https://orcid.org/0000-0002-2447-7220
Figures
Abstract
Despite efforts in improving medication safety, medication administration errors are still common, resulting in significant clinical and economic impact. Studies conducted using a valid and reliable tool to assess clinical impact are lacking, and to the best of our knowledge, studies evaluating the economic impact of medication administration errors among neonates are not yet available. Therefore, this study aimed to determine the potential clinical and economic impact of medication administration errors in neonatal intensive care units and identify the factors associated with these errors. A national level, multi centre, prospective direct observational study was conducted in the neonatal intensive care units of five Malaysian public hospitals. The nurses preparing and administering the medications were directly observed. After the data were collected, two clinical pharmacists conducted independent assessments to identify errors. An expert panel of healthcare professionals assessed each medication administration error for its potential clinical and economic outcome. A validated visual analogue scale was used to ascertain the potential clinical outcome. The mean severity index for each error was subsequently calculated. The potential economic impact of each error was determined by averaging each expert’s input. Multinomial logistic regression and multiple linear regression were used to identify factors associated with the severity and cost of the errors, respectively. A total of 1,018 out of 1,288 (79.0%) errors were found to be potentially moderate in severity, while only 30 (2.3%) were found to be potentially severe. The potential economic impact was estimated at USD 27,452.10. Factors significantly associated with severe medication administration errors were the medications administered intravenously, the presence of high-alert medications, unavailability of a protocol, and younger neonates. Moreover, factors significantly associated with moderately severe errors were intravenous medication administration, younger neonates, and an increased number of medications administered. In the multiple linear regression analysis, the independent variables found to be significantly associated with cost were the intravenous route of administration and the use of high-alert medications. In conclusion, medication administration errors were judged to be mainly moderate in severity costing USD 14.04 (2.22–22.53) per error. This study revealed important insights and highlights the need to implement effective error reducing strategies to improve patient safety among neonates in the neonatal intensive care unit.
Citation: Henry Basil J, Mohd Tahir NA, Menon Premakumar C, Mhd Ali A, Seman Z, Ishak S, et al. (2024) Clinical and economic impact of medication administration errors among neonates in neonatal intensive care units. PLoS ONE 19(7): e0305538. https://doi.org/10.1371/journal.pone.0305538
Editor: Zohour Ibrahim Mahmoud Rashwan, University of Bahrain, EGYPT
Received: January 11, 2024; Accepted: June 1, 2024; Published: July 11, 2024
Copyright: © 2024 Henry Basil et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All relevant de-identified data are within the manuscript and its Supporting Information files.
Funding: This work was supported by the Fundamental Research Grants Scheme by the Ministry of Higher Education (MOHE), Malaysia [FRGS/1/2022/SKK16/UK/02/7]. Sources of funding and funders had no impact on the design, conduct or results of the study. There was no additional external funding received for this study.
Competing interests: The authors have declared that no competing interests exist.
Introduction
Globally, medication errors (MEs) are the leading cause of preventable patient harm [1]. They occur at various stages of the medication-use process, such as prescribing, transcribing and documenting, dispensing, administering, and monitoring [2]. Throughout these stages, errors were most common during administration. Among the 237 million MEs estimated to occur in England annually, the error rate during administration was found to be 54.4% [3]. In a systematic review of MEs in Southeast Asian countries, medication administration errors (MAEs) were also found to be the most common type of ME [4]. Furthermore, the likelihood of MAEs being intercepted was less likely than that of MEs being intercepted during other stages of the medication use process [5]. The fact that MAEs are the most common type of ME and are the least likely to be intercepted means that MAEs may cause more harm to patients and place a significant burden on the healthcare system [3, 6].
Although a plethora of studies have been conducted to investigate MEs, MEs remain a concern, especially among neonates [7]. One of the three key action areas identified by the World Health Organisation in its Third Global Patient Safety Challenge, which aims to globally reduce severe preventable harm as a result of MEs by 50% over a span of 5 years, is high-risk settings such as the Neonatal Intensive Care Unit (NICU) [1, 2]. In the neonatal population, MEs occur at different stages of the medication-use process, with percentages ranging from 14% to 74% for prescribing, 12% to 18.4% for transcription, 11.9% to 25% for dispensing, 31% to 63% for administering, and 1.4% for monitoring [5]. The prevalence of MAEs among neonates in the NICU has been reported to be as high as 94.9% [8]. Given that MAEs are so prevalent in the NICU, it is not surprising that the World Health Organisation has identified such high-risk situations as areas requiring much attention for the reduction of significant harm [2].
A study in the United Kingdom that evaluated almost 60,000 medication incidents reported to the National Patient Safety Agency found that MAEs led to the highest number of medication incidents resulting in severe patient harm or death [9]. In another study, it was estimated that in a hospital where 6 million doses of medications are administered annually, 4,000 patients are expected to be harmed [10]. In a recent systematic review conducted to examine the severity of MAEs among neonates, only three studies were found to have assessed severity using different assessment methods [8]. The challenge in classifying the severity of MAEs is reflected by the limited number of studies [11]. The assessment methods used for these studies were the National Coordinating Council for Medication Error Reporting and Prevention [12], a validated visual analogue scale [13] and a medication error review [14]. MAEs were harmful in 1% to 34.5% of the medication doses observed [15–17]. However, only one of the three studies used an appropriate validated assessment tool to classify potential harm as a result of MAEs in the Netherlands [15]. While the literature on the severity of MAEs among neonates is available [15–17], a valid, reliable, and appropriate tool to assess the severity of potential harm in a resource-restricted setting with a larger sample size will facilitate a better understanding of the clinical impact of MAEs.
