Abstract
Background
One of the most critical steps in the medication process on pediatric wards is the medical prescription. This study aims to investigate the impact of a computerized physician order entry (CPOE) system on Adverse Drug Events (ADEs) and potentially harmful ADEs (pot ADEs) in comparison with paper‐based documentation in a general pediatric ward at a German University hospital.
Methods
A prospective pre–post study was conducted. All patients aged 17 years or younger were observed during the study periods (5 months pre‐ and postimplementation). Issues Regarding Medication (IRM) were identified by intensive chart review. Events were assessed regarding causality (WHO), severity (WHO; Dean & Barber for MEs), and preventability (Shumock) and classified into (pot) ADEs, (pot) Medication errors (ME), Adverse drug Reactions (ADR), and Other incidents (OI) accordingly.
Results
Total of 333 patients with medication were included in the paper‐based prescribing cohort (phase I) and 320 patients with medication in the electronic prescribing cohort (phase II). In each cohort, patients received a median number of four different drugs (IQR 5 and IQR 4). A total of 3966 IRM was observed. During the hospitalization, 2.7% (n = 9) patients in phase I and 2.8% (n = 9) in phase II experienced an ADE. Potentially harmful MEs were less often observed in the cohort with electronic prescribing (n = 228 vs. n = 562). The mean number per patient significantly decreased from 1.69 to 0.71 (p < .01).
Conclusion
The implementation of a CPOE system resulted in a reduction of issues regarding medication, particularly MEs with the potential to harm patients decreased significantly.
Keywords: adverse drug event, adverse drug reaction, computerized physician order entry, electronic prescribing, medication error, medication safety, pediatrics
Total and average (95% CI) Issues Regarding Medication (IRM). 1562 MEs + 29 OIs, 2228 MEs + 13 OIs, 35 ADRs +3 MEs + 1 OI. (pot)ADE, (potential)Adverse Drug Event; ADR, Adverse Drug Reaction; EnCR, Event not Clinically Relevant; IRM, Issue regarding medication; ME, Medication Error.
Abbreviations
- ADE
Adverse Drug Event
- ADR
Adverse Drug Reaction
- ATC
Anatomical Therapeutic Chemical (classification)
- CI
Confidence Interval
- CPOE
Computerized Physician Order Entry
- EnCR
Event not Clinically Relevant
- ICD‐10
International statistical Classification of Diseases and related health problems, 10th version
- IQR
Interquartile Range
- IRM
Issue Regarding Medication
- ME
Medication Error
- OI
Other Incident
- PEG/PEJ
Percutaneous Endoscopic Gastrostomy/ Percutaneous Endoscopic Jejunostomy
- potADE
potential Adverse Drug Event
- potME
potential Medication Error
- ROA
Route Of Administration
- SD
Standard Deviation
- SmPC
Summary of Product Characteristics
- SOP
Standard Operating Procedure
- WHO
World Health Organization
1. BACKGROUND
Adverse Drug Reactions (ADRs) and Medication Errors (MEs) are common and an important reason for the morbidity and mortality of hospitalized patients. 1 Critically ill patients but also newborns, infants, and children are particularly vulnerable. Medication Errors with the potential for harming patients were found to be three times as common in a pediatric hospital compared with a hospital for adults (29 versus 9.1 per 1000 treatment days). 2 , 3 , 4 Common off‐label use, missing evidence‐based dosing recommendations, and lack of age‐appropriate formulations contribute to this situation as well as weight‐based calculations. 7 , 8 , 9 , 10 , 11 , 12
Former studies show that ADRs are quite common in general pediatric wards. 5 , 6 A Study from 2008 showed an overall ADR incidence of 13.0% (95% CI 9.9%–16.6%) for inpatients at a German University Hospital, though events were mainly classified as mild (90.6%). 5
Further, literature shows that one of the most critical steps in the medication process in pediatric wards is medical prescribing. 2 , 3
Studies point out that the use of electronic systems may improve the quality of prescribing and reduces medication errors in pediatric inpatients. 13 , 14 , 15 , 16 , 17 , 18 The American Academy of Pediatrics recommends the implementation of a computerized physician order entry (CPOE) system with specific pediatric functionality, for example, weight‐based dosage calculations. 19
However, data from previously published studies are not necessarily applicable to different settings, countries, and software systems. To our knowledge, in Germany, there are no data reporting on the impact of CPOE on a general pediatric ward using a pre–post design in combination with intensive chart review, which is considered the gold standard to get a comprehensive picture of ADRs and MEs in hospitalized patients.
In the present study, we, therefore, investigated the impact of an electronic prescribing system on the incidence of ADRs and MEs at a Pediatric University Hospital in Germany.
2. METHODS
2.1. Study design
This prospective single‐center, before–after study was conducted in two 5‐month phases to reach the calculated sample size. One before (phase I, July–November 2012) and one after (phase II, May–September 2014) the implementation of a CPOE system on a 23‐bed general ward at a German children's hospital. Patients in the study ward were hospitalized for various diseases with a high proportion of patients with neuropaediatric diseases.
