Prevalence of medication administration errors and its determinants among nurses in Ethiopia: a systematic review and meta-analysis

Abstract

Background

Despite nurses being the backbone of patient care, medication administration errors (MAEs) remain a serious risk to patient safety in low-income countries, including Ethiopia. However, the previous review was outdated, included fewer than 10 studies, focused solely on tertiary hospitals, and did not pool determinants. As a result, this meta-analysis aimed to determine the pooled prevalence of MAEs and their determinants among nurses in Ethiopia, addressing gaps in the study setting, time, and outcomes.

Methods

The study protocol was registered in PROSPERO. Observational studies conducted in Ethiopia and published in English from 2010 to 2024 were included. PubMed, Scopus, EMBASE, and Google Scholar databases were used to search for studies. The Joanna Briggs Institute (JBI) checklist was used to evaluate the quality of the studies, with three authors participating in the process. Data analysis for the pooled magnitude of MAEs and their determinants was conducted using STATA 17 software with the DerSimonian and Laird random-effects model. Heterogeneity was assessed via Cochrane’s Q-test and the I² statistic, while publication bias was evaluated through funnel plots, Egger’s test, and Doi plot.

Results

Of the 264 articles retrieved, 18 studies, including 4,314 nurses, were included in the meta-analysis. The pooled magnitude of MAEs among nurses in Ethiopia was 57% (95% CI: 49–64%). Moreover, inadequate work experience [OR = 3.64; 95% CI: (3.32, 3.96); I²=0.00%], interruptions [OR = 3.53; 95% CI: (3.19, 3.87); I²=0.00%], lack of guideline availability [OR = 2.14; 95% CI: (1.63, 2.66); I²=0.00%], lack of training [OR = 3.22; 95% CI: (2.67, 3.77); I²=0.00%], night shifts [OR = 3.89; 95% CI: (3.37, 4.41); I²=0.00%], and a nurse-patient ratio ≥ 1:10 [OR = 2.82; 95% CI: (2.19, 3.45); I²=0.00%] were identified as determinants of MAEs among nurses.

Conclusion

The magnitude of MAEs in the current review was substantially higher compared to global reports and studies in Africa, highlighting the need for urgent intervention. Furthermore, inadequate work experience, interruptions, lack of guideline availability and training, night shifts, and a nurse-patient ratio ≥ 1:10 were identified as determinants of MAEs among nurses. This suggests that providing training, disseminating guidelines in accessible formats, improving staffing ratios, and fostering a culture of safety are crucial steps to reduce MAEs.

Clinical trial number

Not applicable.

Introduction

Medication administration errors (MAEs) represent a major form of unsafe care practice [1]. They contribute to 50% of adverse events, including morbidity and mortality [2, 3], and can be defined as the occurrence of at least one error made by nurses during medication administration. These errors include drug errors, dose errors, patient errors, route of administration errors, wrong-time administration, technical errors, wrong choices, or administering medication without proper assessment [4, 5]. Nurses spend a significant amount of time with patients and are the primary healthcare providers responsible for MAEs [2]. They dedicate 40% of their time to the daily medication administration process and act as the final safety check [3]. Given their professional, legal, and ethical responsibilities, nurses play a central role in preventing, identifying, and correcting errors [5].

Globally, 1 in 20 patients (5%) experience medication error-related harm, and one-quarter of these cases result in severe or life-threatening conditions [3, 6]. The incidence of medication errors is substantially higher in geriatric and high-acuity care environments, such as surgical, intensive care, and emergency medicine units [6]. Globally, more than half (53%) of medication error-related harm occurs at the prescribing stage, while 36% occurs at the reporting stage [3, 7]. In low- and middle-income countries (LMICs), prescribing-stage medication errors account for up to 80%. The African region contributes the highest proportion of preventable medication error-related harm globally (9%) [3, 6]. Among these incidents, 57.4% occurred during the prescribing stage, and 15.5% involved wrong-dose issues [8].