From an economic standpoint, the World Health Organisation reported that MEs are estimated to cost USD 42 billion annually worldwide [1]. The cost of these preventable adverse drug events was estimated to range between USD 25 and 33 million in a study where 6 million doses of medications were administered annually [10]. In a retrospective review of claims made against the National Health Service in the United Kingdom over a span of 12 years by adult patients alleging that they have suffered from anaesthesia-related MAEs, MAEs were estimated to cost £4 283 677 [18]. To our knowledge, there is also no available literature on the economic impact of MAEs among neonates. Therefore, understanding the clinical and economic impact of MAEs on neonates in the NICU will enable the implementation of targeted interventions to reduce MAEs and highlight specific areas requiring further studies to reduce MAEs for policymakers.
Therefore, the purpose of this study was to determine the potential clinical and economic outcomes of identified MAEs and to identify the factors associated with them. The information obtained from this study will shed light on the consequences of MAEs among neonates in NICUs, enable the implementation of targeted interventions to reduce MAEs and highlight specific areas requiring further studies to reduce MAEs for policymakers.
Materials and methods
Design
A prospective direct observational study was conducted as a preliminary study to identify MAEs. A multidisciplinary expert panel consisting of three neonatologists, a paediatrician, a clinical pharmacist, and a senior nurse specialising in neonatology assessed each MAE for its clinical impact.
Study setting
The national level, multi centre, direct observational study was conducted within the NICUs of five public hospitals affiliated with the Ministry of Health, Malaysia. Each hospital was purposefully selected from one of the five regions in Malaysia—Northern, Central, Southern, East Coast, and East Malaysia—to ensure representation of the two principal categories of public hospitals specialising in neonatology: major specialist hospitals and state hospitals. The NICUs in the selected hospitals were characterised by bed capacities ranging from 16 to 38. While variations existed in the number of nurses and patients across these hospitals, the duration of nurses’ duty hours was the same throughout all the shifts and hospitals. The general characteristics of the study sites are shown in Table 1.
Study sampling
Using a formula without finite population correction, a minimum sample size of 329 was required to achieve a 95% confidence level with a 5% margin of error [19]. The expected prevalence of MAEs in this study was based on the findings of another study with a similar social background and setting, i.e., 31% [20]. With an additional 20% dropout rate, the sample size required would be 395 drug administrations.
Eligibility criteria
Medications prepared and administered by nurses via all routes were included in this study except:
- Enteral feedings, parenteral nutrition, blood-derived products, medical gases and dietary supplement.
- Medications not administered because the patient was absent during the medication rounds, lacked intravenous access or due to clinical reasons such as contraindications.
- Rectal administrations, when neonatal-specific rectal dosage forms were unavailable, and paediatric rectal dosage forms were modified to lower doses.
Data collection
The data were collected at these NICUs between April 2022 and March 2023. In this study, direct observations were conducted by two experienced clinical pharmacists, with a minimum of ten years of clinical experience, acting as observers. Before initiating observations, the observers received comprehensive training in the direct observation method of data collection, as described by Barker and McConnell [21]. Pilot observations were subsequently conducted to familiarise the observers with ward procedures and minimise Hawthorne effects. To further minimise the Hawthorne effect, nurses were informed that the study aimed to identify strategies for enhancing medication supply and distribution systems and understanding their constraints, rather than evaluating individual practices [22].
The NICUs were divided into sections based on ward configuration and patient acuity levels. Random selection using Excel®’s random number generator determined the nurse(s) responsible for drug preparation and administration in each section. Prior to the observation of drug preparation and administration, a written consent was obtained from participating nurses. Observers then shadowed consenting nurses closely during these processes. For ethical reasons, in the event of a potentially harmful MAE such as administering expired drugs [16] or tenfold overdoses [23], observers intervened in a non-judgmental manner. These errors, if left unaddressed, could have adverse consequences for patients and were thus included in the dataset.
To ensure data validity, clinical pharmacists at each study site, separate of the research team, observed 10% of drug preparations and administrations. They concurrently observed these drug preparations and administration with the observer but recoded them independently. Agreement between the findings of pharmacists and observers was essential to validate the data.
The error identification of each sample was independently assessed by two clinical pharmacists with at least six years’ of clinical experience, who were not involved in data collection. They independently assigned errors to the observed samples to minimise bias. Any discrepancies were addressed through a discussion with the research team to reach consensus.
The comprehensive details of the data collection procedure are outlined in our published protocol for the development and validation of a risk prediction tool aimed at identifying neonates at risk of MAEs [24]. In summary, the flowchart for the direct observational study is shown below (Fig 1).
Potential clinical outcome
The potential clinical outcome of the MAEs detected was determined using a validated visual analogue scale [13]. Dean and Barber recommended that at least four experienced healthcare professionals are required for a reliable, valid, and generalizable score of potential harm as a result of MAE [13]. The generalisability coefficient of an expert panel consisting of at least five judges was found to be 0.859. A coefficient of at least 0.8 indicates an acceptable level of reliability [13]. A clear relationship between potential and actual harm when using the validated visual analogue scale was established by Dean and Barber [13]. Hence, this is a suitable measurement scale for determining the severity of MAEs.
Generic anonymised case vignettes were created based on the detected MAEs from the direct observational study. One or more types of MAEs may occur simultaneously for each errored dose, and each type of MAE was represented by a case vignette. Patient-specific information such as the current weight and diagnosis was included in the anonymised case vignettes to better inform the panel when assessing the severity of the MAEs. The panels were briefed to assess the case vignettes based on the following assumptions: (1) the MAE reached the patient; (2) the MAE was a one-time event; (3) the other medications of the patient could be disregarded unless otherwise stated; and (4) the medication processes followed standard procedures of the ward unless otherwise described. MAEs occurring among neonates with the same diagnosis and within the same weight category for the same drug (extremely low birth weight < 1000 g, very low birth weight <1500 g, low birth weight < 2500 g) were combined into a single case vignette. The severity score for each of these combined case vignettes was then applied to the specific dose observed. Examples of case vignettes for each category of MAEs are provided in S1 Appendix. A similar method was employed in other studies [25, 26].
The expert panel was requested to individually and independently rate each case vignette using a visual analogue scale, which was defined as [13]:
- Minimal effect (0–2)—No anticipated lasting effects and only minimal patient discomfort.