2.2. CPOE system
The CPOE system VMobil® (Advanova), which was implemented on the study ward, is an electronic patient record with CPOE. The user interface is based on the previously used paper‐based patient chart but provides additional features to support physicians throughout the prescribing process.
Age‐, weight‐, and body surface area‐based dosing recommendations were prepared for 70 frequently prescribed active substances and integrated into the system including different indications and routes of applications. As there was no national standard formulary at the time of conducting this study, dosing information was gathered from SmPCs, guidelines, and international databases, for example, the British National Formulary for Children, the Dutch Kinderformularium, the Pediatric & Neonatal Dosage Handbook in accordance with a Standard Operating Procedure (SOP), developed for this purpose. Available dosing recommendations for each active substance were compiled by pharmacists and stratified by indication, route of application, and age group. Before the data set was implemented into the CPOE system, all doses were checked by senior physicians.
2.3. Study population
Patients aged 17 years or younger at the time of admission, who were treated for at least 24 h at the study ward, were eligible for the present study under the condition that they were not transferred from another external or internal unit or readmitted to the study ward after inclusion.
Patients with complex diseases such as, for example, neoplasms that did not allow a structured monitoring of the drug therapy, with unobtainable clinical records or patients not recorded in the CPOE system after the implementation as well as patients that participated in a clinical trial less than 1 month ago were excluded.
2.4. Data collection
Demographic and other case‐specific data were collected for all admissions to the study ward. For patients who met the inclusion criteria, additional clinical data on diagnoses (ICD‐10), medication, and lab results were stored in a local Microsoft Access® database.
Medication data included the ATC code, route of administration (ROA), dosage, and frequency of medication. Each drug was included in the analysis only once per hospital admission independent of dosage and ROA.
2.5. Identification of issues regarding medication
Issues regarding medication (IRM) were identified by prospective intensive chart review (paper or electronic). Medication, laboratory data, and notes by physicians and nurses were screened daily by trained pharmacists of the study team as described in the study protocol and were performed throughout the entire hospital stay of the patient to get a full picture. In each phase, there was one lead pharmacist for the study (IT phase I, SW phase II). Additionally, the nursing staff was interviewed based on a standardized questionnaire by one of the pharmacists from Monday to Friday to clarify events and to get hints for potential errors not documented.
All events were assessed with regard to causality (WHO), severity (WHO; Dean & Barber for ME), and preventability (Shumock) 20 , 21 , 22 independently by at least two experienced reviewers. In case of discrepancies, consensus was reached by discussion after consultation with the attending doctor or the chief doctor.
2.6. Definitions
Definitions used in this study are summarized in Table 1.
TABLE 1.
Definitions.
| Issue regarding medication (IRM) | An event/circumstance or problem involving drug therapy that actually or potentially interferes with desired health outcomes, irrespective of its actual impact on the patient. |
|---|---|
| Adverse Drug Reaction (ADR) | A response to a drug that is noxious and unintended and occurs at doses normally used in man for prophylaxis, diagnosis, therapy of disease, or for modification of physiological function. 15 |
| Medication Error (ME) | Any preventable event that may cause or lead to inappropriate medication use or patient harm while the medication is in the control of the healthcare professional, patient, or consumer. 16 The medication process consists of drug prescription, drug transcription, drug dispensing, drug administration, and across settings. 17 A medication error includes also the prescribing of a medication without indication and prescriptions for active substances where no appropriate formulation is available for the stated dose. |
| Potential Medication Error (potME) | Medication errors with a considerable potential to harm (assessed with the severity of at least moderate (average score ≥3) according to Dean and Barber classification). |
| Other Incident (OI) | An Other Incident is any problem, circumstance, incident, or event occurring during drug therapy, which is not classified as ADR or ME (e.g., extravasation). |
| Adverse Drug Event (ADE) | Any harm occurring during drug therapy with at least a reasonable time relationship. ADEs include ADRs, any harm secondary to MEs or Other Incidents which are clinically relevant (outcome assessed with severity of at least moderate according to WHO classification). 18 |
| Potential Adverse Drug Event (potADE) | Any event occurring during drug therapy that had the potential to cause injury but failed to do so (by chance or due to interception). 22 PotADEs comprise MEs and Other Incidents of clinical relevance (assessed with severity of at least moderate (average score ≥3) according to Dean and Barber classification) not resulting in harm or resulting in harm of mild severity (outcome assessed according to WHO classification). |
| Event not Clinically Relevant (EnCR) |
ME or OI, which is judged to have only minor or no potential for injury (assessed with severity of mild or no potential effect (average score <3) according to Dean and Barber classification) 21 or ADRs of mild severity (according to WHO classification). |
Based on the above‐mentioned assessment, each issue was classified as follows:
Issues regarding medication were considered clinically relevant if their severity was equal to or higher than moderate according to (a) Dean and Barber for Medication Errors and Other Incidents (score ≥3.0) or (b) WHO for ADRs, respectively (Figure 1). All other issues regarding medication were classified as Events not Clinically Relevant (EnCR) and excluded from further analysis.