Medication errors impose a significant economic burden on global healthcare systems, costing an estimated US$42 billion annually and accounting for 9% of total avoidable healthcare costs [9]. Moreover, they result in 2 to 4 million cases of short- and long-term disabilities, social and family burdens, and 400,000 deaths annually [10].

Evidence suggests that medication errors are associated with personal, institutional, work-related, and professional factors. Professional factors, such as lack of training, unavailability of guidelines, poor competency, inadequate communication skills, and limited work experience, contribute significantly to medication administration errors among nurses [11,12,13]. Institutional factors, including high patient flow, heavy workload, nurse-patient ratios, working units, and staff shortages, are also key determinants [11,12,13,14]. Furthermore, personal factors, such as a lack of proactivity in seeking clarification on medication concerns, interruptions, fatigue, and burnout, further exacerbate the issue [15,16,17].

To address the critical issue of medication errors, the World Health Organization (WHO) launched a framework in 2017 called Medication Without Harm, with the ambitious goal of reducing severe patient harm by half over five years [18]. Disappointingly, only 21% of countries have committed to specific targets aimed at reducing the devastating impact of medication errors [3]. Despite Ethiopia being part of this global initiative, the proportion of medication administration errors among nurses ranges from 28.93 to 89.9%, remaining a significant patient safety concern [19,20,21,22,23].

Although one systematic review and meta-analysis on medication errors among nurses in Ethiopia was conducted [17], it is outdated (nearly five years old), included only seven studies, focused solely on tertiary hospitals in major cities, and did not examine determinants. In contrast, the current study incorporates 18 studies conducted across various regions of the country, includes hospitals at different levels, provides pooled estimates on the types of medication errors, and includes a meta-regression of the determinants of MAEs. Therefore, this meta-analysis was conducted to determine the pooled magnitude of medication errors and their determinants among nurses in Ethiopia. It aims to provide valuable insights for nurses, higher education institutions, and the Ministry of Health regarding the determinants of medication errors.

Materials and methods

Objectives and review questions

This systematic review and meta-analysis aimed to:

• 1. Determine the pooled prevalence of medication administration errors among nurses in Ethiopia.

• 2. Identify the determinants associated with medication administration errors among nurses in Ethiopia.

This study sought to answer:

• 1. What is the pooled prevalence of medication administration errors among nurses in Ethiopia?

• 2. What are the main determinants contributing to medication administration errors among nurses in Ethiopia?

Protocol registrations

This systematic review protocol is registered in the PROSPERO database (https://www.crd.york.ac.uk/prospero/;CRD42024599726) and adheres to the guidelines outlined in the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 statement (supplementary file 1).

Inclusion criteria

The inclusion criteria were based on following criteria’s:

Population: nurses;

Study setting: health care facilities in Ethiopia.

Study design: observational study (cross-sectional, cohort, and case-control designs), that clearly defined MAEs as the presence of at least one medication error made by nurses, such as drug errors, dose errors, patient errors, route of administration errors, wrong-time medication administration, technical errors, wrong choices, or administering without assessment.

Data collection method: Interviews, self-administered questionnaires, observation, and chart review methods were used to gather medication errors data in the included studies.

Primary Outcome: prevalence of medication administration errors;

Secondary outcome: determinants of medication administration errors;

Publication period: 2010–2024.

Publication type: published and unpublished electronic repository articles.

Publication Language: no language restriction.

Additionally, these studies reported the incidence of diabetic nephropathy and its predictors and were evaluated as having a low to moderate risk of bias according to the Joanna Briggs Institute (JBI) criteria.

Exclusion criteria

Studies that did not clearly define MAEs, as well as those that were qualitative research, case studies, or randomized trials, were excluded.

Information source

Relevant studies were retrieved from PubMed, Scopus, CINAHL, EMBASE, and Google Scholar databases. The search also encompassed dissertations and government reports, including all pertinent publications available from 2010 to November 10, 2024.

