Prioritizing Strategies for Preventing Medication Errors and Adverse Drug Events in Pediatric Inpatients
Objectives. Medication errors in pediatric inpatients occur at similar rates as in adults but have 3 times the potential to cause harm. Error prevention strategies in this setting remain largely untested. The objective of this study was to classify the major types of medication errors in pediatric inpatients and to determine which strategies might most effectively prevent them.
Methods. A prospective cohort study was conducted of 1020 patients who were admitted to 2 academic medical centers during a 6-week period in April and May 1999. Medication errors were characterized by subtype. Physician raters evaluated error prevention strategies and identified those that might be most effective in preventing errors.
Results. Of 10 778 medication orders reviewed, 616 contained errors. Of these, 120 (19.5%) were classified as potentially harmful, including 115 potential adverse drug events (18.7%) and 5 preventable adverse drug events (0.8%). Most errors occurred at the ordering stage (74%) and involved errors in dosing (28%), route (18%), or frequency (9%). Three interventions might have prevented most potentially harmful errors: 1) computerized physician order entry with clinical decision support systems (76%); 2) ward-based clinical pharmacists (81%); and 3) improved communication among physicians, nurses, and pharmacists (86%). Interrater reliability of error prevention strategy assignment was good (agreement: 0.92; κ: 0.82).
Conclusions. Of the assessed interventions, computerized physician order entry with clinical decision support systems; ward-based clinical pharmacists; and improved communication among physicians, nurses, and pharmacists had the greatest potential to reduce medication errors in pediatric inpatients. Development, implementation, and assessment of such interventions in the pediatric inpatient setting are needed.
Iatrogenic injuries are common among hospitalized patients and are often preventable.1–3 A recent Institute of Medicine report estimated that 44 000 to 98 000 people die each year in the United States as a result at least in part of medical error.4 Although the accuracy of these figures is controversial,5–7 medical errors and the injuries to patients that can result are undeniably a major problem in current medical practice. In the Harvard Medical Practice Study, investigators reported that many adverse events experienced by patients in New York State in 1984 were complications of medication use.8,9 Shortly thereafter, the Adverse Drug Event Prevention Study confirmed that adverse drug events (ADEs) were common, costly, and often severe.1–3 Subsequent studies have further substantiated these findings.10–12
Although the problem of medication errors and ADEs has received substantial attention in adults, relatively few published reports have addressed this issue in children. In 1987, using a pharmacy-based review in 2 pediatric hospitals for 6 months, Folli et al13 identified 0.45 to 0.49 physician ordering errors per 100 medication orders. They found that pediatric patients younger than 2 years and pediatric intensive care unit (ICU) patients were particularly susceptible to errors, most of which were dosing errors. In a recent prospective, multicenter study of medication errors using comprehensive error detection methods in 2 academic pediatric hospitals, we found that errors occurred at a rate of 5.7 errors per 100 orders, with most of these errors occurring at the ordering stage (79%).14 Many of these involved incorrect dosing (34%).14 Importantly, although this overall error rate was similar to that found in a previous study of adult inpatients using similar methods,15 errors with the potential to cause harm were 3 times more likely to occur in pediatric inpatients compared with adults.14,15 Patients in the neonatal ICU were particularly susceptible.14,15
Children pose special challenges in the drug ordering and delivery process; for example, drug dosages often must be calculated individually, leading to increased opportunities for error with a relatively high risk of 10-fold errors.16–20 Furthermore, weights can change rapidly and dramatically over time, especially in small infants, requiring frequent dosing recalculations. Medicine dispensing in children is complicated by the fact that stock solutions of medicines are often available only at adult concentrations and must be diluted for use in children. Children, particularly those who are young and critically ill, may be more prone to ADEs than adults because they have less physiologic reserve with which to buffer errors such as overdoses. These issues highlight the importance of pediatric-specific prevention strategies for medication errors.
Several studies in adults have shown that medication errors and ADEs are often preventable.1–3,21–23 Two of the most effective interventions to date in adults have been computerized physician order entry (CPOE)22 and pharmacist participation in physician rounds in the ICU.23 Other information technologies that may be beneficial in preventing errors include computerized medication administration records, the use of robots in dispensing medications, bar coding of medications and patients, and “smart” intravenous devices.24,25 Few studies of medication error prevention strategies have been performed in pediatric inpatients, and research is needed to identify the most effective ways to reduce errors in this setting.13,26,27 Therefore, we performed a subanalysis of medication error data from our recent prospective cohort study of pediatric inpatients at 2 large, tertiary care hospitals. Our goals were 1) to categorize the major types of medication errors and 2) to identify the most effective intervention strategies for reducing medication error rates.