FIGURE 1.
Criteria for classification of events based on WHO 20 and Dean and Barber 21 severity assessment.
ADRs were considered to always cause harm whereas medication errors and other incidents were only considered to cause harm if the severity of the events was equal or higher than moderate according to the Dean and Barber classification scheme (Figure 1).
Clinically relevant issues regarding medication that caused harm according to our definition were considered Adverse Drug Events (ADEs).
Medication errors that were clinically relevant but did not cause harm were classified as potential MEs (potME). Potential ADEs (potADE) additionally included other clinically relevant incidents, for example, indicated drugs that were not available due to a shortage of supply.
2.7. Outcome measures
The primary outcome measure was the incidence of ADEs before and after the implementation of the electronic prescribing system (Figure 1).
Secondary outcome measures were
Incidence of potMEs and ADRs before and after the implementation of an electronic prescribing system.
Risk factors associated with ADEs, ADRs, and potMEs before and after the implementation of an electronic prescribing system.
2.8. Statistical analysis
The incidence of ADEs was defined as a number of patients with at least one ADE divided by the total number of patients receiving medication in the cohort. Additionally, the mean number of ADEs was calculated by dividing the total number of ADEs per cohort by the total number of patients with medication. Similar numbers were calculated for ADRs, potADEs, and potMEs.
Metric variables were reported as mean ± standard deviation (SD) and 95% CI (normally distributed) or median (interquartile range, IQR). Categorical variables were presented as frequency and percentage.
For the comparison of groups, the Mann–Whitney U test was used for metric variables, and the Chi‐squared test or Fisher's exact test was used for categorical variables. Logistic and linear regression were used to analyze the effect of the covariables number of drugs, duration of stay, the age group (infants and toddlers [<2 years], children [2–11 years], adolescents [12–17 years]), gender, and presence of renal insufficiency or a PEG/PEJ tube on incidence and mean number of events. P values of <.05 were considered statistically significant.
IBM SPSS® Statistics for Windows, Version 21.0.0.2 (IBM Corp.), was used to perform the statistical calculations.
2.9. Interrater reliability
In order to validate the detection of IRM in both cohorts, a systematic chart review was additionally performed by three raters independent of each other for 20 randomly chosen patients. Percentage of agreement was calculated for ADE and potADE. The total amount of potential events was defined as the number of events that were identified by at least one of the raters.
Fleiss‐Kappa was calculated for the detection of patients with at least one ADE or potADE. 23 , 24
2.10. Sample size calculation
Sample size was calculated based on the study results of Kunac et al. 25 This study used a similar methodology (intensive chart review, comparable ADE definition) and identified an ADE rate of 12.88% per 100 admissions.
Thus, considering a reduction in ADE rate by 50% (12 vs. 6 per 100 admissions) and a statistical power of 80% and alpha = 0.05, a sample size of 716 (2x358) was calculated.
2.11. Ethics Statement
The Ethics Committee of the Friedrich‐Alexander‐University approved the study protocol including one amendment (Application 110_12 B, 16.05.2012/04.09.2014).
Informed consent was not necessary according to Article 27 of the Bavarian Hospital Act.
3. RESULTS
3.1. Demographics
Overall, 1118 patients (phase I: 560, phase II: 558) were admitted to the ward during the two phases of the study. This corresponded to 684 (phase I) and 634 (phase II) admissions, respectively. Hence, 365 patients were included in the paper‐based prescribing cohort (phase I) and 376 patients in the CPOE cohort (phase II). Reasons for exclusion can be found in Figure 2.
FIGURE 2.
Study flowchart.
Approximately 91.2% (333 of 365) or 85.1% (320 of 376) of the included patients were prescribed and/or administered at least one medication. These patients at risk were included in further analysis.
Median age of patients with medication was 99.0 (IQR 147.0) and 74.5 (IQR 145.25) months, respectively (p = .052). In phase I, 166 of 365 (49.8%) patients were female. In the CPOE cohort, the corresponding number was 165 of 376 (51.6%).
Mean duration of stay in hours was slightly longer in phase II (93.8 95% CI 82.8–104.8) in comparison with phase I (78.5 95% CI 71.3–85.7, p = .088).
There were no statistically significant differences between phase I and II regarding age, gender, and duration of hospitalization (p > .05) (Table 2).
TABLE 2.
