Community Pharmacy Dispensing Errors: A Comprehensive Systematic Review on Trends and Solutions

Abstract

Dispensing errors in outpatient and community pharmacies pose significant safety risks and contribute to increased global healthcare costs. Despite their frequency, comprehensive studies on rates, causes, and interventions remain limited. This review examines current evidence on dispensing error prevalence, key risk factors, and the impact of mitigation strategies. Following PRISMA guidelines, we systematically searched PubMed, Embase, and CINAHL for studies on dispensing error rates in community and outpatient pharmacies, covering the period from January 2010 to June 2025, including English-language studies and grey literature. Eligible studies reported calculable prevalence using cross-sectional, cohort, or longitudinal designs. Two reviewers independently assessed records, extracted data, and evaluated bias using the NHLBI and Grading of Recommendations Assessment, Development, and Evaluation (GRADE) tools. In 22 studies (13 from developing, 9 from developed countries), dispensing error rates ranged from 0.001% to 11.53%, higher in community (mean 4.12%) than in outpatient pharmacies (1.85%). Common errors were wrong dose/ strength (58.6%) and look-/sound-alike drugs (47.90%). Key contributors were workload (55%), staff shortages (22%), and illegible prescriptions (30.70%). While one study reported a modest 1.40% decrease with electronic prescribing, another observed up to a 98.20% reduction using automated systems. GRADE evaluation found very low confidence in the data due to variability in study settings and methods. Dispensing errors remain a major issue, particularly in community pharmacies and underserved areas. Multimodal strategies – integrating technology, standardizing workflows, and training staff – represent a potentially effective approach to reducing errors. Although the certainty of the evidence is very low, the findings may still help guide future policies aimed at introducing error reporting systems, adopting technology, and improving consistency in research methods.

INTRODUCTION

The emergence of new diseases necessitates new drugs, increasing medication errors that significantly impact health and economies. The WHO estimates these errors cost the world around $42 billion annually.[1] Medication errors are a major but preventable cause of patient harm, worsening outcomes, and increasing costs. They range from mild side effects to severe adverse drug events (ADEs), hospitalization, or death, and may occur during prescribing, preparation, dispensing, or administration.[2,3] According to the National Coordinating Council for Medication Error Reporting and Prevention, medication errors are defined as any preventable event that can lead to improper medication use or patient harm. These errors can occur at different stages: prescribing errors happen when the prescriber selects the wrong drug, dispensing errors occur during the preparation or delivery process by the pharmacy, and administration errors take place when the medication is used by the patient or caregiver. This review specifically addresses dispensing errors, which are discrepancies between the prescribed medication and what is dispensed. These errors can result in ineffective therapy or harm to the patient.[4,5] Studies have identified multiple causes of dispensing errors, including distractions, bad prescription handwriting, similar drug names or packaging, pressure, and fatigue. Human factors like multitasking and knowledge gaps, along with system issues such as high workload and end-of-shift weariness and peak workload times, highlight the need for better pharmacy practices.[6,7] Long-term studies in community pharmacies classify dispensing errors into types like wrong drug, strength, quantity, dosage form, and labeling. While safety alerts and prescription verification systems have effectively lowered particular errors, others – especially those linked to generic substitution – have exhibited an increase under changing legal and operational circumstances.[8] Due to the serious effects of dispensing errors on safety and efficiency, structured medication processes, regulations, systematic error reporting, and ongoing monitoring are urgently needed.[9] Although frequent, dispensing errors remain understudied in community and outpatient pharmacies. This review focuses on their prevalence, types, and causes, aiming to provide recommendations to reduce such errors, enhance patient safety, and lower associated healthcare costs.

METHODS

This systematic review followed PRISMA 2020 guidelines.[10] This study aimed to assess the present data, including human participants, on the prevalence of dispensing errors in outpatient settings and community pharmacies. The PEO framework was structured as follows: The population (P) covered studies on dispensing errors; the Exposure (E) related to prevalence; the Outcome (O) for the prevalence study component was the community and outpatient pharmacy environments.

Search strategy

Up to June 11, 2025, Medline (via PubMed), Embase, and CINAHL were systematically searched using MeSH and Emtree terms for “dispensing error,” “community pharmacy,” “outpatient pharmacy,” and their synonyms.