Searching strategy

We included all studies involving human participants conducted up to the search date. Our approach utilized comprehensive searches, applying Boolean operators to combine keywords, all organized according to the Coco Pop and PEO mnemonic framework. The criteria were as follows:

Condition medication administration errors;

Context Healthcare settings in Ethiopia;

Population nurses;

Exposure determinants;

Outcomes medication administration errors.

A Medical Subject Headings (MeSH) thesaurus, along with relevant keyword terms and phrases, was used both independently and in combination with the Boolean operators “OR” and “AND” to identify eligible articles. The authors employed the following search strategy: ((“Medication Errors“[MeSH] OR “medication administration error” OR “drug administration error” OR “medication mistake”) AND (“Nurses“[MeSH] OR “nursing staff” OR “nurse”) AND (“Ethiopia“[MeSH] OR “Ethiopia” OR “Ethiopian hospitals”)) AND (“determinant” OR “associated factor” OR “risk factor” OR “predictor” OR “contributing factor”). Moreover, screening studies in the gray literature were conducted using the phrase ‘Burden of Medication Administration Errors and Its Determinants Among Nurses in Ethiopia.’ Additionally, a cross-referencing search was performed to identify other relevant studies that may not have been retrieved in the initial database search, among the final studies included (supplementary file 2).

Study selection process

Only observational studies (cross-sectional, cohort, and case-control designs) that focus on the burden of medication administration errors and their determinants among nurses in Ethiopia, and that clearly define medication administration errors, were included in this systematic review and meta-analysis. Titles and abstracts were independently screened by three authors (MA, TA, EM) to identify potentially eligible studies. Full-text articles were then independently evaluated for inclusion by the same authors. Duplicate records were removed using EndNote 20. Any discrepancies during the screening or full-text evaluation were resolved through consensus among the authors, with a fourth author consulted if needed.

Data extraction

Data extraction was performed independently by four authors (MA, TA, YE, and EM) using a structured Microsoft Excel spreadsheet, version 2108 (Microsoft Corp). Each author extracted data from assigned studies, and results were cross-checked for consistency. Discrepancies in extracted data were resolved through discussion and consensus among the four authors, with a fifth author consulted if agreement could not be reached. The authors contacted corresponding authors of non-open-access articles via email to obtain full-text access. Publications for which the full text was unavailable were excluded from the study. The data extraction template included the following items: the study region, study setting, study design, hospital level, sample size, number of events, working unit, data collection method, author surname, publication year, predictors, and the 95% confidence interval, standard error of proportion, and odds ratio.

Outcomes

The primary outcome was to determine the pooled magnitude of medication administration errors (MAEs) among nurses in Ethiopia. To achieve this, pooled effect sizes with 95% confidence intervals (CIs) were calculated using the formula: the number of nurses who reported MAEs divided by the total number of nurses who participated in medication administration procedures. Additionally, identifying the determinants of medication administration errors was another outcome, with adjusted odds ratios, 95% confidence intervals (CIs), and standard errors calculated.

Quality assessment and critical appraisal

The quality assessment of included studies was conducted independently by four authors (MA, TA, YE, and EM) using the 8-item Joanna Briggs Institute (JBI) Critical Appraisal Checklist for cohort studies. Prior to the assessment, the JBI tool was piloted on a subset of studies to ensure consistent application of the criteria among reviewers. The assessment evaluated eight criteria: clarity of sample inclusion criteria, descriptions of research subjects and settings, validity and reliability of exposure and outcome measurements, handling of confounding factors, appropriateness of outcome measures, and statistical analysis [24]. Each study was assigned a risk of bias rating as low (≥ 7), medium (5–6), or high (≤ 4) based on the total score from the JBI tool [25]. Independent assessments were cross-checked, and any disagreements were resolved through collaborative discussion among the authors (Supplementary File 3).