- Moderate effects (3–6)—Likely to produce lasting effects and may interfere with treatment.
- Severe effects (7–10)—Likely cause life-threatening or lasting effects and could also result in death.
The expert panel was briefed by the principal investigator on the background of the study and the assumptions to be considered when rating each case vignette. The expert panel was allowed to ask questions about the drugs used in the case vignettes and access any references to assist them in rating the vignettes. However, they were not allowed to discuss or reveal their ratings to avoid influencing each other’s ratings.
Potential economic outcome
Since the decision for a treatment to be rendered to treat MAEs is determined by the physician, only the neonatologists and the paediatrician of the expert panel were required to provide their opinion on the types of resources that would be expended in response to the MAE as a potential economic outcome. In this study, the cost of MAEs detected was defined as the additional costs incurred by hospitals to treat patients with MAEs. In Malaysia, healthcare services in the public sector are heavily subsidised by the government, with patients paying only nominal fees for both outpatient and inpatient services. The public healthcare sector caters to almost 70% of the population [27]. All costs considered in this study were derived from the perspective of the public healthcare provider. Only direct medical costs were considered in this study. Direct medical costs such as observations needed, drug therapy, diagnostic tests, laboratory tests, radiology tests, and additional days of hospitalisation were incurred as an immediate result of the MAE, while labour costs were the time required by the healthcare staff involved, such as the physician, nurse, and pharmacist, to address the consequences of the MAE [28]. The types of costs considered are important when quantifying the economic impact of MAEs, as the resources utilised for the treatment of these patients would not have been required had there been no MAEs.
The costs were derived based on the fee order [29–32], salary schedules [33–35], and local inventory drug price list. A consensus was reached among the experts for the amount of time spent by the physicians, nurses, and pharmacists treating the consequences of the MAEs. The labour costs of the healthcare team were then derived based on the salary schedules. Since the wages of neonatologists, general physicians, and pharmacists in the public sector depend on years of experience, these wages are based on the most common salary grade in Malaysia. This conservative approach was used to avoid inflation of the expected cost since the years of experience of these healthcare professionals varied greatly among the study sites. On the other hand, the wages for the nurses were based on the median nursing experience of 10 years among the nurses observed in the direct observational study.
In each of the case vignettes, the expert panel was asked to document the impact of the MAEs by choosing the consequences shown in Table 2.
Research variables and outcome measures
Study outcome.
The outcome of interest in the direct observational study was the occurrence of MAEs among neonates in the NICU. MAE was defined as any deviation observed during the preparation or administration of medications from the medication order, hospital policy, or manufacturer’s instructions, as provided in the product leaflet [15]. The outcome of interest in this study was the potential clinical outcome, also known as the severity of MAEs detected, and the potential cost associated with these MAEs.
Independent variables.
Various variables were identified through the literature and then reviewed by an expert panel consisting of a paediatrician with 14 years of clinical experience, a clinical pharmacist with 11 years of clinical experience, and a senior nurse who specialises in neonatology with 16 years of clinical experience. The identified variables were then categorised and defined in Table 3.
Ethical and research approvals
The study was conducted in accordance with the Declaration of Helsinki and approved by the Medical Research and Ethics Committee, Ministry of Health Malaysia (NMRR-21-1484-59494 [IIR]) and the Medical Ethics Committee, Universiti Kebangsaan Malaysia (JEP-2022-038). A written informed consent was obtained from the nurses prior data collection. We also obtained written approval and permission to conduct the study from the hospital director and the head of department of each study site.
Potential clinical outcome
The ratings from the expert panel for the potential clinical outcome of MAEs were calculated based on the severity of MAEs, and their average was then categorised as mild (0–2), moderate (3–6), or severe (7–10). Univariate multinomial logistic regression was conducted to evaluate the associations between each independent variable. Mildly severe MAEs were assigned to the reference group. Variables with a p value of less than 0.05 according to the univariate analysis were included in the multivariate analysis using multinomial logistic regression. Variables with a p value of less than 0.05 were then selected using the backward stepwise method and retained in the final model. All the significant variables were assessed for multicollinearity, and variables with variance inflation factor values greater than 10 were excluded from the final model. Pearson chi-square statistics and classification results were assessed for model fitness. The strength of association for each of the independent variables was reported using crude and adjusted odds ratios (AORs). All the statistical tests were performed using IBM SPSS Statistics for Windows, version 28.0. Armonk, NY (IBM Corp.).
Potential economic outcome
All cost estimates were provided in the local currency units of Ringgit Malaysia (MYR). It was then adjusted to the United States Dollar (USD) based on the World Bank’s growth domestic product per capita (PPP) of 1.58 [37]. This index reflects the units of the MYR needed, based on the country’s purchasing power, to purchase the same goods and services in the U.S. using USDs.
Simple linear regressions were conducted to examine the association between the cost of MAEs and individual independent variables. Variables with a p value of less than 0.05 were then included in the multiple linear regression analysis. The backward stepwise method was applied in the analysis for the final model, and variables were retained if their p values were less than 0.05. All the significant variables were assessed for multicollinearity and interactions. The assumptions of homoscedasticity and a normal distribution of variance for the final model were then assessed. The strength of association for each independent variable was reported using both crude and adjusted regression coefficients (B). All the statistical tests were performed using IBM SPSS Statistics for Windows, version 28.0. Armonk, NY (IBM Corp.).
Results
Direct observational study
The direct observational study included 170 patients, who were predominantly male (61.8%). The median (interquartile range, IQR) age was 35.0 weeks (IQR 9.0), while the median (IQR) birth weight was 1,950 g (IQR 1,665). The median (IQR) for the number of medications prescribed per patient was 2.0 (IQR 1.0).
A total of 1,093 medication doses were administered in the direct observational study. The rate of MAE was 68.0% (95% confidence interval (CI) 65.1%-70.7%). Among the 1,093 observed doses, 743 had at least one error, affecting 92.4% (157/170) of the neonates. Of the 743 observed doses with errors, a total of 1,288 types of MAEs were detected. The detailed MAE types detected are presented in Table 4.