The search approach included Pharmacy Subject Headings (MeSH) and pertinent medication error-related terms (e.g., error, mistake, wrong, incident), medication dispensing (e.g., supply, filling, refill), and community pharmacies (e.g., community, pharmacist). Expert advice helped to refine the search method. Using Google and Google Scholar, together with a review of citations from pertinent papers, a comprehensive search was done to find any possibly ignored studies. Moreover, the examination included gray literature, citation tracking, and reference lists from the selected papers to find any possibly ignored research. The search was limited to English-language studies involving humans, published between 2010 and 2025.

Eligibility criteria

Studies published from January 2010 to June 2025 were included if they assessed dispensing error prevalence in outpatient or community pharmacies using cross-sectional, cohort, or longitudinal designs with sufficient data to compute dispensing error prevalence.

Regarding exclusion criteria, studies were excluded if they used designs such as case–control, qualitative, intervention, RCTs, case reports, or mixed methods. Also excluded were reviews, abstracts, trade journal articles, and commentaries. Inpatient pharmacy studies, those without relevant error data or sufficient denominators, and articles lacking full-text access were removed.

Screening and data collection

Search results were imported into EndNote X9, and duplicates were removed. In the first screening phase, two independent reviewers assessed titles and abstracts to identify relevant studies. Only cross-sectional, cohort, or longitudinal studies were included. Full texts were reviewed based on inclusion and exclusion criteria, and those that qualified were included. A review of the gray literature – Google and Google Scholar – was done to make sure no papers were overlooked. The PRISMA criteria were used to create a checklist from the data.[10] The two reviewers stayed oblivious of one another’s assessments during all stages.

Extracted data included pharmacy setting (community or outpatient), study characteristics (author, year of publication, geographic location, and study design), number of dispensing errors, and the total number of medications dispensed to determine error prevalence. Where many articles used the same dataset, we chose the one with the greatest sample size or the longest follow-up period. To get data not included in the paper, we reached out to the relevant author. Disagreements were resolved by the third reviewer. No protocol was registered for this review in registries such as PROSPERO.

Risk of bias assessment and the level of evidence

Two reviewers assessed bias risk using the NIH’s NHLBI Quality Assessment Tool for Observational Cohort and Cross-Sectional Studies,[11] settling conflicts through discussion. The Grading of Recommendations Assessment, Development, and Evaluation (GRADE) approach evaluated evidence quality, with an additional reviewer for conflict resolution.[12] Although GRADE was originally developed for intervention studies, we applied adapted criteria suitable for prevalence research, evaluating domains such as risk of bias, inconsistency, indirectness, imprecision, and publication bias. For risk of bias, we examined each included study using the NHLBI tool for cohort and cross-sectional studies. For inconsistency, we assessed the variability in prevalence estimates across studies by considering differences in direction and magnitude of the results, overlap of confidence intervals, and potential methodological heterogeneity, rather than relying solely on statistical measures of heterogeneity. Indirectness was evaluated by comparing the study populations, settings, and definitions of dispensing errors with our review question; studies conducted in settings other than community pharmacies, involving nonrepresentative populations, or using outcome definitions substantially different from our inclusion criteria were considered to have higher indirectness. Imprecision was judged based on the width of the confidence intervals around prevalence estimates in each study and whether the sample size was adequate to provide a reliable estimate; for studies without reported confidence intervals, we considered the total sample size and event counts. Publication bias was explored qualitatively by considering the likelihood that small or clinically nonsignificant prevalence findings were underreported, and by reviewing the balance of studies across different geographical regions and time periods. As no meta-analysis was performed, the GRADE assessment was based on a qualitative synthesis of the evidence, with certainty ratings reflecting methodological limitations, consistency of findings, applicability to the target context, and precision of reported prevalence measures.

RESULTS

Study selection

The first search produced 4948 records from PubMed, Embase, and CINAHL [Figure 1]. After removing duplicates and screening titles and abstracts, 206 articles were selected for full-text review. Of these, 184 papers qualified for exclusion and were deleted. In the end, 22 full-texts qualified for inclusion and were added to the systematic review.[6,7,8,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31]

Figure 1.

PRISMA 2020 flow diagram of the systematic review process

Study characteristics

This review included 22 studies from 2010 to 2025, with most published in 2014 and 2024 (three each), and none in 2016, 2017, or 2025. Of these, 59.09% (n = 13) were from developing countries (e.g., China, India, Nigeria, Lebanon, the United Arab Emirates, Jordan, Egypt, Malaysia, Indonesia, and Saudi Arabia,) and 40.91% (n = 9) from developed ones (e.g., Australia, Finland, the United States, Spain, and Sweden).