Data synthesis and analysis

Data analysis was conducted using STATA version 17, applying the Der Simonian–Laird random-effects model. A significant degree of heterogeneity was observed in the magnitude of medication administration errors (MAEs), as indicated by an I² statistic of 96.37% (Cochrane Q-test, p < 0.001). Moreover, meta-regression and subgroup analysis were performed to identify the source of heterogeneity using variables such as study region, study setting, hospital level, sample size, department, data collection method, and publication year. Publication bias was assessed using a funnel plot, Egger’s test, and Doi plot. Furthermore, sensitivity analysis demonstrated the stability of the pooled effect size, suggesting that the overall findings were not influenced by any single study.

Results

This review began with 264 identified studies. Following a thorough evaluation, 164 studies were excluded before the screening stage, 74 studies were excluded after reviewing titles and abstracts, 2 study was inaccessible for review, and 6 studies were excluded due to unclear study participants. Ultimately, 18 studies met the inclusion criteria and were included in the meta-analysis (Fig. 1).

Fig. 1

Study selection for systemic review and meta-analysis of MAEs and determinants among nurses in Ethiopia, 2024

Characteristics of the included studies

This meta-analysis synthesized data from 18 studies published since 2010, involving a total of 4,314 participants. Regarding the publication years, six studies [19,20,21,22,23, 26] were conducted between 2010 and 2015, eight studies [27,28,29,30,31,32,33, 38] were conducted between 2016 and 2020, and the remaining four studies [34,35,36,37] were conducted between 2021 and 2023.The studies were conducted in diverse regions of Ethiopia, including Amhara, Oromia, Addis Ababa, Tigray, and the Southern region. In terms of level of hospitals, 8 studies [19,20,21,22, 27, 30,31,32, 38] were conducted in teaching hospitals, 7 studies [23, 26, 28, 33, 35, 36, 38] in comprehensive specialized hospitals, and the remaining 3 [29, 34, 37] in general hospitals. Moreover, 9 studies [19,20,21,22,23, 26–27, 30, 31] were conducted in single-institution settings, while the remaining 9 [28, 29, 32,33,34,35,36,37,38] were multi-center studies. Regarding working units, 13 studies [19, 23, 26,27,28, 30,31,32,33,34,35, 37, 38] were conducted in all inpatient units, 2 studies [21, 36] in intensive care units (adult, neonatal, and pediatric), 2 studies [20, 29] in pediatric inpatient units, and the remaining study in an emergency unit [22]. Furthermore, regarding data collection methods, eleven studies [19, 27, 28, 30, 35, 37, 38] used self-administered methods, five [20,21,22, 29, 36] used observation and chart review, and the remaining two studies [23, 26] used interviews and chart review (Table 1).

Table 1 Characteristics of individual studies in Ethiopia, 2024 (N = 18)

Quality assessment

Based on the JBI Critical Appraisal Checklist, 9 of the 18 studies were assessed as having a low risk of bias, whereas the remaining 9 were rated as moderate risk (supplementary file 3).

The pooled magnitude of MAEs

The pooled medication administration error rate among nurses in Ethiopia was 57% (95% CI: 49–64%), revealing substantial heterogeneity among the studies (I² = 96.37%, p < 0.000). The lowest and highest weights were 5.22 and 5.75, respectively (Fig. 2).

Fig. 2

Pooled magnitude of MAEs among nurses in Ethiopia, 2024

The pooled estimates of common types of MAEs

The most common pooled estimates of medication administration errors (MAEs) were: no or documentation error 56% (95% CI: 35-77%), wrong time 42% (95% CI: 36-49%), omission/missed dose 30% (95% CI: 22-36%), and wrong dose 29% (95% CI: 19-40%) (Table 2).