Patient characteristics (patients with at least 1 drug).
| Paper‐based (Phase I) N = 333 | CPOE (Phase II) N = 320 | p | |
|---|---|---|---|
| Gender | .662 a | ||
| N (%) | f 166 (49.8%) | f 165 (51.6%) | |
| m 167 (50.2%) | m 155 (48.4%) | ||
| Age (months) | .052 b | ||
| Mean (SD) | 100.90 (±72.07) | 90.25 (±73.11) | |
| Median (Q1–Q3) | 99 (25.5–172.5) | 74.5 (18–163.25) | |
| Duration of stay (hours) | .088 a | ||
| Mean (SD) | 78,5 (±66,7) | 93.8 (±100,1) | |
| Median (Q1–Q3) | 50.9 (41.8–96.5) | 59.0 (42.7–98.5) | |
| No. of drugs | .160 a | ||
| Mean (SD) | 5.20 (±3.75) | 4.74 (±3.46) | |
| Median (Q1–Q3) [min ‐ max] | 4 (2–7) [1 ‐24] | 4 (2–6) [1 – 26] | |
| Other risk factors | |||
| Renal insufficiency N (%) | 10 (3.0%) | 6 (1.9%) | .351 b |
| PEG/PEJ | 18 (5.4%) | 17 (5.3%) | .958 b |
Mann–Whitney U‐test.
Chi‐squared test.
In both cohorts, most patients were admitted to the hospital due to an emergency (80.8% vs. 81.3%).
Overall, 3262 drugs were recorded (phase I: n = 1730, phase II: n = 1517), including drugs prescribed to treat a chronic disease state. Mean numbers per patient were 5.2 ± 3.7 and 4.7 ± 3.5, respectively (p > .05). Maximum numbers of drugs per patient were 24 and 26, respectively.
Following IV solutions affecting the electrolyte balance (ATC B05BB), ibuprofen (n = 123 vs. n = 117 patients), metamizole (n = 97 vs. n = 83 patients), and paracetamol (n = 75 vs. n = 48 patients) were the drugs used most often. Next in frequency were rectal diazepam (n = 44 vs. n = 53 patients) and buccal midazolam (n = 43 vs. n = 28 patients) prescribed as on‐demand medication for patients with epileptic disorders.
3.2. Issues regarding medication
A total of 2756 (phase I) and 1210 (phase II) issues regarding medication were detected in the two study cohorts. Overall, 310 of 365 (93.1%) patients in phase I and 268 of 376 (83.8%) patients in phase II experienced at least one Issue (p < .01). The majority of Issues were classified as events without clinical relevance (EnCR): 2128 of 2756 (77.2%) or 931 of 1210 (76.9%), respectively (Figure 3).
FIGURE 3.
Total and average (95% CI) Issues Regarding Medication (IRM). 1562 MEs + 29 OIs, 2228 MEs + 13 OIs, 35 ADRs +3 MEs + 1 OI. (pot) ADE, (potential) Adverse Drug Event; ADR, Adverse Drug Reaction; EnCR, Event not Clinically Relevant; IRM, Issue regarding medication; ME, Medication Error.
3.3. Adverse drug events
A total of 37 ADEs (phase I) and 38 ADEs (phase II) were recorded (Figure 3). Among all ADEs, there were 24 ADRs and 13 MEs in phase I and 17 ADRs, 17 MEs, and 4 Other Incidents (OIs) in phase II, respectively.
Among the 20 ADEs, which occurred during hospitalization, five (two in phase I and three in phase II) originated from MEs (p = .68) (Table 3), whereas 11 (phase I) and 14 (phase II) out of the ADEs present prior to hospital admission were MEs (Figure 3). MEs with relevant harm during hospitalization were, for example, refusal of parents to start antiepileptic drugs, overlooked contraindications, underdosing, and incorrect application of oral drugs by the child's mother leading to insufficient plasma concentration.
TABLE 3.
Incidence and total numbers of Adverse Drug Events with clinically relevant harm in paper‐based and CPOE cohorts.
| Type of ADE | Incidence per 100 patients | Mean per patient | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Paper‐based (Phase I) | CPOE (Phase II) | Paper‐based (Phase I) | CPOE (Phase II) | |||||||||||
| % | N | % | N | p | Events | Mean | SD | 95%CI | Events | Mean | SD | 95%CI | p | |
| Adverse Drug Event | 9.9% | 33 | 11.3% | 36 | .578 a | 37 | 0.11 | 0.35 | 0.07–0.15 | 38 | 0.12 | 0.34 | 0.08–0.16 | .599 c |
| At admission | 7.2% | 24 | 9.1% | 29 | .385 a | 26 | 0.08 | 0.29 | 0.05–0.11 | 29 | 0.09 | 0.29 | 0.06–0.12 | .400 c |
| During hospitalization | 2.7% | 9 | 2.8% | 9 | .932 a | 11 | 0.03 | 0.21 | 0.01–0.06 | 9 | 0.03 | 0.17 | 0.01–0.05 | .942 c |
| Adverse Drug Reaction | 6.6% | 22 | 5.0% | 16 | .381 a | 37 | 0.11 | 0.35 | 0.07–0.15 | 17 | 0.05 | 0.24 | 0.03–0.08 | .378 c |
| At admission | 4.2% | 14 | 3.8% | 12 | .767 a | 15 | 0.05 | 0.22 | 0.02–0.07 | 12 | 0.04 | 0.19 | 0.02–0.06 | .761 c |
| During hospitalization | 2.7% | 8 | 1.6% | 5 | .442 a | 9 | 0.03 | 0.18 | 0.01–0.05 | 5 | 0.02 | 0.12 | 0.00–0.03 | .440 c |
| Medication Error with harm | 3.9% | 13 | 5.3% | 17 | .390 a | 13 | 0.04 | 0.19 | 0.02–0.06 | 17 | 0.05 | 0.23 | 0.03–0.08 | .390 c |
| At admission | 3.3% | 11 | 4.4% | 14 | .476 a | 11 | 0.03 | 0.18 | 0.01–0.05 | 14 | 0.04 | 0.20 | 0.02–0.07 | .476 c |
| During hospitalization | 0.6% | 2 | 0.9% | 3 | .681 b | 2 | 0.01 | 0.08 | 0.00–0.01 | 3 | 0.01 | 0.10 | 0.00–0.02 | .622 c |
| Other incidents | 0.0% | 0 | 1.3% | 4 | .041 a | 0 | 0.00 | 0.00 | 0.00–0.00 | 4 | 0.01 | 0.11 | 0.00–0.02 | .041 c |
| Potential Medication Error | 56.7% | 189 | 34.7% | 111 | .000 a | 562 | 1.69 | 2.39 | 1.43–1.94 | 228 | 0.71 | 1.52 | 0.55–0.88 | .000 c |
Chi‐squared test.