Regarding study design, most (90.91%, n = 20) used a cross-sectional method; 4.54% (n = 1) were cohort studies, and 4.54% (n = 1) were observational without sufficient detail to specify the exact design. Concerning methodology, 54.55% (n = 12) of the studies were prospective, whereas 45.45% (n = 10) were retrospective. Pharmacy settings differed; 54.55% (n = 12) were in community pharmacies and 45.45% (n = 10) in outpatient hospital pharmacies.

Reported rates of dispensing errors showed great diversity between studies. While the lowest was reported in China (0.001% [7]), the highest error rate was found in Jordan (11.53% [13]), a remarkable difference of 11.53% between the extremes. With observation being the most prevalent (n = 8), followed by reporting systems (n = 4), chart reviews (n = 3), and patient or pharmacist self-reports (n = 4), error identification techniques varied across studies. A minority of studies relied on voluntary self-reporting (n = 1) or electronic health record review (n = 1).

While some used patient counts (n = 2) or dispensed items (n = 8), most research (n = 12) calculated prevalence using dispensed prescriptions as the denominator. Of the 14 errors, 63.64% (n = 14) concentrated only on actual errors; 18.18% (n = 4) looked at near misses, and the other 18.18% (n = 4) noted both occurrences and near misses. Table 1 offers a thorough overview of study features, including author, nation, design, environment, error identification technique, and prevalence.

Table 1.

Summary characteristics of the included studies

Author, year, country	Study design	Study type	Study period	Pharmacy setting, Incidents or near misses	Error identification method, number of errors	Type of denominator, denominator	Prevalence (%)
Abdel-Qader, 2020[13] Jordan	Cross-sectional	Prospective	October 2019–February 2020	Community pharmacy, near misses	Observation, 17,352	Dispensed items, 150,442	11.53
Abdu-Aguye, 2021[14] Nigeria	Cross-sectional	Prospective	January 2020–March 2020	Outpatient hospital pharmacy, both	Observation, 100	Dispensed items, 2060	4.85
Abdulah, 2014[15] Indonesia	Cross-sectional	Prospective	February 2013–March 2013	Community pharmacy, incidents	Observation, 47	Dispensed items, 4732	0.99
Alakhali, 2014[16] Saudi Arabia	Cross-sectional	Retrospective	September 2013–October 2013	Outpatient hospital pharmacy, incidents	Review of records, 7	Dispensed Prescriptions, 1850	0.38
Allen, 2012[17] USA	Cross-sectional	Retrospective	November 2008–October 2009	Community pharmacy, incidents	Electronic health record, 1218	Dispensed items, 83,902	1.45
Baranowski, 2015[18] USA	Cross-sectional	Retrospective	June 2012–August 2013	Community pharmacy, incidents	Chart review, 223	Patients, 7406	3.01
Chand, 2022[19] India	Cross-sectional	Prospective	February 2020–April 2021	Outpatient hospital pharmacy, incidents	Observation, 26	Dispensed Prescriptions, 1852	1.40
Chou, 2024[20] China	Cross-sectional	Retrospective	January 2012–December 2022	Outpatient hospital pharmacy, incidents	Patient or family reported, 376	Dispensed Prescriptions, 25,769,660	0.00146
Cochran, 2014[21] USA	Cross-sectional	Retrospective	May 2012–January 2013	Community pharmacy, incidents	Observation, 14	Dispensed Prescriptions, 602	2.32
Copi, 2018[22] USA	Cross-sectional	Retrospective	July 2015–March 2016	Outpatient hospital pharmacy, incidents	Chart review, 526	Dispensed Prescriptions, 10,649	4.94
de Las Mercedes Martínez Sánchez, 2013[23] Spain	Cross-sectional	Prospective	February 2010–March 2011	Community pharmacy, both	Observation, 990	Dispensed Prescriptions, 42,000	2.35
Friesner, 2011[24] USA	Cross-sectional	Retrospective	January 2005–September 2008	Community pharmacy, both	Reporting system, 1633	Dispensed Prescriptions, 170,424	0.95
Gao, 2023[7] China	Cross-sectional,	Retrospective	January 2008–December 2021	Outpatient hospital pharmacy, incidents	Reporting system, 1340	Dispensed Prescriptions, 67,735,968	0.0019
Ibrahim, 2020[25] UAE	Cross-sectional	Prospective	November 2019–April 2020	Community pharmacy, near misses	Observation, 12274	Dispensed items, 244,862	5.01
Kenawy, 2019[26] Egypt	Cross-sectional,	Retrospective	November 2014–March 2015 and May 2015-September 2015	Outpatient hospital pharmacy, near misses	Reporting system, 845	Dispensed items, 24,057	3.51
Kiron, 2013[27] India	Cross-sectional,	Prospective	August 2008–January 2009	Outpatient hospital pharmacy, incidents	Chart review, 6	Patients, 1630	0.37
Mäkinen, 2024[8] Finland	Cross-sectional	Retrospective	January 2015–December 2020	Community pharmacy, incidents	Voluntary self-reporting, 6030	Dispensed Prescriptions, 122,900,000	0.0049
Norden-Hagg, 2010[28] Sweden	Observational Longitudinal	Retrospective	July 2004–December 2007	Community pharmacy, incidents	Reporting system, 19.41	Dispensed items, 100,000	0.019
Rajah, 2019[29] Malaysia	Cross-sectional	Prospective	September 2015–October 2015	Outpatient hospital pharmacy, near misses	The pharmacist reported 187	Dispensed items, 720,000	0.026
Soubra, 2021[6] Lebanon	Cross-sectional	Prospective	July 2017–August 2017	Community pharmacy, both	Observation, 376	Dispensed items, 12,860	2.92
Talay, 2024[30] Australia	Cohort	Retrospective	June 2023–December 2023	Community pharmacy, incidents	Patients self-reported, 99	Dispensed Prescriptions, 28,165	0.35
Thomas, 2011[31] India	Observational	Prospective	October 2008–May 2009	Outpatient hospital pharmacy, incidents	Observation, 53	Dispensed Prescriptions, 4988	1.06