Table 2 Pooled estimates of common types of medication administration errors among nurses in Ethiopia, 2024 (N = 18)

Meta-regression

Univariate meta-regression was conducted to explore heterogeneity using study setting, study design, data collection methods, study region, publication year, hospital level, and sample size. No covariate was statistically significant (all p > 0.05), and I² remained high at 96.37%, suggesting that these factors do not account for the observed heterogeneity. In contrast, multivariate meta-regression incorporating working unit, hospital level, and study setting explained 46.47% of the heterogeneity (R² = 46.47%, I² = 93.52%, p = 0.0893). Pediatric inpatient units exhibited a significantly higher medication administration error (MAE) prevalence (β = 0.338, p = 0.046), while general hospitals showed a lower prevalence (β = -0.214, p = 0.024), compared to their respective reference categories. These findings identify working unit and hospital level as partial sources of heterogeneity (Table 3). However, residual heterogeneity may reflect unmeasured confounders or limited statistical power (n = 18).

Table 3 Multi-variate analysis for identifying source of heterogeneity for MAEs among nurses in Ethiopia, 2024 (N = 18)

Subgroup analysis

Sources of heterogeneity were evaluated based on factors such as study region, study setting, hospital level, sample size, working unit, data collection method, and publication year. However, despite these evaluations, heterogeneity persisted (Table 4).

Table 4 Subgroup analysis of studies on the prevalence of medication administration errors among nurses in Ethiopia, 2024 (N = 18)

Assessment of publication bias

To assess publication bias, both subjective (funnel plot) and objective (Egger’s test) methods were employed. The funnel plot indicated that the equal distribution of studies on both sides of the middle line suggests no significant publication bias. Moreover, the presence of studies at the bottom of the plot indicates larger standard errors, which is common in smaller studies. However, their alignment with the middle line suggests that their effect sizes are consistent with those of larger studies. Overall, this pattern suggests that the results of the current meta-analysis are likely robust and not significantly influenced by publication bias (Fig. 3).

Fig. 3

The funnel plot for analysis of publication bias for MAEs among nurses in Ethiopia, 2024

The Egger’s test revealed a p-value of 0.309, and the confidence interval (-13.56524 to 4.579698) (p-value > 0.05, with the confidence interval including zero) suggests that there is no statistically significant evidence of publication bias, strengthening confidence in the robustness and reliability of this meta-analysis’ findings (Table 5).

Table 5 Egger’s test analysis of the prevalence of medication administration errors among nurses in studies in Ethiopia, 2024 (N = 18)

The Doi plot was employed to detect bias and evaluate the certainty of the evidence. The analysis revealed a Luis Furuya-Kanamori (LFK) index of -0.37. The closer the value is to zero, the less concern there is for publication bias, which indicates minor asymmetry and suggests a tolerable potential for publication bias, though it is not extremely pronounced (Fig. 4).

Fig. 4

Doi plot for Reporting bias and certainty of evidence for MAEs among nurses in Ethiopia, 2024

Sensitivity analysis

The sensitivity analysis indicated that the pooled estimate for MAEs remained stable, unaffected by any single study. The point estimate of each individual study fell within the confidence interval of the pooled estimates, and the consistently significant p-values (p = 0.000) further supported the robustness of the findings (Fig. 5).

Fig. 5

Sensitivity analysis for MAEs among nurses in Ethiopia, 2024

Determinants of MAEs

In the meta-regression, factors such as inadequate work experience, interruptions during medication administration, lack of guideline availability, lack of training, night shifts, and a nurse-patient ratio ≥ 1:10 were identified as determinants of MAEs among nurses.