As previously described,14 the study was performed at 2 large, tertiary care academic medical centers. Hospital A is a free-standing pediatric hospital where study units included 2 randomly selected general medical wards, 1 randomly selected general surgical ward, a short-stay medical ward, and the pediatric medical/surgical ICU. Hospital B treats both adult and pediatric inpatients with a pediatric service that is independent of the adult service both geographically and administratively. At this hospital, all pediatric wards were included as study sites, which included the general medical/surgical wards, the pediatric medical/surgical ICU, and the neonatal ICU. Nine wards were studied in total, among which there were differences in case mix and staffing.
As previously described,14 medication errors were defined as errors in medication ordering, transcribing, dispensing, administering, or monitoring. For example, an order written for ondansetron without a route of administration was a medication error. A potential ADE was defined as any medication error with significant potential to harm a patient. An example of a potential ADE would be an order for dilaudid at 10 times the appropriate dose as a result of a calculation error. Such an event could be intercepted, meaning that the error did not reach the patient because someone recognized the problem before administration of the drug, or nonintercepted, meaning that the error reached the patient but caused no measurable harm. An ADE was any medication-related patient injury. ADEs could be either preventable or nonpreventable; an example of the latter would be a patient who develops Clostridium difficile colitis after appropriate antibiotic use. In this study, potentially harmful errors were defined as preventable ADEs and potential ADEs, regardless of whether they were intercepted. Nonpreventable ADEs were not included in these analyses because they are not associated with errors.
Sources for error detection included the medication order sheet, the medication administration record (MAR), and chart review of all patients on study wards during the study period. Data collected included medication name, dosage, route, and category. Additional data collected included stage in the medication delivery system at which the error occurred and the type of error. Data collectors also identified medication errors, potential ADEs, and ADEs by voluntary and verbally solicited reports from hospital staff, including house officers, nurses, and pharmacists. One data collector was assigned to each study ward on a given day. After completion of the study, administrative data were obtained on all patients hospitalized on the study wards, including age, gender, race, and payer. The study was performed over 6 weeks from April to May 1999, after Institutional Review Board approval was obtained at each hospital.
Case Review and Error Classification
Two physicians independently reviewed all medication errors and ADEs; these events were then classified as ADEs, potential ADEs, medication errors, or rule violations. The ADEs and potential ADEs were classified according to severity of injury or potential injury to the patient using a 4-point Likert scale. The ADEs were also rated according to their preventability using a 5-point Likert scale, and the likelihood the incident was attributable to the specific drug in question was determined using the Naranjo algorithm.28 All disagreements between reviewers were resolved by discussion and consensus.
Prevention Strategy Determination and Classification
As previously described,14 preventability assessment of all medication errors was performed by physician reviewers during initial data collection. However, at that time, fewer potential prevention strategies were considered than in the present study. We conducted a thorough review of the literature regarding previously successful error prevention strategies, most of which have been tried only in adults, and created an expanded list of potential error prevention strategies in pediatric inpatients. Of the studies that assessed preventability of ADEs in adults,1–3,21–23 2 of the most effective interventions have been CPOE22 and pharmacist participation in physician rounds in the ICU.23 Other potentially useful strategies include improved communication among health care providers, computerized MARs, the use of robots in dispensing medications, bar coding of medications and patients to prevent drug and patient identification errors, and “smart” intravenous pumps that can reliably perform dilutions.24,25 A total of 10 potential error prevention strategies were considered in this study: 1) basic CPOE, 2) CPOE with clinical decision support systems (CDSSs), including checks of drug ordering with regard to drug factors including dose, route, and frequency, and patient factors, including weight, allergies, renal function, age, and pregnancy status; 3) a clinical pharmacist on physician rounds or monitoring medication ordering, transcribing, and delivery; 4) changes in communication between health care providers, such as increasing nursing involvement during physician work rounds; 5) computerized MARs; 6) robots in drug dispensing; 7) “smart” intravenous devices for performing dilutions; 8) bar coding of drugs and patients; 9) automated bedside dispensing devices; and 10) unit-based dosing. Each erroneous order could have >1 prevention strategy assigned to it.