Characteristics of patients and nurses involved in the MAEs
A total of 157 patients experienced MAEs in the direct observational study. Ninety-seven of the patients were male. The median (IQR) gestational age was 35.0 weeks (IQR 8.8), while the median (IQR) birth weight was 1,930 g (IQR 1,660).
A total of 127 nurses were observed to have committed MAEs during the direct observational study. The average years of nursing experience at the NICU of the study site was 7.05 years (SD = 5.07), while the average overall years of nursing experience was 10.94 years (SD = 6.67). Half (66/127, 52.0%) of the nurses who committed MAEs were nurses with advanced diplomas in neonatology. The characteristics of the patients who experienced MAEs and the nurses who committed the MAEs are shown in Table 5.
Potential clinical outcome
A total of 1,288 types of MAEs involving 66 drugs were included in this study. The expert panel reviewed 423 anonymised case vignettes to ascertain the potential clinical outcome of MAEs. The overall mean severity score in this study was 4.15, which is moderate for all MAEs detected. The majority of the errors, 79.0% (1,018), were found to be potentially moderate in severity, while 18.6% (240) of the errors were found to be potentially mild. Only 2.3% (30) of the errors were believed to be potentially severe. A detailed analysis of the types and severities of the errors is provided in Table 6, while examples of the errors according to their severity are provided in Table 7.
According to the univariate analysis, factors significantly associated with moderately severe MAEs compared to mildly severe MAEs were the route of administration, number of medications administered, availability of protocol related to the preparation and administration of medications, gestational age, and nurses’ experience at the study site. While factors significantly associated with severe MAEs compared to mildly severe MAEs were route of administration, birth weight, gestational age, high alert medications, number of medications prescribed, and nurses’ experience at the study site. Details of the crude odds ratios (AORs), 95% confidence intervals (CIs), and p values are presented in Table 8.
After all the possible factors were adjusted as mentioned in the methods section, the probability of occurrence of moderately severe MAEs was found to be significantly greater for patients treated with intravenous medications (AOR = 2.84; 95% CI = 2.05–3.93; p<0.001), younger neonates (AOR = 0.95; 95% CI = 0.92–0.98; p = 0.001), and higher number of medications administered (AOR = 1.06; 95% CI = 1.02–1.11; p = 0.005). The significant risk factors for severe MAEs were intravenous medication (AOR = 5.82; 95% CI = 2.01–16.84; p = 0.001), high alert medications (AOR = 5.81; 95% CI = 1.58–21.33; p = 0.008), unavailability of a protocol related to the preparation and administration of medications (AOR = 3.04; 95% CI = 1.05–8.83; p = 0.041), and younger neonates (AOR = 0.85; 95% CI = 0.78–0.93; p<0.001). The adjusted odds ratios (AORs), 95% confidence intervals (CIs), and p values are shown in Table 9.
The likelihood ratio chi-square test indicated that the full model had a good model fit, showing a significant improvement in fit compared to the intercept-only model (χ2 (10) = 90.68, p < 0.001). Overall, the accuracy of the model was 79.0%. Based on McFadden’s R2 statistics, the full model containing the predictors represented a 6% improvement in fit relative to the null model.
Potential economic outcome
In this study, the potential cost of the 743 errored doses observed was estimated at MYR 43,664.16 (USD 27,452.10). Half of the potential cost, which amounted to MYR 21,861.84 (USD 13,749.58), was attributed to the additional observation needed as a result of the MAEs, while only 2.7% of the potential cost was attributed to the change in drug therapy (Table 10). The median (IQR) cost per MAE in this study was 22.19 (32.10) for MYR, which was 14.04 (2.22–22.53) for USD.
According to the univariate analysis, the route of administration and the use of high-alert medications were found to be associated with the cost of MAEs and were retained for the multiple linear analysis (p < 0.05) (Table 11). According to the multiple linear regression analysis, the regression model was found to be statistically significant (F = 8.022, p < 0.001; adjusted R2 = 0.011), and the statistically significant independent variables retained in the model were the intravenous route of administration (Adj. B = 9.08, 95% CI = 1.71–16.44; p = 0.016) and the presence of high-alert medications (Adj. B = 28.12, 95% CI = 10.59–45.64; p = 0.002) (Table 12).
Discussion
The overall mean severity score in this study was 4.15 for all MAEs detected, with 85.5% of these MAEs being potentially moderate to severe. It was estimated that the MAEs detected could have potentially cost the Ministry of Health, Malaysia, close to MYR 45,000 (USD 28,000). Medications administered intravenously, the presence of high-alert medication, unavailability of a protocol related to the preparation and administration of medications, and younger neonates were significant factors associated with severe MAEs, while medications administered intravenously, an increased number of medications administered, and younger neonates were factors associated with moderately severe MAEs. Factors significantly associated with costlier MAEs were the intravenous route of administration and the presence of a high-alert medication.
The mean severity score and the proportion of moderately severe and severe MAEs in this study were greater than those in another study in the Netherlands conducted using a similar methodology, which reported that 57% and 1% of the MAEs were potentially moderately severe and severe MAEs, respectively [15]. The MAE in the study by Chedoe et al. [15] was 48.0%, whereas the error rate was 68.0% in our direct observational study. The difference in severity could be due to the difference in the error rate and the expert panel reviewing the MAEs detected. Despite having the same setting and methodology, the sample size in Chedoe et al was smaller than that in our study. Furthermore, instead of employing a minimum of four healthcare professionals to assess the severity of MAEs as recommended by Dean and Barber, an expert panel consisting of three healthcare professionals was used. Hence, a true representation of the population and the reliability of the severity scores may be questioned, and the results may not be extrapolated to the entire population.