Fisher's exact test.
Mann–Whitney U‐test.
Four OIs were recorded in phase II and rated as clinically relevant (ADEs). These included three hospitalizations that were caused by drug abuse or suicide attempts. Another OI was an infant that did not swallow oral antibiotics most likely because of the taste. IV administration was necessary and led to a prolonged hospitalization.
Overall, 33/333 patients in phase I and 36/320 in phase II experienced at least one ADE resulting in a cumulative incidence of 9.9% and 11.3%, respectively. In both cohorts, the majority of those patients had an ADE present at the time they were admitted to the hospital (phase I: 73% [24/33]; phase II: 81% [29/36]). During hospitalization, 9 patients in each cohort experienced clinically relevant ADEs (RR 1.04, 95% CI 0.42–2.59). (Table 3).
After logistic regression analysis of the covariables, the primary outcome measure incidence of ADEs during hospitalization still showed no statistically significant differences between both cohorts (p = .993). The number of drugs and duration of stay were the only covariates with significance (p = .032 and p = .002).
ADRs occurred in 6.6% (phase I) and 5.0% (phase II) of patients, respectively, whereas MEs with relevant harm occurred in 3.9% (phase I) and 5.3% (phase II). (Table 3).
3.4. Potential adverse drug events
Totally 591 (phase I) and 241 (phase II) potADEs were recorded (Figure 3) including 562 potME (phase I) and 228 potME (phase II), respectively.
The incidence of MEs, which were potentially harmful but did not cause actual harm, was 56.7% (n = 189) in phase I and 34.7% (n = 111) in phase II (p < .001). This means that patients in phase I were significantly more affected by potential events than in phase II. The relative risk for potME was 0.6 (95% CI 0.5–0.7). The mean number of potMEs per patient significantly decreased from 1.69 to 0.71 (p < .01) post‐CPOE implementation.
A linear regression analysis with the same covariates as above was conducted for potentially harmful MEs per patient. After adjustment to these covariants, the number of potentially harmful MEs decreased highly significant after the introduction of the electronic prescribing system (p = .000). Number of drugs, duration of stay, renal insufficiency, and PEG/PEJ tube were covariates significantly related to a higher incidence of potADEs (all p < .005).
Number of drugs showed the highest impact on the number of potME in both regression models. Details can be found in Tables 4 and 5.
TABLE 4.
Logistic regression for Adverse Drug Events during hospitalization.
| Variate (reference) | ADE | ||
|---|---|---|---|
| B | SE | p | |
| Cohort (Phase I) | 0.005 | 0.548 | .993 |
| Number of drugs | −0.093 | 0.043 | .032 |
| Duration of stay in days | 0.125 | 0.041 | .002 |
| Renal insufficiency (no) | −0.476 | 1.264 | .706 |
| PEG/PEJ (no) | −1.880 | 1.042 | .167 |
| Gender (male) | 0.611 | 0.574 | .287 |
| Age group (infants and toddlers): | |||
| Children | 0.800 | 0.836 | .338 |
| Adolescents | 0.986 | 0.863 | .253 |
TABLE 5.
Linear regression for potential Medication Errors.
| potADE R 2 = 0.495 | ||||
|---|---|---|---|---|
| Variate (reference) | B | SE | Beta | p |
| Cohort (Phase I) | −0.882 | 0.117 | −0.214 | .000 |
| Number of drugs | 0.269 | 0.021 | 0.470 | .000 |
| Duration of stay in days | 0.004 | 0.001 | 0.155 | .000 |
| Renal insufficiency (no) | 1.569 | 0.431 | 0.118 | .000 |
| PEG / PEJ a (no) | 0.840 | 0.293 | 0.092 | .004 |
| Gender (male) | −0.122 | 0.118 | −0.030 | .303 |
| Age group (infants and toddlers): | ||||
| Children | 0.148 | 0.146 | 0.035 | .309 |
| Adolescents | 0.119 | 0.155 | 0.027 | .443 |
PEG/PEJ, percutaneous endoscopic gastrostomy tube/percutaneous endoscopic jejunostomy tube.