One physician reviewer (E.B.F.) assigned potentially effective prevention strategies to each error using preprinted forms. To assess interrater reliability in potential prevention strategy assignment, a second physician reviewer (R.K.) reabstracted 10% of the orders and assigned potential prevention strategies. Degree of rater agreement was then calculated using the κ statistic.
All errors were stratified according to the stage at which the error occurred, error type, and services involved in the error. Stages at which errors could occur include drug ordering, transcribing, pharmacy dispensing, administering, or monitoring. Ordering error subtypes included errors involving dosing, frequency, route, allergy, and missing information. Orders could have errors occurring at >1 stage and involving >1 type, although this was rare. Error subtypes and stages were analyzed for their correlation with potentially harmful versus nonharmful errors using univariate analyses SAS for Windows version 6.12 (SAS Institute, Cary, NC). Error prevention strategies were analyzed according to the proportion of total errors that might have been prevented, as well as the proportion of potentially harmful errors that might have been prevented. The unit of analysis was the medication order.
Medication Error Rates
During the study period, a total of 10 778 medication orders were written for the 1020 patients on study floors. Of these, 616 (5.7%) orders involved an error at 1 or more of the following stages of the medication delivery process: drug ordering, transcribing, dispensing, administering, or monitoring. There were 5 (0.8%) preventable ADEs, 115 (18.6%) potential ADEs, and 496 (80.5%) errors with relatively little potential for harm. A total of 320 (31%) patients experienced a medication error; 118 (12%) patients experienced 2 or more errors.
For the 616 medication errors, we performed detailed analyses of error characteristics (Table 1). The majority of errors occurred at the ordering stage (77.8%), followed by administering (12.8%) and transcribing (5.8%). The most frequent type of medication errors were dosing errors (28.4%). Of these, the majority involved failure to document the dose on the order (38.3%), missed doses (28.6%), and overdoses (22.9%). There were 6 10-fold dosing errors. The second most common error type was errors in drug route (17.7%); transcription errors were the third most common (15.9%). Frequency errors were another common error type (12.5%), of which the most common mistake was failure to document frequency on the order (61%). Potentially harmful errors were more likely to involve errors in dosing, allergies, or frequency than nonharmful errors (Table 1). For example, potentially life-threatening intercepted potential ADEs encountered in this study included an order for a heparin overdose and another for a digoxin overdose. Factors associated with lower rates of potentially harmful medication errors compared with nonharmful errors included errors occurring at the transcription stage, errors involving nursing, and missed doses.
Error Prevention Strategies
In this study, we analyzed 10 medication error prevention strategies for their potential effectiveness in reducing both overall error rates and potentially harmful error rates. The mean number of strategies assigned to each order was 4 (standard deviation: 2). In reducing overall medication error rates, 3 types of interventions had the greatest potential impact. First, the presence of clinical pharmacists to monitor ordering might have prevented 58.3% of errors, whereas pharmacists’ monitoring transcribing and administering might have prevented an additional 19.6% and 5.8% of errors, respectively (Table 2). In all, clinical pharmacists’ performing these functions could have resulted in a total medication error rate reduction of 81.3%. Second, computerized physician order entry could have prevented a large majority of all errors. Basic CPOE, which ensures legibility and completeness of orders but would not include decision support, had the potential to prevent 65.9% of all errors, whereas CPOE with decision support could have prevented an additional 6.8% of all errors for a total error rate reduction from CPOE of 72.7% (Table 2). Third, improved communication between health care practitioners had a significant impact in potential error prevention. Specifically, physician-pharmacist communication improvements might have prevented 47.4% of all medication errors; however, this impact largely overlaps with clinical pharmacist monitoring of ordering. In addition, physician-nurse communication improvements, such as increased nursing involvement in physician morning rounds, could have prevented 17.4% of all medication errors (Table 2). Together, these 3 interventions—ward-based clinical pharmacists; CPOE with CDSSs; and improved communication between physicians, pharmacists, and nurses—could potentially have prevented 98.5% of errors.