A local study by Zainal et al reported that the median daily admission cost at two NICUs in Malaysia ranged from MYR 409 (USD 111) to MYR 496 (USD 135) [38]. The cost incurred for a median of two doses per drug administration round ranged between MYR 22.19 (USD 14.04) and MYR 110.95 (USD 70.22) per neonate in the direct observation study. Consequently, given the frequent drug administration rounds scheduled in a day, the potential cost incurred by MAEs alone could cost more than the median daily admission cost. The direct observational study was conducted at five study sites over a total of 60 days. Extrapolating these findings to a year suggests that the MAEs detected could approximately cost MYR 270,000 (USD 168,000) annually at these study sites. Although this sum may not seem significant, extending it nationwide to the 45 NICUs under the Ministry of Health, Malaysia, where healthcare is heavily subsidised and neonatal care is predominantly funded, the potential cost incurred by MAEs would be a significant burden. A recent study conducted in Denmark reported an annual incremental cost of €1,808,600 for an intervention implemented in the entire hospital, administering 2.3 million medications annually over a ten-year period [39]. Therefore, policymakers and stakeholders could consider redirecting the potential costs identified in this study towards the implementation of such error-reducing strategies.
Although the proportion of neonates potentially affected by severe MAEs is small, the number of neonates admitted to the NICU annually is approximately 3,400 across the five study sites, and the number of intravenous medications administered daily indicates that MAEs may be more common than expected and may cause significant patient harm [40]. In our study, drugs administered intravenously increased the risk of severe and moderately severe MAEs by 5.8 and 2.8 times, respectively. These were expected to cost MYR 9.08 (USD 5.75) more than drugs that are administered orally. The intravenous route of administration has been associated with an increased frequency of MAEs [8, 41]. This is mainly due to the increased opportunities for errors during administration as a result of the complex multistep preparation processes of the medications [5, 15]. According to a 5-year review of incident reports, the percentage of harmful intravenous-related MEs was greater than the percentage of all harmful errors combined [42]. Thus, it has been agreed universally that attention should be given to reducing intravenous MAEs, especially since they are associated with greater patient harm [41, 43]. Such strategies include implementing and enforcing protocols relating to the preparation and administration of intravenous medications and implementing a centralised intravenous admixture service [44, 45].
High-alert medications are medications with a heightened risk of causing significant patient harm when these medications are used erroneously [46]. Severe MAEs were almost 6-fold more likely to occur when high-alert medications were administered. In our study, significant linear relationships between the cost of MAEs and the use of high-alert medications (p = 0.004) were found, increasing the cost by MYR 28.12 (USD 17.80). A similar relationship between the severity of MAEs and high-alert medication use was found in a study by Sakowski et al. [47]. The use of these medications was also found to be more likely to result in harmful MEs among neonates in NICUs [48]. Therefore, various risk reduction strategies are recommended to reduce harm when high-alert medications are used during medication administration. Examples of these strategies include standardising the preparation and administration of these medications with the involvement of a clinical pharmacist [46, 49], implementing barcode medication administration safety and limiting the administration of high-alert medications to special authorised staff [46, 50], improving access to information about these medications [46] and enforcing double checking when administering medications [51].
In this study, we found that for each increase in the number of medications administered, the odds of having moderately severe MAEs increased by 6%. Studies have also suggested that an increase in nursing workload, indicated by the number of medications administered, is associated with errors such as omissions and situational violations, leading to severe MAEs and patient harm [52, 53]. Another factor significantly associated with both moderately severe and severe MAEs was gestational age. With an increase of one unit in gestational age (weeks), the odds of moderately severe and severe MAEs decreased by 5% and 15%, respectively. Extremely premature neonates are more susceptible to acute complications than their older and heavier counterparts [54, 55].
Our study identified the absence of a protocol related to the preparation and administration of medications as a significant factor associated with severe MAEs, with severe MAEs being three times more likely to occur in the absence of a protocol. This finding is consistent with previous studies that also found a significant association between absence of protocol and the occurrence of MAEs [56, 57]. Our results suggest that the absence of protocols may contribute to an increased risk of severe MAEs, highlighting the importance of implementing and adhering to established guidelines in healthcare settings. The significance of this finding is further emphasized by the World Health Organisation’s recognition of the importance of guidelines in enhancing healthcare professionals’ capacity and improving patient care quality [1]. By providing clear instructions and procedures, guidelines can help standardize medication administration practices, reduce variability, and minimize the likelihood of severe errors occurring [50].
Several predictors, while potentially holding practical importance in this study, were found to be solely significant in the univariate analysis. The inclusion of multiple variables in the multinomial logistic regressions revealed a more nuanced picture, where the significance of certain variables may be reverse when accounting for other predictors. Nevertheless, the clinical relevance of some of these predictors should be taken into consideration.
The relationship between nurses’ length of experience and MAEs in various acute-care settings has been extensively studied in the literature, with varying findings across different studies [58]. Most studies found that the greater nurse experience is associated with reduced occurrence of MAEs. This could be attributed to the advanced pharmacological knowledge typically possessed by experienced nurses, leading to enhanced medication safety [58]. While our study specifically investigated the severity of MAEs, our findings did not reveal a significant relationship between nurses’ experience and the occurrence of moderately severe and severe MAEs when compared to mild MAEs. Despite expectations that increased experience would correlate with decreased error severity, mirroring the observed trend in error occurrence, our results did not reveal this. One possible explanation to this discrepancy lies in the differing behaviors exhibited by junior and senior nurses in medication administration practices. Junior nurses, perhaps more cautious in adhering to medication orders, may demonstrate greater accuracy in medication administration. Conversely, senior nurses, while possessing extensive experience and knowledge, may not consistently adhere to protocols with the same precision, potentially leading to errors of greater severity [59]. This may contribute to the non-significant relationship observed between nurse experience and MAE severity in our study.