3.5. Characterization of events
The main cause of potentially harmful MEs in phase I (n = 301 of 562) was missing information in a prescription (e.g., missing dosage unit). The corresponding number in phase II was 40 of 228 only.
In the CPOE cohort, 12 of 228 (5.3%) potMEs resulted, at least in part, from software user errors.
Clinically relevant dosing errors decreased from 28 to 21 (p = .943).
In the paper‐based cohort, ibuprofen (n = 70), metamizole (n = 34), and furosemide (n = 21) were most often involved in potMEs. After the implementation of the CPOE system, valproic acid (n = 16), diazepam (n = 15), and ibuprofen (n = 11) were the three drugs that caused most potMEs.
Drugs with the highest rate of potMEs per prescription were furosemide (2.1) salbutamol (1.4), valproic acid (0.9) in phase I and valproic acid (0.8), furosemide (0.6), and levothyroxine (0.6) in phase II. However, these drugs were prescribed only 10 times (furosemide and salbutamol) and 16 times (valproic acid) in phase I and 21 times (valproic acid), 7 times (furosemide), and 14 times (levothyroxine) in phase II, respectively.
No deaths occurred in both study cohorts. One event in each cohort was assessed as life‐threatening.
3.6. Interrater reliability
Phase I: 80.0% (4/5) of ADEs were identified by all three raters; the fifth ADE was identified by two of the raters. Overall, 50 potential harmful events were found. 109 (72.7%) of the 150 (3x50) possible events were found by the three raters and the calculated Fleiss‐kappa for the detection of patients with at least one ADE or potADE was 0.72 (95% CI 0.47–0.97).
Phase II: Both ADEs were found by all three raters (100%). Overall, 22 potentially harmful events were found, and 55 (83.3%) of the 66 (3x22) possible events were found by three raters and the calculated Fleiss‐kappa for the detection of patients with at least one ADE or potADE was 0.91 (95% CI 0.66–1.17).
4. DISCUSSION
This was the first study in Germany investigating the impact of an electronic patient record with CPOE in a general pediatric ward setting on patient safety.
We found a significant impact particularly on potentially harmful MEs which confirms the positive effects of a CPOE system in pediatrics. As potential ADEs have a significant chance to cause harm to patients, the prevention of these MEs is important to avoid unnecessary harm to this vulnerable patient group. However, we could not prove the impact of the CPOE on ADEs (Adverse Drug Events with patient harm).
The number of ADEs in total and MEs with harm in particular during treatment on the ward was, in both cohorts, lower than reported in the studies that were used for the sample size calculation. 25 The study of Kunac et al. found 25.0 ADEs per 100 admissions whereof 14.3 were classified as preventable and therefore as MEs. The corresponding number (ME with harm during hospitalization) in our study was noticeably lower in both of our cohorts (0.6 and 0.9 per 100 admissions). This can be explained by the fact that Kunac et al. defined ADEs as “actual injuries resulting from medical interventions related to a medicine” regardless of the severity, whereas in our study, only events with at least moderate harm were defined as ADEs.
Holdsworth et al. investigated the impact of a CPOE system that was implemented in a whole children's hospital in the USA. ADEs per 100 admissions significantly declined from 6.3 to 3.1 (RR 0.64, 95%CI 0.43–0.95). 16 Based on the published data, the incidence of ADEs for the general pediatric unit was calculated at 5.1% before and 1.4% after the implementation of a CPOE system. The corresponding numbers in our study were 2.7% and 2.8% on a general pediatric ward which is considerably similar although no significant decrease for this type of event could be shown in our study.
Furthermore, the same study identified 7.9 vs. 2.9 potential ADEs per 100 admissions (0.37, 95% CI: 0.25–0.55) for the whole hospital. 16 The corresponding numbers for the general pediatric unit were 4.6% and 2.9% which was significantly lower than potADEs in our study (61.3% and 38.4%).
This huge difference is to be explained by varying definitions. The study by Holdsworth et al. defined potential ADEs as “an error that had the potential to result in at least a significant injury,” however, there was no definition for “significant difference” in this publication.
In our study, we defined that an event is considered to be relevant if according to Dean and Barber, the score is at least 3. This is the lower limit for a classification as moderate (range from ≥3 to ≤7) 21 and 98.1% of the potADEs in the AVOID study had a score between 3 and 5. If our cutoff had been different, the numbers would also be different. This underlines that only small differences in the definition of potential ADEs can have a strong impact on the comparability of study results. Nevertheless, independent of the definition, both studies showed a significant decrease in the number of potADEs which indicates that CPOE systems are useful in pediatric care.