In addition, we performed subanalyses to identify which strategies might be most effective in preventing potentially harmful errors. Although all errors have some adverse consequences, prevention of those errors that pose the greatest risk of harm to patients is especially important. Again, we found that 3 interventions could have been most effective: 1) the presence of a clinical pharmacist to monitor medication ordering, transcribing, and administering (88.3% of errors); 2) CPOE with CDSSs (75.8%; Table 2, Fig 1); and 3) improved communication between physicians, nurses, and pharmacists, such as increased nursing participation in morning rounds (85.8%). Together, these 3 interventions might have prevented 96.7% of potentially harmful errors. Interventions such as unit-based dosing and pharmacists’ monitoring administering tended to prevent potentially harmful and nonharmful errors equally, whereas interventions such as computerization of the medication administration record and clinical pharmacist monitoring of transcribing tended to prevent nonharmful errors at a greater rate than potentially harmful errors (Table 2, Fig 1).
Finally, we analyzed whether different subtypes of potentially harmful medication errors were likely to be prevented by similar interventions. Of the 120 potentially harmful errors, 115 were potential ADEs, 68 of which were intercepted and 47 were not. The remaining 5 errors were preventable ADEs that actually caused harm to the patient. Compared with all other errors, the most effective potential strategies for preventing these 52 serious medication errors (nonintercepted potential ADEs and preventable ADEs) were CPOE with drug-allergy and drug-frequency checks, automated bedside dispensing devices, and enhanced communication between physicians and nurses. Of the 5 preventable ADEs, CPOE with drug-dose and drug-allergy checking would have prevented 4; the last preventable ADE was a drug administration error.
Medication errors and ADEs are serious problems in pediatrics as well as in adult medicine. The relatively higher rates of potentially harmful errors in hospitalized children compared with adults probably occurs primarily because dosing is more complex in pediatrics and underscores the need for safer systems in this setting.14 However, until recently, the incidence of pediatric medication errors has received relatively little scrutiny compared with adults, and even less has been done to assess their preventability. Errors with little potential to cause harm tend to be easiest to prevent21,22,29; however, prevention strategies that can prevent harmful errors would be expected to have more clinical impact.
In this study, we analyzed potential prevention strategies for medication errors in general as well as for potentially harmful errors. We found that 3 strategies had the greatest likelihood of reducing all errors as well as potentially harmful errors: CPOE with CDSSs; presence of a ward-based clinical pharmacist; and improved communication among physicians, nurses, and pharmacists. The first 2 of these strategies are demonstrated effective error prevention strategies in adult inpatients,21–23 and the present study strengthens the importance of these interventions. In addition, our findings suggest that improved communication among physicians, nurses, and pharmacists, perhaps through increased nursing involvement in physician work rounds, may be an effective error prevention strategy.
To date, relatively few studies of medication errors in children have focused on systems factors associated with errors.16,17,30–32 Although competence at the level of medical staff is essential in any clinical setting, medication errors are most effectively prevented by a systems approach rather than by singling out error-prone individual providers.1–3,21–23 Human factors research in general typically focuses on problems in systems rather than on individual blame.33–35 This approach reduces the likelihood of error and introduces checks into the system that can intercept errors before they reach the patient.
System Improvements 1: Clinical Pharmacists in Inpatient Pediatrics
Limited published data are available regarding the role of the clinical pharmacist in reducing medication errors in children, but reports to date have favored this intervention.13,36–38 In this study, we found that a clinical pharmacist’s monitoring ordering might have prevented 58% of all medication errors and 72% of potentially harmful errors. Pharmacists’ monitoring transcribing might have prevented an additional 20% of all errors and 13% of potentially harmful errors. Ward-based pharmacists may be very effective in medication-intensive areas such as ICUs. Although centralized pharmacist presence is important, increased pharmacist presence in the clinical setting such as on work rounds and with physicians as they write orders would likely have a greater impact in reducing errors by incorporating real-time feedback and teaching into the order writing process. Having pharmacists at the point of care may lead to more informed clinical decisions by physicians and greater interception of erroneous orders before they are implemented. In adults, clinical pharmacist presence at work rounds in the ICU has been demonstrated to be effective in reducing medication errors; in 1 study by Leape et al,23 this 1 intervention reduced preventable ordering ADEs in an adult ICU by 66%. Other studies have reported similar findings.39,40
System Improvements 2: CPOE in Inpatient Pediatrics
Computerization of medication ordering is a powerful intervention for improving drug safety because most errors occur at the ordering stage. Medication ordering in inpatient pediatrics poses many challenges: wide variation in patient weights necessitates individualized, weight-based dosage calculations; the smaller quantities of drugs required for younger children and infants necessitate working with smaller numbers; multiple calculations lead to an increased risk of 10-fold errors as a result of mistakes in decimal point placement; and medications are often not supplied in dosages appropriate for small children and must be diluted. Each of these issues increases the opportunity for error, and CPOE with CDSSs could effectively reduce the vast majority of these types of errors.