NR: Not reported, USA: United States of America, UAE: United Arab Emirates

Risk of bias in studies

Using the NHLBI quality assessment tool [Table 2], 14 predefined methodological criteria were evaluated for each included study. The majority of studies demonstrated a low risk of bias in critical methodological aspects, including clearly stated research objectives (C1), well-defined study populations (C2), consistent application of inclusion and exclusion criteria (C4), appropriate temporal relationship between exposure and outcome (C6), sufficient study duration (C7), and consistent outcome measurement methods (C11). All studies adequately controlled for important confounders (C14) and reported loss-to-follow-up rates of <20% (C13).

Table 2.

Risk of bias assessment of included studies based on the NHLBI’s quality assessment tool for observational, cohort, and cross-sectional studies

Study	Criteria 1	Criteria 2	Criteria 3	Criteria 4	Criteria 5	Criteria 6	Criteria 7	Criteria 8	Criteria 9	Criteria 10	Criteria 11	Criteria 12	Criteria 13	Criteria 14	Overall
Abdel-Qader, 2020[13]	L	L	L	L	L	L	L	L	L	L	L	L	NA	L	L
Abdu-Aguye, 2021[14]	L	L	NR	L	H	L	L	L	L	L	L	H	NA	L	L
Abdulah, 2014[15]	L	L	NR	L	H	L	L	L	L	H	L	H	NA	L	L
Alakhali, 2014[16]	L	L	NR	L	H	L	L	L	L	H	L	H	NA	L	L
Allen, 2012[17]	L	L	NR	L	H	L	L	L	L	H	L	H	NA	L	L
Baranowski, 2015[18]	L	L	NR	L	H	L	L	L	L	H	L	H	NA	L	L
Chand, 2022[19]	L	L	NR	L	L	L	L	L	L	L	L	L	NA	L	L
Chou, 2024[20]	L	L	NR	L	H	L	L	L	L	L	L	H	NA	L	L
Cochran, 2014[21]	L	L	NR	L	H	L	L	L	L	H	L	H	NA	L	L
Copi, 2018[22]	L	L	NR	L	H	L	L	L	L	H	L	H	NA	L	L
de Las Mercedes Martínez Sánchez, 2013[23]	L	L	NR	L	H	L	L	L	L	H	L	H	NA	L	L
Friesner, 2011[24]	L	L	NR	L	H	L	L	L	L	L	L	H	NA	L	L
Gao, 2023[7]	L	L	NR	L	H	L	L	L	L	L	L	H	NA	L	L
Ibrahim, 2020[25]	L	L	L	L	L	L	L	L	L	L	L	L	NA	L	L
Kenawy, 2019[26]	L	L	NR	L	H	L	L	L	L	L	L	H	NA	L	L
Kiron, 2013[27]	L	L	NR	L	H	L	L	L	L	H	L	H	NA	L	L
Mäkinen, 2024[8]	L	L	L	L	H	L	L	L	L	L	L	H	NA	L	L
Norden-Hagg, 2010[28]	L	L	L	L	H	L	L	L	L	L	L	H	NA	L	L
Rajah, 2019[29]	L	L	NR	L	H	L	L	L	L	H	L	L	NA	L	L
Soubra, 2021[6]	L	L	NR	L	H	L	L	L	L	H	L	H	NA	L	L
Talay, 2024[30]	L	L	NR	L	H	L	L	NA	L	H	L	H	NR	L	Some concern
Thomas, 2011[31]	L	L	NR	L	H	L	L	L	L	L	L	H	NA	L	L