The highest impact of the CPOE in our study was on the completeness of prescriptions, for example, dosage forms or dosage units. This was also shown in the study by Sin et al. 26
On the other hand, we did also see errors that were caused by the use of the CPOE. This strongly indicates the importance of permanent training of healthcare professionals, especially new colleagues. 26 , 27
Interestingly, a large proportion of our IRM were MEs. This type of event is generally preventable which means that, if appropriate measures are being implemented, there is a great potential to improve the safety of drug therapy in this population.
4.1. Strengths and limitations
The prospective study design in combination with intensive chart review and daily interviews was well suited for the detection of ADRs and MEs that occurred during prescribing and documentation of medication. The large number of events identified supports the comprehensiveness of data collection. By applying different definitions regarding clinical relevance and severity, the impact of the CPOE system on different types of events could be identified.
Errors during drug administration or preparation were only detected if mentioned in the documentation or during interviews. Nevertheless, the expected impact of VMobil® on these MEs was very low.
The major limitation of this study is probably its sample size. Our primary endpoint was the incidence of clinically relevant events, that is, which caused real harm to the patients. The incidence of these events was much lower in our study than in the study which was used for the sample size calculation despite apparently similar definitions. This underlines the importance of clear definitions in this research field, ideally by applying established tools and scores.
Furthermore, an additional control ward without intervention would have strengthened our results. This was, however, not possible because VMobil® was already implemented at all other general pediatric wards in the hospital prior to the AVOID study and a comparable ward with similar patients and diseases was not available.
Due to the unblinded study design, unintentional bias regarding the detection and rating of events cannot be excluded completely. Nevertheless, the interrater reliability showed substantial/(almost) perfect agreement. 28
Nevertheless, as there were no other major modifications to the medication process during both time periods and both study cohorts showed no significant differences in the study population, the effects were most likely related to the implementation of the CPOE system.
5. CONCLUSION
Our study showed that the implementation of an electronic prescribing system at a general pediatric ward in Germany can significantly reduce the number of MEs with the potential to harm patients. Studies in larger pediatric cohorts are necessary to investigate the impact on the frequency and incidence of ADEs with clinically relevant patient harm.
More attention should be paid to clear definitions and scores to make results comparable among studies.
AUTHOR CONTRIBUTIONS
SW, IT, and AN made substantial contributions to the conception and design of the study. The data were acquired by SW, IT, and SB. Adverse events and medication errors were rated by SW, IT, AN, WR, and RT. The analysis and interpretation of the data were performed by SW, IT, and AN. SW, IT, and AN wrote the draft of the article. WR, RT, and SB critically revised the content of the article. The final version was approved by all authors.
CONFLICT OF INTEREST STATEMENT
All authors declare that they have no conflict of interest.
FUNDING INFORMATION
This study received funding from the ELAN Fonds of the Friedrich‐Alexander Universität Erlangen‐Nürnberg. We acknowledge financial support by Deutsche Forschungsgemeinschaft and Friedrich‐Alexander‐Universität Erlangen‐Nürnberg within the funding program “Open Access Publication Funding.”
ETHICS STATEMENT
This study was approved by the local Ethics Committee of the Friedrich‐Alexander Universität Erlangen‐Nürnberg.
ACKNOWLEDGMENTs
Open Access funding enabled and organized by Projekt DEAL.
Wimmer S, Toni I, Botzenhardt S, Trollmann R, Rascher W, Neubert A. Impact of a computerized physician order entry system on medication safety in pediatrics—The AVOID study. Pharmacol Res Perspect. 2023;11:e01092. doi: 10.1002/prp2.1092
Stefan Wimmer and Irmgard Toni have contributed equally.
DATA AVAILABILITY STATEMENT
The data sets used and/or analyzed during the current study are available from the corresponding author upon reasonable request.