Basic CPOE can reduce the frequency of certain types of error by ensuring that orders are complete, legible, and in standard format. CDSSs add substantial value; physicians are much less likely to err when initially directed to an appropriate dose, route, or frequency, and computerized forcing functions can prevent potentially harmful errors. CPOE with CDSSs is somewhat more difficult to implement than basic CPOE, however, because of the need for software development, testing, and maintenance. This is especially true in the pediatric setting, where individualized dosage calculations, rapid fluctuations in weight and certain laboratory values, and changing physiologic parameters over the pediatric age spectrum require more sophisticated software. Although more challenging to develop, the computerization of ordering may be especially beneficial in pediatrics given these same issues.
In this study, basic CPOE could have prevented 66% of all errors and 60% of potentially harmful errors. CPOE with CDSSs could have prevented 73% of all errors and 76% of potentially harmful errors. In the adult setting, Bates et al29 demonstrated a 64% reduction in medication errors using a CPOE system with only basic decision support and an 83% reduction with more advanced decision support. In addition, a group from Latter-Day Saints Hospital in Utah has demonstrated that a computerized clinical decision support program significantly reduced antibiotic-associated medication errors and ADEs as well as costs.41 The superior projected effectiveness of CPOE with CDSSs compared with basic CPOE in preventing potentially harmful medication errors in this study (76% vs 60%, respectively) provides additional evidence in this regard.
In this study, we did not separately analyze the effectiveness of computerized pharmacist order entry. This is an important intervention already in place at many institutions and involves electronic entry of orders by pharmacists. Like CPOE, there is great variability among level of decision support in available systems. Theoretically, a pharmacist order entry system should be as effective as a physician order entry system with 1 important caveat. Physicians are more receptive to feedback at the time of decision making rather than after a decision has been made. Therefore, physicians will likely be more receptive to suggestions arising from decision support in CPOE systems rather than pharmacy systems. Ideally, a hospital would electronically link a CPOE system with a computerized pharmacy system, thereby deriving benefits from both and eliminating transcription.
System Improvements 3: Improved Provider Communication
The third strategy that we found most effective in potentially reducing medication error rates was improved communication among physicians, nurses, and pharmacists. Although previous studies have found that improved interactions between physicians and pharmacists can reduce error rates substantially in both adult and pediatric inpatient care,13,23 few have evaluated similarly enhancing formal interactions between physicians and nurses, although 1 study that included this as part of a compound intervention found no benefit.22 It may be that by increasing nursing staff participation in physician work rounds or in the order-writing process, error rates can be further reduced. In this study, improved communication between physicians and nurses might have prevented 17.4% of all errors and 29.2% of potentially harmful errors. Given the current structure of most inpatient pediatric medical settings, an intervention such as this might be relatively cost-effective and easy to implement.
Other information technology solutions for reducing medication errors that deserve mention include automated dispensing devices and robots in the pharmacy. Robots can recognize medications using bar codes to automate the prescription filling process; in 1 unpublished study, a robot decreased the dispensing error rate from 2.9% to 0.6% in an adult hospital.42 In addition, of the many errors that can occur with delivery of intravenous medications, especially in children, for whom dilutions of stock solutions are often needed, “smart” intravenous devices can reduce the chance of error through simplified programming and computerized checks. Such pumps are especially important for reducing the likelihood of 10-fold overdoses, a major problem in pediatrics.16,43 Finally, another potentially useful role of computers will likely be in coupling computerized MARs with CPOE. Such a computerized link can eliminate many transcription errors and allow for cumulative dose checking, which is particularly important for medications that are administered on a per-need basis. However, few data evaluating the impact of computerizing this process are available. In this study, computerization of the MAR would have prevented 29% of all errors and 19% of potentially harmful errors.
Two important caveats must be kept in mind in considering the findings in this study. First, the error detection method that we used is better at detecting prescribing errors than administration errors; many administration errors that were not detected undoubtedly occurred. If such administration errors were included, then they might have changed the relative effectiveness of each prevention strategy. Second, we were assessing potential preventability, not actual proportion prevented. Some strategies, particularly information technology-based strategies, would likely be more consistently effective in preventing errors than others. This study is also limited in that it was performed at only 2 institutions. Finally, the error prevention strategy categorization in this study treated each strategy as if it were monolithic; in fact, there is great variation. For example, clinical decision support systems vary widely among different CPOE systems.