L: Low risk, H: High risk, NR: Not reported, NA: Not applicable, NHLBI: National heart, lung, and blood institute

Nevertheless, several methodological limitations were identified. Nineteen studies (86.4%) provided insufficient justification for sample size (C5), representing a high risk of bias in this domain. Most studies did not address variation in exposure assessment (C8), as dispensing errors were typically recorded as binary outcomes. All studies lacked blinding of outcome assessors (C12), a high-risk factor considering the observational nature of error detection methods (e.g., chart reviews, self-reports). Consistency of exposure measurement (C9) varied across studies; ten (45.5%) were rated high risk due to reliance on nonstandardized reporting systems or single-point assessments (C10).

Three studies – characterized by prospective data collection and direct observation – achieved consistently low risk of bias across almost all domains.[8,13,25] In contrast, the study by Talay (2024) had two unreported domains (C8 and C13), indicating methodological shortcomings.

Overall, 21 studies (95.4%) were classified as having a low overall risk of bias, while one studies (5.6%) were rated as some concern. Although high-risk ratings in sample size justification (C5), repeated measures of outcome (C10), and blinding of outcome assessment were frequent (C12), these domains were judged noncritical for the primary outcomes. Consequently, the majority of studies were considered to have a low overall risk of bias.

Prevalence of dispensing errors

With a median prevalence of 1.40%, the systematic analysis of 22 included studies showed variation in medicine dispensing error rates, with reported prevalences ranging from 0.001%[7] to 11.53%.[13] The mean error rate in community pharmacies was 4.12%, while in outpatient hospital pharmacies it was 1.85%.

Geographic distribution

Geographic analysis revealed that research done in underdeveloped countries (59.09%, n = 13) indicated greater average error rates (5.20%) than those from industrialized nations (40.91%, n = 9; mean: 1.10%). This variance could suggest variations in pharmacist staffing ratios and electronic prescribing infrastructure.

Common error types and contributing factors

Error type analysis revealed improper dose/strength as the most often reported dispensing error, with peaks of 58.60%[30] and 50.40%.[8] Automated systems’ implementation was found to cut such errors by 96.20%.[7] Often linked to manual counting techniques, wrong quantity errors accounted for 15%–47% of recorded occurrences.[13,20] In one study, look-alike/sound-alike (LASA) medication errors made up 47.90% of occurrences, highlighting the importance of consistent naming conventions.[20] While expired medications contributed 21.20% of errors in one study,[30] labeling problems accounted for 15%–22% of errors,[14,19] highlighting inventory control weaknesses.

Error severity and outcomes

Severity ratings across studies categorized 8.60%–17.30% of errors as severe (including erroneous prescription or dose errors), 38.80%–52.60% as moderate, and 44.50% as minor.[13,25] With one research reporting financial losses of $188,406,[20] the most significant errors led to hospitalizations[22] or ADEs.[17]

Impact of interventions

Though it did not eliminate them, intervention studies showed electronic prescribing cut errors by 1.40%.[26] Automated systems proved more effective, cutting strength-related errors by 95% and total errors by 98.20%.[7,28] Workflow changes like double-check protocols lowered near-miss rates by 30%.[23]

Key risk factors

Main risk elements included workload pressure (55.00% of peak-time errors), illegible prescriptions (30.70% of handwritten system errors), and staffing shortages, raising single-pharmacist shift errors by 22.00%.[6,13,31]

Methodological differences in error detection

The studies showed big differences in error detection techniques that affected prevalence rates. The Abdel-Qader study reported an 11.53% error rate using covert observation known only to pharmacy managers, much higher than the 0.001% discovered in Gao’s study based on passive electronic system data.[7,13] This means passive systems miss more errors than active methods. Several studies verified that observational techniques found 2–3 times more errors than system-generated reports, probably because human observers may spot minute flaws that automated systems overlook.[6,23] When comparing error rates, these method differences must be considered as detection directly affects prevalence.

