patient (P < .001). Three nonintercepted near misses and no preventable ADEs involved ward clerks.
Potential Impact of an Idealized Pediatric CPOE System
We judged whether or not an idealized pediatric CPOE system with highly sensitive and comprehensive clinical decision support could intercept additional potentially harmful errors in an inpatient setting with pediatric clinical pharmacists (Table 3). Two physician reviewers judged that the addition of CPOE could increase the interception of near misses and preventable ADEs from 54% to 73% (P < .001).
Implementation of CPOE in addition to a system of pharmacists significantly improved the interception of both potentially harmful prescribing errors (78% vs 93%; P = .002) and transcription (24% vs 64%; P = .005) but not potentially harmful administration (0% vs 6%; P = .246) and monitoring errors (43% vs 71%; P = .296). No particular medication class was significantly associated with an increased error interception rate with the addition of CPOE, but errors associated with intravenously (53% vs 70%; P = .004) and oral/enterally (58% vs 81%; P = .036) administered medications would be intercepted more frequently. Potentially harmful errors initiated or perpetuated by physicians (76% vs 93%; P = .001) and nurses (12% vs 43%; P = .001) would more likely be intercepted with the use of a CPOE system added to a system of pediatric clinical pharmacists.
Our study adds a number of important findings to the existing literature on pediatric medication errors. First, we demonstrate the epidemiology of both prescribing and nonprescribing medication errors and ADEs in a pediatric hospital setting with a system of pediatric clinical pharmacists in place. Second, we show that although prescribing mistakes are the most frequent cause of harmful medication errors, most are intercepted by pharmacists before patient harm is realized. On the other hand, our study demonstrated that harmful medication administration errors were not prevented by a system of pediatric clinical pharmacists. Third, our data suggest that although the addition of a stringent CPOE system to pediatric clinical pharmacists can reduce medication errors overall, it does not directly fill the largest remaining safety gap: harmful medication administration errors.
A limited number of studies have systematically quantified both prescribing and nonprescribing medication errors and ADEs in the inpatient pediatric environment.6,7,24 A recent review of the primary literature surrounding pediatric medication errors noted a significant variation in published medication error rates, depending on the study methodology and definitions used.25 Kaushal et al6 found that the rate of near misses was 3 times higher in hospitalized infants and children at 2 Boston, MA, hospitals than in a historical adult population. This study was performed before the implementation of a system of pediatric clinical pharmacists, and in this setting 79% of near misses occurred at the ordering stage, whereas 4.3% occurred during the administration stage. Using a similar methodology, we observed similar overall rates of nonharmful medication errors (4.1% vs 5.7%), near misses (0.96% vs 1.1%), and preventable ADEs (0.09% vs 0.05%) but demonstrate that a majority of potentially harmful medication errors were intercepted by a system of pediatric clinical pharmacists.
In most hospitals caring for pediatric patients, nonpharmacist members of the clinical team also play an important role as “human systems” that prevent inpatient medication errors. While ordering medications, physicians routinely calculate drug doses and frequencies by considering differences in age, weight, and/or body surface area. Nurses often double check medication orders before administration but may not recalculate the doses uniformly. By using a system of pediatric-trained pharmacists to participate in and monitor important steps of medication use, hospitals benefit from an additional line of defense against iatrogenic injury caused by medication errors in children. Limited published data demonstrate the important role of pediatric pharmacists in preventing medication error-related patient injury.14 Even less is known about errors undetected by pharmacists or other members of the clinical team.
One study examined potentially harmful medication errors and predicted the potential efficacy of different error prevention strategies including pediatric pharmacists.22 The investigators predicted that a system of pediatric pharmacists could potentially intercept 81% of all harmful errors, including nonprescribing errors. In our study, we found that although a system of pediatric clinical pharmacists intercepted a majority (78%) of potentially harmful prescribing errors, none of the 32 potentially harmful administration errors were intercepted, resulting in one half of the preventable ADEs.
Therefore, strategic improvements in medication safety should target errors with the highest potential for morbidity and focus on the potential harmful errors that are unlikely to be averted by existing systems. Clinical information technology, including CPOE with advanced clinical decision support, has the potential to significantly impact prescribing errors in a mechanism similar to medication order review by clinical pharmacists. This assumption has been substantiated by recent reports of CPOE use in pediatrics, which have shown a reduction in prescribing errors.11,12 However, these studies did not examine near misses or ADEs occurring subsequent to medication ordering. Other studies have measured the impact of CPOE on both prescribing and nonprescribing errors but are limited by a retrospective methodology and reliance on voluntary reporting for case finding.10,13 Together, these studies build an evidence base that CPOE in the pediatric hospital setting can be a useful mechanism for reducing prescribing errors and sets an important foundation for further investigation.
However, the benefit of adding CPOE to a hospital setting with a system of pediatric clinical pharmacists is unknown, especially because both systems primarily target medication prescribing errors. Our data suggest that pediatric CPOE can incrementally improve the effectiveness of pharmacy staff by preventing additional potentially harmful errors, especially medication prescribing and transcribing errors. This is consistent with data in the adult literature that predict CPOE benefit when used in concert with a system of clinical pharmacists.26 Our findings echo the sentiment that the addition of CPOE to a system of clinical pharmacists in the pediatric hospital setting could help reduce the overall incidence of potentially harmful errors. But notably, the addition of CPOE to a system of pediatric pharmacists would not likely impact the largest group of nonintercepted potentially harmful medication errors, those that occur during the administration phase.
