Prevalence and Nature of Medication Administration Errors in Health Care Settings

A Systematic Review of Direct Observational Evidence

Richard N Keers MPharm; Steven D Williams MPhil; Jonathan Cooke PhD; Darren M Ashcroft PhD


The Annals of Pharmacotherapy. 2013;47(2):237-256. 

In This Article


To our knowledge, this is the first multinational systematic review of the prevalence and nature of MAE in a range of health care settings. We found that MAEs remain a common threat to patient safety, occurring with a median and interquartile range of 19.6% (8.6–28.3%) of TOE with timing errors and 8.0% (5.1–10.9%) without timing errors using the OME numerator, and a median (IQR) of 25.6% (20.8- 41.7%) of TOE with timing errors and 20.7% (9.7- 30.3%) without timing errors using TNE.

The intravenous route of administration appeared to be associated with a higher error rate than when all administration routes were studied (median MAE rate 48.0% vs 8.2% minus timing errors using OME, and 53.3% vs 20.1% using TNE). A Bayesian analysis found a 73% overall probability of making at least 1 error in intravenous therapy[123] and some[73,87,93,96,102,107] but not all[62,81,85,86,89,92] of the studies included in this review found that the intravenous route was associated with greater MAE rates/risk than other routes. Others have acknowledged the difficulty in comparing intravenous MAE rates because of differences in study methodology.[32] However, agreement is universal in that the intravenous administration requires much attention in ME reduction strategies, as it is associated with considerable complexity and patient risk,[26,32,95,124,125] with the percentage of harmful intravenous MEs being greater than that of all harmful errors combined in 1 incident report review.[126]

Although there are several different denominators used in ME research,[8,10] total opportunity for error is well established as the most frequently used measure by observation-based MAE studies (explicitly used by 62% of studies). Within each OE, a dose is typically described as being correct or in error,[7] although there are examples where multiple errors per OE have been reported.[69,75,81,83] As direct observation

MAE research focuses on a snapshot of the medication administration process, the collected data often preclude calculation of error rates in different forms, such as per patient day.[118]

The observed variability in error rates, as demonstrated by wide range and overlap in the IQRs both within and between denominators and numerators, may be attributed to a umber of factors. Error rates varied between the institution studied,[17,32–34,43,44,50,58,89,91,92,101,110,112] between units within one institution,[44,49,53,54,60,62,68,75,80,82,85,104,113] and at the level of the individuals being observed.[68,72,83,86] Some of this variation may be explained by the presence of different medication distribution systems and specialties of the units studied within each institution; for example, critical care settings may operate under disruptive conditions[53] and involve frequent intravenous administrations, which may affect MAE rates.[89]

A number of studies chose not to include WTE in their definition or considered them separately from other MAEs. Presenting a median MAE rate without timing errors does not portray an entirely accurate image of the systems in which MAEs occur.[7] Wrong-time errors can be viewed as a product of systems failures such as workload,[47] which in themselves may drive institutional changes to minimize MEs. Timing errors were by far the most common MAE subtype observed in this review, which may therefore indicate a widespread systems issue which is multinational. Errors associated with time-sensitive medications have recently moved to the forefront of health care practice due to their potential for patient harm,[127,128] and along with addressing the wider timing error issue these error types should form a focus of future research interventions.

Omission and wrong dosage errors were also found to be commonly occurring MAEs. Another, albeit smaller, review of observational MAEs in critical care settings found similar results,[9] as did a review of prescribing errors in hospitals,[8] which found that dosage errors were most commonly reported by the majority of studies. Wrong or unclear dosage/frequency errors, along with omission/delayed medication and wrong medication, make up the 3 error subtypes associated with 71% of MEs causing death or serious harm from incident reports in the UK.[129] The US has also identified omission and wrong dosage errors as the most common MEs from incident reports.[130]

Wrong time, administration rate, and preparation errors were among the 3 most common MAE subcategories observed with regard to intravenous administration (when WTE were studied). Intravenous medications may require complex multistep preparation processes[26] and can be administered over prolonged periods, which may lead to additional opportunities for error when compared to other routes of administration. Some studies considered intravenous errors to include poor aseptic technique,[33,67,69] which may have contributed to greater numbers of preparation errors and to greater MAE rates in general. As separating aseptic errors from other preparation errors was not always possible,[96,107] a matter for future debate should perhaps be whether a distinction should be made between them. Other potential contributory factors to intravenous MAE include poor knowledge of equipment associated with this route of administration.[131]

Due to the disparity in method of severity assessment, a detailed comparative analysis between studies was not possible. Others have found similar difficulties in this type of ME analysis.[8] Errors resulting in no harm or requiring monitoring/intervention to reclude harm were most commonly observed by studies using NCC MERP criteria,[119] although harm requiring intervention or hospitalization was observed for a smaller proportion of errors. The proportion of MAEs leading to actual or potential harm varied depending upon the criteria used, thus highlighting standardization of severity assessment as an important future research goal.

Medications affecting cardiovascular, nutrition and blood, and the gastrointestinal systems were most frequently reported as being in the 3 most commonly observed medication groups. These 3 were closely followed by medications falling into the groups' central nervous system and antiinfectives. Similar results have been found in reviews of MAEs[9] and prescribing errors.[8] Given that many of the medications in these groups are considered time sensitive,[127,128] future initiatives could focus on reducing the incidence of errors (in particular, timing errors) in these medication groups.

Those responsible for administration of medication highlight a multitude of causes behind or factors associated with MAEs, ranging from individual failures such as working knowledge and concentration to more organizational deficiencies such as unclear/impractical procedures and heavy workload/understaffing.[16,131–134] Numerous studies have evaluated specific interventions aimed at known causes/factors behind MAEs. Bar-code assisted medication administration (BCMA) could reduce misidentification (patient, medication) failures[53] and has been reported to reduce the number of wrong-time[52,57] omission and wrong dose errors,[45] although these results are often conflicting.[24] Others have found that users of BCMA may find ways to circumvent its safeguards, leading to novel MAE risks.[135] Generally positive results were obtained using educational packages designed to increase working knowledge.[46,54,70,93,97] Computerized physician order entry can help minimize communication- and knowledge-based errors,[65,98,136] but does create additional ME challenges.[136] Use of different drug distribution systems has shown some improvements in MAE rates.[81,83,88] Suggestions have been made on how to minimize intravenous MEs[26,67,95] and some have demonstrated that implementation of a medication administration protocol reduces intravenous process errors.[137] Although these various interventions may reduce the incidence of certain types of MAEs, they may not be alone sufficient to minimize omission, wrong time, and wrong dosage errors, and further research could consider a multifaceted approach to error rate reduction.[24,93,131]

Important strengths of this review were the time period and number of electronic databases searched, which spanned over 25 years. Hand-searching of reference lists of included studies ensured that the risk of eligible studies being missed by the electronic searches would be minimized. However, as only publications in English were included, relevant articles printed in other languages could have been missed. The impact of the Hawthorne effect on the validity of direct observation data must also be considered, whereby the presence of the observer may cause those being observed to become more careful or error prone.[16,138] However, provided that the observer is unobtrusive and nonjudgmental,[138] participants often return to regular behavioral habits after they become accustomed to being observed.[26,84,138] Observer inference is another factor that may affect the validity of MAEs detected, although appropriate training may help to minimize this effect.[7] Studies that have tested interrater reliability between observers have shown consistency if thorough training was provided,[13,35,43,53,112] although these results have not always been noted for other ME types.[139]