Demographic variations in error reporting

Several studies showed that patient demographics affect error reporting. Patients aged 18–39 reported more errors than older groups. Talay and Vickers found women reported errors at 0.41%, men at 0.12%.[30] The Abdel-Qader study found pharmacies with over 60 daily prescriptions reported more errors.[13] Ibrahim et al. reported that patients caused 50.00% of errors in community pharmacies, linking this to health literacy and involvement.[25] These demographic trends highlight the consideration of patient traits in error analysis and prevention.

Temporal trends in error rates

Longitudinal data demonstrated significant reductions in error rates following technological installations. Mäkinen et al. reported a 46.00% decrease in Finnish community pharmacy errors during 2015–2020 following digitalization and the 2019 Medication Verification System launch.[8] Gao et al. likewise claimed a 98.20% drop over 14 years following the use of improved prescription verification and pharmacist training.[7] These developments were placed even with rising prescription numbers (67.1 million in 2020 statistics compared to 55.8 million in 2015), proving that technology interventions can offset concerns connected to workload. The technical barrier implementation of the Norden-Hagg trial in Sweden directly lowered dosage errors by 75.00%, implying that some interventions can yield quick changes.[28]

System-level versus human factors

Root cause studies classified errors as system (47.90% LASA in Chou et al.), human (36.00% staff weariness in Mäkinen), and environmental variables (55.00% workload pressure in Soubra).[6,8,20] While human elements ruled community pharmacies (illegible prescriptions 30.70% in Abdel-Qader), system aspects ruled outpatient environments (LASA errors in Copi’s study).[13,22] Inadequate workspaces and peak hours among environmental stresses aggravated these problems.[31] Especially, the Kenawy study found that electronic prescribing only lowered human factor errors by 1.40% while Gao’s system-wide interventions accomplished a 98.20% reduction, implying that system solutions outperform individual-focused strategies.[7,26]

Near-miss versus actual error patterns

Martínez Sánchez found that near-misses (1.80%) occurred 3.6 times more often than actual errors (0.50%).[23] Rajah et al. reported that 59.40% of near-misses occurred during filling and 40.60% during labeling in Malaysian hospitals.[29] These data suggest that near-miss analysis could detect up to 78% of potential errors before reaching patients, as shown in Martínez Sánchez’s quality control model.[23] The high near-miss ratio shows system weaknesses as well as current safeguards catching many errors.

Cost and resource implications

Several studies have revealed significant economic consequences of dispensing errors. Baranowski et al. found that discontinued medication errors cost a total of 9397 (37.15 per error), while Chou et al. calculated losses of $188,406.6 from 376 errors (501 per error).[18,20] The Copi trial showed that 60.46% of errors had the potential to cause harm, and among 379 patients who received discontinued prescriptions, 78 were hospitalized, and 3 cases were possibly admitted due to receiving discontinued medications.[22] Talay noted that 11.1% of errors were serious, reflecting both direct costs and future healthcare needs.[30] The productivity consequences are emphasized even more by Abdel-Qader’s study, which found Sundays and high-volume times as risk factors.[13]

Technology adoption thresholds

A comparative study showed the tiered efficacy of technical solutions. Integrated systems improved errors by 98.20% through better verification, pharmacist training, and monitoring, while basic e-prescribing cut errors by only 1.40%.[7,26] Targeted technical obstacles – dose verification prompts – showed 75% reductions in some error kinds without complete system overhauls, according to the Norden-Hagg study.[28] Copi’s results, on the other hand, indicated that certain pharmacy system integration still permitted 4.94% discontinued prescription errors, hence demonstrating that total interoperability is required.[22] Effective projects typically require 4–5 years to fully succeed, implying that adoption schedules significantly affect the results.