Organizations developing a strategy to improve pediatric medication safety involving the use of clinical information technology must, therefore, weigh the costs and benefits of implementation and consider the existing safeguards, both human and technical. Although CPOE in pediatrics is likely to offer organization benefits, significant barriers exist to the successful implementation of CPOE, including overall cost, physician acceptance, and the necessary redesign of patient care processes.27 In addition, the use of CPOE in pediatrics has historically been limited by a lack of available functionality and clinical decision support designed for weight- and age-based drug dosing, as well as pediatric intravenous fluids and total parenteral nutrition. Our data support the notion that a CPOE system enabled with decision support could further reduce errors associated with intravenous medication use, including errors associated with total parenteral nutrition and intravenous fluids.
Although some institutions have begun to demonstrate the use of CPOE systems customized for pediatric use, it is unclear whether hospitals with pediatric clinical pharmacists would best benefit from CPOE or other technology-based systems, such as an electronic medication administration records system or medication bar coding. These systems could directly target harmful administration errors and perhaps with lower cost or less risk than CPOE.
The present study has a number of limitations. The data were collected in a single hospital setting and may not be representative of other pediatric hospital environments, each with presumed differences in the milieu of medication safety systems, as well as the participation of clinical pharmacists. Even so, our study strongly suggests that a system of clinical pharmacists can significantly reduce medication-related injury, but a stronger conclusion is limited by its observational methodology and lack of a comparison group.
Our study examines the potentially harmful medication errors not intercepted by a system of pediatric health care workers, including clinical pharmacists, to predict the potential benefit of CPOE. However, a CPOE system could indirectly reduce medication administration errors by helping to avert medication errors at the point of prescription and allow clinical pharmacists to devote increased effort toward monitoring potentially harmful medication administration errors.
Finally, our estimations of the impact of a pediatric CPOE system were based on a hypothetical system containing maximally sensitive and full-featured decision support. Although such a stringent CPOE system would result in an intolerable level of false-positive alerts and a progressive decline in impact on physician behavior, our predictive analysis provides an estimate of the maximal incremental safety benefit of CPOE when added to a system of pediatric clinical pharmacists.
Optimal reduction of medication errors and ADEs in the pediatric inpatient setting requires a multifaceted approach, ideally integrating human safety systems with well-designed clinical information systems into hospital workflow. A system of pediatric clinical pharmacists is effective in capturing hospital-based prescribing errors, but with this system in place, the largest residual risk arises from administration errors. CPOE may support clinical pharmacists in incrementally improving medication safety, but when used in concert, should not be expected to reduce patient morbidity that arises during medication administration. In light of budgetary constraints and evolving evidence, hospitals providing care for infants and children must carefully consider which additional systems, whether human- or technology-based, would best improve pediatric medication safety.
This work was supported by a grant from the California HealthCare Foundation, a nonprofit philanthropic organization (based in Oakland, CA).
We thank John Wong for statistical support and M. Michael Shabot, MD, and Michael Langberg, MD, for their critical review of the article.
- Accepted August 9, 2006.
- Address correspondence to Jerome K. Wang, MD, Departments of Pediatrics, Medicine, and Enterprise Information Services, Cedars-Sinai Health System, 8700 Beverly Blvd, EIS, Steven Spielberg Building, 3rd Floor, Los Angeles, CA 90048. E-mail:
The authors have indicated they have no financial relationships relevant to this article to disclose.
- ↵Kohn LT, Corrigan J, Donaldson MS. To Err Is Human: Building a Safer Health System. Washington, DC: National Academy Press; 2000
- ↵Rowe C, Koren T, Koren G. Errors by paediatric residents in calculating drug doses. Arch Dis Child.1998;79 :56– 58
- ↵Koren G, Barzilay Z, Greenwald M. Tenfold errors in administration of drug doses: a neglected iatrogenic disease in pediatrics. Pediatrics.1986;77 :848– 849
- ↵Bates DW, Teich JM, Lee J, et al. The impact of computerized physician order entry on medication error prevention. J Am Med Inform Assoc.1999;6 :313– 321
- ↵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
- ↵Potts AL, Barr FE, Gregory DF, Wright L, Patel NR. Computerized physician order entry and medication errors in a pediatric critical care unit. Pediatrics.2004;113 :59– 63
- ↵Folli HL, Poole RL, Benitz WE, Russo JC. Medication error prevention by clinical pharmacists in two children's hospitals. Pediatrics.1987;79 :718– 722
- ↵American Pharmaceutical Association. Medication Errors. Washington, DC: American Pharmaceutical Association; 1999
- ↵World Health Organization. Requirements for Adverse Reaction Reporting. Geneva, Switzerland: World Health Organization; 1975
- 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
- ↵Fortescue EB, Kaushal R, Landrigan CP, et al. Prioritizing strategies for preventing medication errors and adverse drug events in pediatric inpatients. Pediatrics.2003;111 :722– 729
- ↵Wang JK, Shabot MM, Duncan RG, Polaschek JX, Jones DT. A clinical rules taxonomy for the implementation of a computerized physician order entry (CPOE) system. Proc AMIA Symp.2002;860– 863
- ↵Han YY, Carcillo JA, Venkataraman ST, et al. Unexpected increased mortality after implementation of a commercially sold computerized physician order entry system [published correction appears in Pediatrics. 2006;117:594]. Pediatrics.2005;116 :1506– 1512
- Copyright © 2007 by the American Academy of Pediatrics