REFERENCES
- 1. Bates DW, Cullen DJ, Laird N, et al. Incidence of adverse drug events and potential adverse drug events. Implications for Prevention. ADE Prevention Study Group. JAMA. 1995;274(1):29‐34. [PubMed] [Google Scholar]
- 2. Fortescue EB, Kaushal R, Landrigan CP, et al. Prioritizing strategies for preventing medication errors and adverse drug events in pediatric inpatients. Pediatrics. 2003;111(4 Pt 1):722‐729. [DOI] [PubMed] [Google Scholar]
- 3. Kaushal R, Bates DW, Landrigan C, et al. Medication errors and adverse drug events in pediatric inpatients. JAMA. 2001;285(16):2114‐2120. [DOI] [PubMed] [Google Scholar]
- 4. Bates DW, Boyle DL, Vander Vliet MB, Schneider J, Leape L. Relationship between medication errors and adverse drug events. J Gen Intern Med. 1995;10(4):199‐205. [DOI] [PubMed] [Google Scholar]
- 5. Oehme AK, Rashed AN, Hefele B, Wong IC, Rascher W, Neubert A. Adverse drug reactions in hospitalised children in Germany are decreasing: results of a nine year cohort‐based comparison. PloS One. 2012;7(9):e44349. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Rashed AN, Wong IC, Cranswick N, et al. Adverse drug reactions in children ‐ international surveillance and evaluation (ADVISE): a multicentre cohort study. Drug Saf. 2012;35(6):481‐494. [DOI] [PubMed] [Google Scholar]
- 7. Smyth RL, Peak M, Turner MA, et al. ADRIC: adverse drug reactions In children—a programme of research using mixed methods. Programme Grants Appl Res. 2014;2(3):1‐184. [PubMed] [Google Scholar]
- 8. Gallagher RM, Mason JR, Bird KA, et al. Adverse drug reactions causing admission to a paediatric hospital. PloS One. 2012;7(12):e50127. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Standing JF, Tuleu C. Paediatric formulations ‐ getting to the heart of the problem. Int J Pharm. 2005;300(1–2):56‐66. [DOI] [PubMed] [Google Scholar]
- 10. Doherty C, Mc Donnell C. Tenfold medication errors: 5 years' experience at a university‐affiliated pediatric hospital. Pediatrics. 2012;129(5):916‐924. [DOI] [PubMed] [Google Scholar]
- 11. Crouch BI, Caravati EM, Moltz E. Tenfold therapeutic dosing errors in young children reported to U.S. poison control centers. Am J Health Syst Pharm. 2009;66(14):1292‐1296. [DOI] [PubMed] [Google Scholar]
- 12. Chappell K, Newman C. Potential tenfold drug overdoses on a neonatal unit. Arch Dis Child Fetal Neonatal Ed. 2004;89(6):F483‐F484. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Maaskant JM, Vermeulen H, Apampa B, et al. Interventions for reducing medication errors in children in hospital. Cochrane Database Syst Rev. 2015;3:Cd006208. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Walsh KE, Landrigan CP, Adams WG, et al. Effect of computer order entry on prevention of serious medication errors in hospitalized children. Pediatrics. 2008;121(3):e421‐e427. [DOI] [PubMed] [Google Scholar]
- 15. 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(3 Pt 1):506‐509. [DOI] [PubMed] [Google Scholar]
- 16. Holdsworth MT, Fichtl RE, Raisch DW, et al. Impact of computerized prescriber order entry on the incidence of adverse drug events in pediatric inpatients. Pediatrics. 2007;120(5):1058‐1066. [DOI] [PubMed] [Google Scholar]
- 17. Rinke ML, Bundy DG, Velasquez CA, et al. Interventions to reduce pediatric medication errors: a systematic review. Pediatrics. 2014;134(2):338‐360. [DOI] [PubMed] [Google Scholar]
- 18. Ehlers B. Untersuchungen über Empfehlungen zu Funktionen von elektronischen Verordnungssystemen für KinderDosierung.de. Dissertation Heinrich‐Heine‐Universität Düsseldorf. 2014.
- 19. American Academy of Pediatrics . Electronic prescribing in pediatrics: toward safer and more effective medication management. Pediatrics. 2013;131(4):824‐826. [DOI] [PubMed] [Google Scholar]
- 20. Uppsala Monitoring Centre . The use of the WHO‐UMC system for standardised case causality assessment. https://www.who.int/medicines/areas/quality_safety/safety_efficacy/WHOcausality_assessment.pdf. Accessed Mai 15, 2019
- 21. Dean BS, Barber ND. A validated, reliable method of scoring the severity of medication errors. Am J Health Syst Pharm. 1999;56(1):57‐62. [DOI] [PubMed] [Google Scholar]
- 22. Schumock GT, Thornton JP. Focusing on the preventability of adverse drug reactions. Hosp Pharm. 1992;27(6):538. [PubMed] [Google Scholar]
- 23. Fleiss JL. Measuring nominal scale agreement among many raters. Psychol Bull. 1971;76(5):378‐382. [Google Scholar]
- 24. Fleiss JL, Nee JC, Landis JR. Large sample variance of kappa in the case of different sets of raters. Psychol Bull. 1979;86(5):974‐977. [Google Scholar]
- 25. Kunac DL, Kennedy J, Austin N, Reith D. Incidence, preventability, and impact of adverse drug events (ADEs) and potential ADEs in hospitalized children in New Zealand: a prospective observational cohort study. Paediatr Drugs. 2009;11(2):153‐160. [DOI] [PubMed] [Google Scholar]
- 26. Sin CM, Young MW, Lo CCH, Ma PK, Chiu WK. The impact of computerised physician order entry on prescribing in general paediatric units in Hong Kong. Int J Pharm Pract. 2021;29(2):164‐169. [DOI] [PubMed] [Google Scholar]
- 27. Han YY, Carcillo JA, Venkataraman ST, et al. Unexpected increased mortality after implementation of a commercially sold computerized physician order entry system. Pediatrics. 2005;116(6):1506‐1512. [DOI] [PubMed] [Google Scholar]
- 28. Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics. 1977;33(1):159‐174. [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
The data sets used and/or analyzed during the current study are available from the corresponding author upon reasonable request.