Medication-class specific risks

Some medication classes showed unequal hazards throughout the research. Baranowski reported that 71.90% of errors were associated with cardiovascular agents, while Copi et al. identified anticoagulants (enoxaparin, 12.02%) and insulins as the primary contributors to serious errors.[18,22] Talay’s investigation on GLP-1 receptor agonists (semaglutide) emphasized inappropriate medication (6.10%) and incorrect dosage (58.60%) as particularly harmful. These findings underscore the necessity of enhanced safeguards for high-alert medications, corroborated by Allen’s report showing 34% of discontinued medication errors involved drugs with narrow therapeutic indices.[17] Kiron’s LASA study further confirmed class-specific hazards linked to naming conventions.[27]

Error detection cascades

Several studies charted when and how errors were found. While Martínez Sánchez discovered staff caught 30% of problems in processing, Mäkinen found patients found 50% of faults in community pharmacies.[8,23] Unlike Abdel-Qader’s observation approach, which found staff members found problems proactively, the digital platform of the Talay research enabled 100% error documentation via patient reports. These detection mechanisms have significant consequences: patient-reported errors averaged 0.35% prevalence against 11.53% in staff-observed studies, implying both under-reporting by patients and the need for professional attention.[13,30] Timing’s influence on detection effectiveness is shown even more by the Thomas study’s discovery that 62% of errors happened during 11 AM–3 PM shifts.

Policy and workflow implications

Successful intervention synthesis produces practical policy ideas. Despite a 20% rise in prescriptions, Finland reduced errors by combining national reporting with phased technology implementation over 5 years.[8] A structure relevant across environments, Gao’s hospital put into place: (1) monthly training, (2) verification checkpoints, and (3) real-time monitoring. Thomas et al. showed that including pharmacists during peak hours helped to lower errors for workflow; Soubra reported that separate LASA medications cut errors by 18%.[6,31] For resource-constrained environments, the technical solution of the Norden-Hagg study – forced dosage verification – offers a low-cost choice. These results taken together imply that ideal methods mix technology, staffing changes, and process reengineering according to setting-specific risk profiles.[28]

Publication bias

Egger’s test showed no notable publication bias for studies on dispensing error prevalence in community and outpatient pharmacies (P = 0.599).

Certainty of evidence

Using the GRADE method, evidence certainty for dispensing error prevalence was rated very low [Table 3]. Despite 22 studies showing no major risk of bias, imprecision, indirectness, or publication bias, substantial inconsistency existed due to wide variation in reported prevalence (ranging from 0.001% to 11.53%). Methodological variations in error detection, pharmacy settings, and regional healthcare systems produced this variety. Studies from less developed countries reported higher error rates (mean 5.20%) than industrialized countries (mean 1.10%). Therefore, even with a strong research design, the data calls for very low certainty, underlining the importance of uniform measuring techniques in future studies to enhance dependability.

Table 3.

The certainty of evidence regarding the prevalence of dispensing errors in community and outpatient hospital pharmacies

Outcome	Number of experiments	Risk of bias	Imprecision	Inconsistency	Indirectness	Publication bias	Level of evidence
Prevalence	22	Not serious	Not serious	Serious	Not serious	Not present	Very low

DISCUSSION

This systematic review assesses error rates in community and outpatient hospital pharmacies, revealing significant variation and contributing factors. Most errors involved incorrect medication, dosage, and form due to high workload, staffing shortages, and medication similarities, underscoring the need for broader studies and effective error reduction strategies.[32] Pre-typed prescriptions, large medication counts, injectable medications, and specific professional teams were linked by the Brazilian hospital study to an 81.80% dispensing error rate, thus stressing systematic weaknesses in pharmacy dispensing procedures.[33] Another Australian poll of 209 pharmacists found that the majority think dispensing errors are on the rise, mostly because of heavy workload, tiredness, and ambiguous drug names. Pharmacists pushed for systematic workflow checks and regulatory limits on daily dispensing volumes as the main preventive measures.[34] James et al. reviewed 60 studies, finding that methodological variations led to significant dispensing errors in medication across environments, influenced by look-alike drugs, staffing, workload pressures, and distractions.[35] This difference may be due to variations in workflow patterns, prescription quantities, and levels of technological adoption across different settings.