Medication errors are an important problem in pediatrics, and effective strategies for preventing them are needed. In this study, we found that the vast majority of medication errors, including potentially harmful errors, may have been prevented by 3 interventions: 1) CPOE with CDSSs; 2) ward-based clinical pharmacists; and 3) improved communication among physicians, nurses, and pharmacists. Information technology interventions such as CPOE with CDSSs and organizational improvements such as including ward-based pharmacists on physician rounds are powerful tools that have already been proved to decrease medication errors in adults. The development and testing of pediatric-specific information technology, as well as the implementation of organizational changes in hospitals to increase communication among pharmacists, nurses, and physicians, are essential in reducing medication error rates and should be a top priority in the current practice of hospital-based pediatrics.
We thank John Orav, PhD, for statistical assistance. In addition, we thank Priscilla Dasse, RN, MPH; Alan Ezekowitz, MBCH, DPhil; Constance Crowley Ganzer, MS, RN; and Ronald Kleinman, MD, for support.
- Received March 21, 2002.
- Accepted September 27, 2002.
- Reprint requests to (R.K.) Division of General Internal Medicine, PBBA3, BWH, 75 Francis St, Boston MA 02115. E-mail:
Dr Bates is a coinventor on patent No. 6029138 held by Brigham and Women’s Hospital on the use of decision support software for medical management, licensed to the Medicalis Corporation. He holds a minority equity position in the privately held company Medicalis, which develops web-based decision support for radiology test ordering, and serves as a consultant to Medicalis.
- ↵To Err Is Human: Building a Safer Health System. Washington, DC: National Academy Press; 1999
- ↵Folli HL, Poole RL, Benitz WE, et al. Medication error prevention by clinical pharmacists in two children’s hospitals. Pediatrics.1987;79 :718– 722
- ↵Koren G, Barzilay Z, Greenwald M. Tenfold errors in administration of drug doses: a neglected iatrogenic disease in pediatrics. Pediatrics.1986;77 :848– 849
- ↵Leape LL, Lawthers AG, Brennan TA, et al. Preventing medical injury. Qual Rev Bull.1993;19 :144– 149
- ↵Bates DW. Using information technology to reduce rates of medication errors in hospitals. BMJ.2000;320 :788– 791
- ↵Naranjo CA, Busto O, Sellers EM, et al. A method for estimating the probability of adverse drug reactions. Clin Pharmacol Ther.1981;30 :238– 245
- ↵Rowe C, Koren T, Koren G. Errors by pediatric residents in calculating drug doses. Arch Dis Child.1998;79 :56– 58
- ↵Glauber J, Goldmann DA, Homer CJ, Berwick DM. Reducing medical error through systems improvement: the management of febrile infants. Pediatrics.2000;105 :1330– 1332
- Tisdale JE. Justifying a pediatric critical-care satellite pharmacy by medication error reporting. Am J Hosp Pharm.1986;43 :368– 371
- ↵Ellis MD. The clinical pharmacist on a pediatric unit. Hosp Top.1969;47 :81– 89
- ↵Kaushal R, Bates DW. Computerized Physician Order Entry with Clinical Decision Support Systems. In: Shojania KG, Duncan BW, McDonald KM, Wachter RM, eds. Making Health Care Safer: A Critical Analysis of Patient Safety Practices. 2001 (Evidence Report/Technology Assessment No. 43, AHRQ Publication No. 01-E058). Available at: http://www.ahrq.gov/clinic/ptsafety
- ↵Kaushal R, Bates DW. Clinical Pharmacists. In: Shojania KG, Duncan BW, McDonald KM, Wachter RM, eds. Making Health Care Safer: A Critical Analysis of Patient Safety Practices. 2001 (Evidence Report/Technology Assessment No. 43, AHRQ Publication No. 01-E058)
- ↵Weaver PE: Random sampling for quality assurance of the RxOBOT dispensing system. Available at: www.haas.berkeley.edu/citm/publications/briefings, mckcase.htm
- ↵Bates DW, Cousins DD, Flynn E, Gosbee JW, Richardson L, Schneider PJ. Consensus Development Conference Statement on the Safety of Intravenous Drug Delivery Systems: balancing safety and cost. Hosp Pharm.2000;35 :150– 155
- Copyright © 2003 by the American Academy of Pediatrics