A 6-month study conducted in the United States identified that there was a curvilinear relationship between MAEs and the proportion of nurses with degree. While an increase in the number of nurses with degree was associated with a decrease in severe MAEs, no corresponding decrease was observed in non-severe MAEs [60]. The study also identified an optimal proportion of degree-prepared nurses at 54%, indicating that exceeding this percentage did not lead to further improvements; instead, the relationship became curvilinear. This finding may explain the higher incidence of moderately severe MAEs (79.0%) as compared to severe MAEs (2.3%), as 52.0% of the observed nurses held an advanced diploma in neonatology. Similar results were found in another direct observational study, where 68.6% of the detected MAEs were moderate, and 2.9% were severe [61].
Interestingly, interruptions were not found to be significantly associated with the severity of MAEs in our study, contrary to findings from previous research [62, 63]. For instance, prior study has reported a doubling of the estimated risk of majorly severe errors in the presence of four interruptions [63]. One plausible explanation for this contradicting finding lies in the differing prevalence of interruptions across studies. In our investigation, only 11.8% of nurses experienced interruptions or distractions during medication preparation and administration, compared to a markedly higher rate of 90.8% reported in another study [62]. Moreover, despite the implementation of interventions such as the ’Do Not Disturb’ vests and their inclusion as a parameter in the annual nursing audit in this study, adherence to wearing the safety vest remained low at 4.8%. This finding aligns with a recent study in France, which similarly found no impact of safety vest implementation on MAEs or interruption rates [64]. Therefore, we hypothesize that the significantly lower frequency of interruptions observed in our study may be attributed to the Hawthorn effect, whereby nurses modified their behaviour in response to being observed in a research setting [65].
Strengths and limitations
This study has several limitations. Since the MAEs evaluated by the expert panel in this study were considered distinct events, the potential cumulative effects of multiple MAEs occurring simultaneously in a neonate were not included for review by the expert panel. The cost analysis conducted in this study may not be generalizable due to the variability in costs among the different geographic regions and payers. Direct nonmedical costs, indirect costs, and intangible costs were not included in this study. However, when considering these costs, the overall cost of MAEs would be much greater. Despite these limitations, this study is the largest direct observational study conducted among neonates in the NICU to assess the potential clinical outcome of the MAEs detected and the first to estimate the potential economic outcome of these MAEs.
Recommendations for further research
Future research should focus on interventions overcoming the significant factors associated with the severity and economic impact of MAEs to improve patient safety.
Implications for policy and practice
We anticipate that the insights attained from this study will assist policymakers and stakeholders in prioritising and implementing MAE-reducing strategies such as the implementation of a centralised intravenous admixture service and standardised national protocols related to the preparation and administration of medications by targeting the identified risk factors to prevent the occurrence of severe MAEs. Cost estimates from this study also provide a better understanding of the costs involved and further stress the need to identify cost-effective interventions to improve patient safety.
Conclusion
The majority of MAEs detected were judged to be moderate in severity. The potential costs of the 743 errored doses observed were MYR 45,000 (USD 28,000) over 60 days, possibly leading to a significant economic burden. Factors associated with severe MAEs were the intravenous route of administration, the use of high-alert medications, the increased number of medications prescribed, and neonates with lower birth weights. Moreover, factors associated with moderately severe MAEs were intravenous medication and a greater number of medications prescribed. The use of high-alert medications and the intravenous route of administration were significantly associated with costlier MAEs. Understanding these factors is essential in implementing effective error-reducing strategies targeting the identified factors to improve patient safety among neonates in the NICU.
Acknowledgments
The authors would like to thank the Director-General of Health Malaysia for his permission to publish this article. We would also like to thank James Voo Yau Hon for his expert opinion.
References
- 1. World Health Organization. Global patient safety challenge: medication without harm 2017 [1–16]. Available from: https://www.who.int/patientsafety/medication-safety/en/.
- 2. World Health Organization. Medication safety in high-risk situations 2019 [Available from: https://apps.who.int/iris/handle/10665/325131].
- 3. Elliott RA, Camacho E, Jankovic D, Sculpher MJ, Faria R. Economic analysis of the prevalence and clinical and economic burden of medication error in England. BMJ Qual Saf. 2021;30(2):96–105. pmid:32527980
- 4. Salmasi S, Khan TM, Hong YH, Ming LC, Wong TW. Medication errors in the Southeast Asian countries: a systematic review. PLoS One. 2015;10(9):e0136545. pmid:26340679
- 5. Krzyzaniak N, Bajorek B. Medication safety in neonatal care: a review of medication errors among neonates. Ther Adv Drug Saf. 2016;7(3):102–19. pmid:27298721
- 6. Härkänen M, Vehviläinen-Julkunen K, Murrells T, Rafferty AM, Franklin BD. Medication administration errors and mortality: incidents reported in England and Wales between 2007–2016. Research in Social and Administrative Pharmacy. 2019;15(7):858–63.
- 7. Alghamdi AA, Keers RN, Sutherland A, Ashcroft DM. Prevalence and nature of medication errors and preventable adverse drug events in paediatric and neonatal intensive care settings: a systematic review. Drug Saf. 2019;42(12):1423–36. pmid:31410745
- 8. Henry Basil J, Premakumar CM, Mhd Ali A, Mohd Tahir NA, Mohamed Shah N. Prevalence, Causes and Severity of Medication Administration Errors in the Neonatal Intensive Care Unit: A Systematic Review and Meta-Analysis. Drug Saf. 2022;45(12):1457–76. pmid:36192535
- 9. Cousins DH, Dewsbury C, Matthew L, Nesbitt I, Warner B, Chamerlain J, et al. NPSA Safety in doses: medication safety incidents in the NHS: the fourth report of the patient safety observatory. London.2007.
- 10. Kale A, Keohane CA, Maviglia S, Gandhi TK, Poon EG. Adverse drug events caused by serious medication administration errors. BMJ Qual Saf. 2012;21(11):933–8. pmid:22791691
- 11. Gates PJ, Baysari MT, Mumford V, Raban MZ, Westbrook JI. Standardising the Classification of Harm Associated with Medication Errors: The Harm Associated with Medication Error Classification (HAMEC). Drug Saf. 2019;42(8):931–9. pmid:31016678
- 12.