Seven studies [24, 29, 30, 32,33,34,35] indicated that nurses with inadequate work experience were nearly four times more likely to make MAEs compared to those with adequate work experience [OR = 3.64; 95% CI: (3.32, 3.96); I²=0.00%]. Similarly, eight studies [24, 26, 30,31,32,33,34,35] revealed that nurses who experienced interruptions during medication administration were almost four times more likely to make MAEs compared to nurses who remained focused during medication administration [OR = 3.53; 95% CI: (3.19, 3.87); I²=0.00%]. Moreover, five studies [27, 30, 31, 33, 35] demonstrated that nurses without guideline availability in their work settings were twice as likely to make MAEs as those with guidelines in their work settings [POR = 2.14; 95% CI: (1.63, 2.66); I² = 0.00%]. Likewise, four studies [30, 31, 33, 35] showed that nurses lacking training were three times more likely to make MAEs compared to those who had received training [OR = 3.22; 95% CI: (2.67, 3.77); I² = 0.00%]. Furthermore, three studies [24, 34, 36] indicated that nurses working in settings with a nurse-patient ratio ≥ 1:10 were nearly three times more likely to make MAEs compared to those in settings with a nurse-patient ratio of less than 1:10 [OR = 2.82; 95% CI: (2.19, 3.45); I² = 0.00%] (Table 6).

Table 6 Determinants of medication administration errors among nurses in Ethiopia, 2024 (N = 18)

Discussions

The magnitude of medication errors has increased, resulting in a growing risk of patient harm and a threat to patient safety in health institutions, particularly in low-resource settings [39]. Nurses are the frontline health workforce who interact with patients and spend significant time with them. Medication administration is also considered a major responsibility of nurses in healthcare settings [6]. Therefore, this review aimed to determine the pooled magnitude of MAEs among nurses in Ethiopia and identified inadequate work experience, interruptions during medication administration, lack of guideline availability, lack of training, night shifts, and a nurse-patient ratio ≥ 1:10 as determinants of MAEs among nurses.

Accordingly, the pooled estimate of MAEs among nurses in Ethiopia was 57% (95% CI: 49–64%). This figure is similar to findings from China, Nigeria, and Sudan [40,41,42]. However, the magnitude of MAEs in this meta-analysis was higher than the rates reported in other studies: 22% in the WHO report [6], a worldwide systematic review [43], 8.32–12% in the United Kingdom [44, 45], 20% in Saudi Arabia [46], 8.4% in Africa [8], and 39.3% in a previous study in Ethiopia [17]. On the other hand, the magnitude of MAEs in this meta-analysis was lower than the rates reported in studies from India (81%) and China (74.3%) [47, 48]. The discrepancy might be due to variations in sociodemographic factors, sample size, infrastructure of health facilities, patient flow, availability of error reporting systems, availability of medication reviews and reconciliation, automated information systems, pre-service education, and multicomponent interventions.

Nurses with inadequate work experience were nearly four times more likely to make MAEs compared to those with adequate work experience in this review. The finding is comparable with result from systematic review and meta-analysis of global practice [5, 6, 12], lower middle-income countries of Asia [11], and Australia [13]. The possible explanation might be the lack of clinical exposure and practical skills, leading to difficulty in interpreting prescriptions, calculating doses, or understanding medication safety protocols [1, 3]. Similarly, limited knowledge of medications and procedures results in a less comprehensive understanding of drug mechanisms, interactions, and side effects [2, 43]. Moreover, stress and anxiety, poor adaptation to workload and environment, insufficient training (both in-service and on-the-job training), over-reliance on protocols without critical thinking, work overload, and lack of confidence may increase the incidence of MAEs [4,5,6,7, 11, 12, 14]. This implies that comprehensive orientation programs, mentorship and supervision, robust training, regular competency assessments, a supportive work environment, and effective workload management need to be strongly implemented to ensure medication safety and reduce errors in Ethiopia.

Nurses who experienced interruptions were four times more likely to make MAEs compared to those who did not experience interruptions in this meta-analysis. This finding is similar to results from a systematic review of global practices [5, 6, 11], studies in India [47], and Africa [8]. The possible explanation might be due to the fact that disruptions affect the focus and attention of nurses, impairing their ability to follow the medication administration process accurately and increasing the incidence of MAEs [1,2,3]. Similarly, interruptions increase the cognitive load on nurses, causing them to lose track of critical details, such as the correct dosage, patient identity, or administration route [6, 7, 11]. Furthermore, interruptions lead to task switching and memory errors (e.g., failing to double-check the five rights), increase the risk of procedural errors, raise stress and anxiety, and impair decision-making and judgment. Overall, this condition increases the incidence of MAEs due to interruptions [12, 13, 15]. This highlights the critical need to establish “No Interruption Zones,” implement structured communication protocols, educate staff and patients, use technology to reduce interruptions (such as barcode scanners and electronic medication administration records [eMAR]), improve staffing ratios, encourage teamwork, and foster a culture of safety.