Furthermore, a systematic review and meta-analysis of 62 studies revealed a worldwide pooled prevalence of 1.60% for dispensing errors across pharmacy environments, with rates differing greatly depending on denominator type, study design, and error identification techniques, therefore stressing the need for standardized definitions to lower heterogeneity in error reporting.[36] A UK study also revealed dispensing errors in 1.70% of medications and 1.60% of labels; authentication systems connected to patient data or electronic prescriptions could help to prevent 25%–60% of mild errors, while none would have stopped catastrophic errors.[37] In the end, a meta-analysis and methodical evaluation showed that programs such as staff training and dispensing technology greatly lowered hospital pharmacy dispensing errors by 34%–68%, hence stressing their efficacy in enhancing medication safety.[38]

These findings significantly impact health policy and medication safety, showing that tailored interventions based on pharmacy workload and technology are necessary. Integrated electronic prescribing, real-time monitoring, pharmacist training, and workflow changes effectively reduce errors. Data indicate that while electronic prescribing modestly reduces some errors, comprehensive system-wide interventions yield far greater improvements, emphasizing the need for a full pharmacy system overhaul rather than isolated technical fixes.

Dispensing errors, leading to hospitalizations and financial losses, cause significant economic and clinical harm, emphasizing the urgent need for solutions. National-level policy projects like Finland’s compulsory technology adoption program show how regulatory frameworks mixed with technical investment can produce continuous decreases in error prevalence even with rising prescription volumes.

Despite its strengths, this study has limitations mainly due to substantial methodological and reporting variability among included papers. Differences in error detection methods – from covert observation to passive electronic systems – affected reported error rates and reduced comparability. Many studies also lacked an explanation for sample size and outcome assessor blinding, which could lead to bias. Cross-sectional studies predominate, which restricts longitudinal knowledge and causal inference. In addition, by including only English-language publications, this review may have introduced language bias and potentially excluded relevant studies published in other languages. Moreover, although Egger’s test did not indicate significant publication bias, the exclusion of non-English language studies and grey literature – despite a thorough search – may still have introduced bias.

Although the review followed PRISMA criteria, uneven reporting and missing data in some studies complicated the process, which required author contact efforts with mixed results. Inconsistency and indirectness led to very low confidence in the evidence, highlighting the urgent need for consistent definitions and approaches to harmonize future studies.

This review highlights multiple future research avenues. Consistent global prevalence data require approved dispensing error detection and reporting systems. Longitudinal and interventional studies with strong designs are needed to clarify causes and assess technology and workflow impacts. Research should also explore patient engagement to improve error detection. Finally, cost-effectiveness studies of different error reduction programs would inform resource allocation.

To improve performance, pharmacies should implement integrated e-prescribing, combined with pharmacist training and workflow redesign, to reduce errors. In resource-limited settings, policymakers should promote technology use and national error-reporting systems. Embedding safety culture and improving medication results call for cooperative initiatives across healthcare sectors.

The current study partially aligns with El Hajj’s review but differs in key aspects. While El Hajj included prescription and OTC drugs and various error types (prescribing, administration, and dispensing), our study focused only on dispensing errors of prescription drugs. El Hajj reviewed 73 studies without language limits; we included 22 English studies. The periods differ: 1995–2023 versus 2010–June 2025. Both found “incorrect dose or strength” common, but only El Hajj noted “wrong drug selection.” El Hajj focused on community pharmacies; we included community and hospital outpatient settings. Unlike El Hajj, we examined contributing factors and consistent error rates. In sum, El Hajj gave a broad overview of medication errors; we provide a focused analysis of dispensing errors with detailed prevalence, causes, and interventions.[39]

CONCLUSION

This systematic review provides a comprehensive synthesis of dispensing error prevalence, causes, and mitigation techniques in community and outpatient hospital pharmacies. It highlights the complexities of dispensing errors, emphasizing human and system factors. The study calls for integrated technological solutions, enhanced pharmacist training, and workflow changes to effectively reduce errors. Its findings advocate for systematic improvements with uniform reporting and continuous quality initiatives. Future research should focus on standardizing detection methods and involving patients in error detection to enhance medication safety and improve healthcare outcomes.

AUTHORS’ CONTRIBUTIONS

N. Madani actively participated in searching databases and shaped the initial draft of the manuscript. N. Madani, M. Farasatinasab, and A. Eshraghi are conducting formal analyses and investigations essential for interpreting the findings. M. Farasatinasab, A. Eshraghi played a pivotal role in conceptualizing the study, contributing to developing research methodologies. Additionally, A. Eshraghi meticulous review and editing efforts enhanced the clarity and coherence of the manuscript. H. Aslani contributed considerably to the search databases and technical aspects of the study. All authors have approved the final submission.

Conflicts of interest

There are no conflicts of interest.

Funding Statement

Nil.