National Co-ordinating Council for Medication Error Reporting and Prevention. NCCMERP index for categorizing medication errors [Available from: http://www.nccmerp.org/types-medication-errors].
- 13. 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. pmid:10048880
- 14. Socié té française de pharmacie clinique. Revue des erreurs lié es aux mé dica- ments et dispositifs associés–REMED. Une méthode d’amélioration de la qualité 2008. 2008 [Available from: http://www.adiph.org/sfpc/manuel_de_la_REMED_ Version_2008.pdf].
- 15. Chedoe I, Molendijk H, Hospes W, Van den Heuvel ER, Taxis K. The effect of a multifaceted educational intervention on medication preparation and administration errors in neonatal intensive care. Arch Dis Child Fetal Neonatal Ed. 2012;97(6):F449–55. pmid:22491014
- 16. Ramirez-Camacho M, Ake N, Gloria A, Torres-Romero J. Medication errors of intravenous therapy in the neonatal intensive care unit of a second-level hospital in southeastern Mexico. Latin American Journal of Pharmacy. 2020;39(3):604–11.
- 17. Cherif A, Sayadi M, Ben Hmida H, Ben Ameur K, Mestiri K. [Evaluation of administration errors of injectable drugs in neonatology]. Ann Pharm Fr. 2015;73(6):461–70.
- 18. Cranshaw J, Gupta KJ, Cook TM. Litigation related to drug errors in anaesthesia: an analysis of claims against the NHS in England 1995–2007. Anaesthesia. 2009;64(12):1317–23. pmid:19849678
- 19. Naing L, Nordin RB, Abdul Rahman H, Naing YT. Sample size calculation for prevalence studies using Scalex and ScalaR calculators. BMC Medical Research Methodology. 2022;22(1):1–8.
- 20. Raja Lope RJ, Boo NY, Rohana J, Cheah FC. A quality assurance study on the administration of medication by nurses in a neonatal intensive care unit. Singapore Med J. 2009;50(1):68–72. pmid:19224087
- 21. Barker KN, McConnell WE. The problems of detecting medication errors in hospitals. Am J Hosp Pharm. 1962;19:360–9.
- 22. Van Der Veen W, Van Den Bemt PM, Wouters H, Bates DW, Twisk JW, De Gier JJ, et al. Association between workarounds and medication administration errors in bar-code-assisted medication administration in hospitals. Journal of the American Medical Informatics Association. 2018;25(4):385–92. pmid:29025037
- 23. Tse Y, Tuthill D. Incidence of paediatric 10-fold medication errors in Wales. Archives of disease in childhood. 2020. pmid:33109522
- 24. Henry Basil J, Premakumar CM, Mhd Ali A, Mohd Tahir NA, Seman Z, Mohamed Shah N. Development and validation of a risk prediction model for medication administration errors among neonates in the neonatal intensive care unit: a study protocol. BMJ Paediatr Open. 2023;7(1). pmid:36754439
- 25. Assunção-Costa L, Pinto CR, Machado JFF, Valli CG, de Souza LEPF. Assessing the severity of medication administration errors identified in an observational study using a valid and reliable method. Journal of Pharmaceutical Policy and Practice. 2023;16(1):143.
- 26. Geeson C, Wei L, Franklin BD. High‐risk medicines associated with clinically relevant medication‐related problems in UK hospitals: A prospective observational study. British Journal of Clinical Pharmacology. 2020;86(1):165–9. pmid:31495965
- 27.
Ministry of Health M. Health White Paper for Malaysia. In: Ministry of Health M, editor. 2023. p. 63.
- 28.
Jacobs P, Rapport J. The economics of health and medical care. 5th ed. Sudbury, Massachusetts: Jones and Bartlett; 2004.
- 29.
Federal Government of Malaysia. Fees (Medical) (Cost of Services) Order 2014. In: Ministry of Health M, editor.: Attorney General’s Chambers of Malaysia; 2014.
- 30. Federal Government of Malaysia. Fees (Medical) Order 1982. Attorney General’s Chambers; 1982.
- 31. Federal Government of Malaysia. Fees (Medical) (Amendment) Order 2017. In: Ministry of Health M, editor.: Attorney General’s Chambers of Malaysia; 2017.
- 32. Federal Government of Malaysia. Fees (Medical) (Amendment) Order 1994. Attorney General’s Chambers; 1994.
- 33. Federal Government of Malaysia. Pekeliling Perkhidmatan Bilangan 7 Tahun 2011. In: Public Service Department M, editor. 2011.
- 34. Federal Government of Malaysia. Pekeliling Perkhidmatan Bilangan 7 Tahun 2019. In: Public Service Department M, editor. 2019.
- 35. Federal Government of Malaysia. Pekeliling Perkhidmatan Bilangan 8 Tahun 2015. In: Public Service Department M, editor. 2015.
- 36. Meissner B, Nelson W, Hicks R, Sikirica V, Gagne J, Schein J. The Rate and Costs Attributable to Intravenous Patient-Controlled Analgesia Errors. Hospital Pharmacy. 2009;44(4):312–24.
- 37. Bank World. PPP conversion factor, GDP (LCU per international $): World Development Indicators database, The World Bank Group, 2023.; 2023 [Available from: https://data.worldbank.org/indicator/PA.NUS.PPP?locations=MY].