Nurses without guideline availability in their work settings were twice as likely to make MAEs compared to those with guideline availability in their work settings, according to the current review. This finding is similar to the results from a systematic review of global practices [5, 6, 11], as well as studies conducted in India and China [47, 48], the United Kingdom [44], and Africa [8]. The possible explanation might be that the lack of guideline availability in the workplace results in issues such as a lack of standardized protocols, knowledge gaps, increased cognitive load, reduced confidence, and inconsistent practices, all of which increase the likelihood of MAEs [1, 3, 7, 18]. This suggests that developing updated medication administration guidelines, disseminating them in accessible formats, providing training on the guidelines, integrating them into technology, fostering a culture of adherence, and monitoring and evaluating compliance can significantly reduce these errors and improve patient safety.

Nurses lacking training were three times more likely to make MAEs compared to those who had received training in the current review. This finding is consistent with results from studies, including a systematic review of global practices [5, 6, 11], and studies conducted in the United Kingdom [44, 45], India [47], China [48], Saudi Arabia [46], Africa [8], and Sudan [42]. The possible explanation might be that the lack of training significantly increases the risk of medication administration errors among nurses due to gaps in knowledge, skills, and confidence [1, 3, 7, 15, 16, 18]. This suggests that training programs focusing on medication safety, error prevention strategies, and the use of safety tools are essential to reducing MAEs in Ethiopia.

Nurses working in settings with a high nurse-patient ratio (≥ 1:10) were nearly three times more likely to make MAEs compared to their counterparts working with a nurse-patient ratio of less than 1:10 in the current meta-analysis. This finding is consistent with results from studies, including a systematic review of global practices [5, 6, 11] and studies conducted in Africa [8]. This might be due to the fact that high nurse-patient ratios significantly increase the risk of MAEs, driven by factors such as workload, fatigue, cognitive overload, and reduced compliance with safety protocols [3, 7, 16, 18]. This suggests that addressing the issue requires systemic changes, such as increasing staffing levels, implementing supportive technologies, and fostering a culture of safety.

Strengths and limitations of the study

This is the first to estimate the pooled common types of MAEs and their determinants among nurses in Ethiopia. Additionally, the meta-analysis includes a sufficient number of primary studies, providing a comprehensive representation of the data and enhancing the credibility of its findings. Although this research offers valuable insights, it has several limitations. These include significant heterogeneity among studies, which remained substantial and statistically significant even after subgroup analysis. Additionally, the study was limited to observational studies, excluded non-English publications, and could not provide pooled estimates for medication types due to insufficient reporting in most studies.

Conclusions

The magnitude of MAEs in the current review was substantially high compared to WHO report and Africa, contradicting the global action plan medication without harm in 2030, highlighting the need for urgent intervention. Moreover, inadequate work experience, interruptions during medication administration, lack of guideline availability, lack of training, night shifts, and a nurse-patient ratio ≥ 1:10 were identified as determinants of MAEs among nurses.

Practical implication for nurses and health care systems

This suggests that nurses can improve safety by pairing less-experienced colleagues with seasoned mentors during medication administration, establishing ‘no-interruption zones’ for critical tasks, increasing vigilance and double-checking medications during night shifts, and prioritizing tasks while seeking support when nurse-patient ratios exceed safe limits. Additionally, healthcare systems should standardize and distribute medication administration guidelines across all facilities, invest in continuous professional development programs, optimize staffing and support for night shifts, and enforce safe staffing ratios by hiring additional personnel.