- 38. Zainal H, Dahlui M, Soelar SA, Su TT. Cost of preterm birth during initial hospitalization: A care provider’s perspective. PLoS One. 2019;14(6):e0211997. pmid:31237874
- 39. Jessurun JG, Hunfeld NGM, van Dijk M, van den Bemt P, Polinder S. Cost-effectiveness of central automated unit dose dispensing with barcode-assisted medication administration in a hospital setting. Res Social Adm Pharm. 2022;18(11):3980–7. pmid:35853809
- 40. O’Brien F, Clapham D, Krysiak K, Batchelor H, Field P, Caivano G, et al. Making Medicines Baby Size: The Challenges in Bridging the Formulation Gap in Neonatal Medicine. Int J Mol Sci. 2019;20(11). pmid:31159216
- 41. Sutherland A, Canobbio M, Clarke J, Randall M, Skelland T, Weston E. Incidence and prevalence of intravenous medication errors in the UK: a systematic review. Eur J Hosp Pharm. 2020;27(1):3–8. pmid:32064081
- 42. Hicks RW, Becker SC. An overview of intravenous-related medication administration errors as reported to MEDMARX, a national medication error-reporting program. J Infus Nurs. 2006;29(1):20–7. pmid:16428997
- 43. Figueredo IB, Passos SdSS, Opitz SP, de Castro CT, dos Santos DB. Prevalence of errors in the preparation and administration of intravenous drugs in adults: Meta-analysis with meta-regression. Research, Society and Development. 2024;13(1):e13213144834–e.
- 44. Chen H, Guo Y, Wei H, Chen X. The impact of pharmacist oriented mode on risk control in a Chinese centralized intravenous admixture service centre. Sci Rep. 2021;11(1):5445. pmid:33686207
- 45. Marufu TC, Bower R, Hendron E, Manning JC. Nursing interventions to reduce medication errors in paediatrics and neonates: Systematic review and meta-analysis. Journal of Pediatric Nursing. 2022;62:e139–e47. pmid:34507851
- 46. Institute for Safe Medication Practices (ISMP). ISMP List of High-Alert Medications in Acute Care Settings 2024 [Available from: https://www.ismp.org/recommendations/high-alert-medications-acute-list.
- 47. Sakowski J, Newman JM, Dozier K. Severity of medication administration errors detected by a bar-code medication administration system. Am J Health Syst Pharm. 2008;65(17):1661–6. pmid:18714114
- 48. Stavroudis TA, Shore AD, Morlock L, Hicks RW, Bundy D, Miller MR. NICU medication errors: identifying a risk profile for medication errors in the neonatal intensive care unit. J Perinatol. 2010;30(7):459–68. pmid:20043010
- 49. De Basagoiti A, Fernandez A, Mendiola S, De Miguel M, Guerra E, Loureiro B, et al. Intravenous drug use in neonatal intensive care units. Eur J Hosp Pharm. 2021;28(6):341–5. pmid:34697051
- 50. Koeck JA, Young NJ, Kontny U, Orlikowsky T, Bassler D, Eisert A. Interventions to reduce medication dispensing, administration, and monitoring errors in pediatric professional healthcare settings: a systematic review. Frontiers in Pediatrics. 2021;9:633064. pmid:34123962
- 51. Ghezaywi Z, Alali H, Kazzaz Y, Ling CM, Esabia J, Murabi I, et al. Targeting zero medication administration errors in the pediatric intensive care unit: A Quality Improvement project. Intensive and Critical Care Nursing. 2024;81:103595. pmid:38101213
- 52. Schroers G, Ross JG, Moriarty H. Nurses’ perceived causes of medication administration errors: a qualitative systematic review. The Joint Commission Journal on Quality and Patient Safety. 2021;47(1):38–53.
- 53. Härkänen M, Vehviläinen‐Julkunen K, Murrells T, Paananen J, Franklin BD, Rafferty AM. The contribution of staffing to medication administration errors: A text mining analysis of incident report data. Journal of Nursing Scholarship. 2020;52(1):113–23. pmid:31763763
- 54. Al-Turkait A, Szatkowski L, Choonara I, Ojha S. Drug utilisation in neonatal units in England and Wales: a national cohort study. Eur J Clin Pharmacol. 2022;78(4):669–77. pmid:35028673
- 55. Davidesko S, Wainstock T, Sheiner E, Pariente G. Long-term infectious morbidity of premature infants: is there a critical threshold? Journal of Clinical Medicine. 2020;9(9):3008. pmid:32961963
- 56. Tsegaye D, Alem G, Tessema Z, Alebachew W. Medication administration errors and associated factors among nurses. International journal of general medicine. 2020:1621–32. pmid:33376387
- 57. Wondmieneh A, Alemu W, Tadele N, Demis A. Medication administration errors and contributing factors among nurses: a cross sectional study in tertiary hospitals, Addis Ababa, Ethiopia. BMC nursing. 2020;19(1):1–9. pmid:31956293
- 58. Kerari A, Innab A. The Influence of Nurses’ Characteristics on Medication Administration Errors: An Integrative Review. SAGE Open Nurs. 2021;7:23779608211025802. pmid:34222653
- 59. Fahimi F, Sefidani Forough A, Taghikhani S, Saliminejad L. The Rate of Physicochemical Incompatibilities, Administration Errors. Factors Correlating with Nurses’ Errors. Iran J Pharm Res. 2015;14(Suppl):87–93.
- 60. Chang YK, Mark BA. Antecedents of severe and nonsevere medication errors. Journal of Nursing Scholarship. 2009;41(1):70–8. pmid:19335680
- 61. Kelly J, Wright D. Medicine administration errors and their severity in secondary care older persons’ ward: a multi‐centre observational study. Journal of clinical nursing. 2012;21(13–14):1806–15. pmid:21733025
- 62. Ali S, Sherali S. Association of medication administration errors with interruption among nurses in public sector tertiary care hospitals. Pakistan journal of medical sciences. 2019;35(5):1318. pmid:31488999
- 63. Westbrook JI, Woods A, Rob MI, Dunsmuir WT, Day RO. Association of interruptions with an increased risk and severity of medication administration errors. Arch Intern Med. 2010;170(8):683–90. pmid:20421552
- 64. Berdot S, Vilfaillot A, Bezie Y, Perrin G, Berge M, Corny J, et al. Effectiveness of a ‘do not interrupt’vest intervention to reduce medication errors during medication administration: a multicenter cluster randomized controlled trial. BMC nursing. 2021;20(1):153. pmid:34429095
- 65. Owen S, Menzies J, Pontefract S. Educational interventions to reduce nurse medication interruptions: A scoping review. Nurse Education Today. 2023;121:105665. pmid:36527755