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

Disclosures

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

In This Article

Literature Search

Search Strategy

Ten electronic databases were searched (all conducted by RNK): MEDLINE, EMBASE, International Pharmaceutical Abstracts, Cumulative Index to Nursing and Allied Health Literature, Scopus, Health Management Information Consortium, PsycINFO, Cochrane Reviews and Trials (all 1985-May 2012), British Nursing Index (1994-May 2012), and Applied Social Sciences Index and Abstracts (1987-May 2012).

Search terms used fell into 3 groups: error definition (including error, medication error(s), incident report, near miss, drug error, treatment error, medication safety, drug safety, preventable adverse event, adverse event, medical error, clinical incident, adverse drug event, medication incident), epidemiology (rate, prevalence, incidence, epidemiology), and error type (including drug/medication/medicine administration, dose/drug/medicine/medication preparation, drug/medication/medicine delivery, omission, drug utilization, commission, drug/medication/medicine supply, drug/medication/medicine handling, self medication, self administration), and were broadly similar between databases. The reference lists of included studies were hand-searched to identify additional articles. Relevant reviews identified from the electronic databases were searched for additional studies.

Inclusion and Exclusion Criteria

Studies were included if they were published (including early online) in English between 1985 and May 2012 and reported on the rate of medication administration errors derived only from direct observation. Studies needed to be conducted in long-term care or inpatient hospital settings; those carried out in patients' homes were excluded. Review articles were excluded, although relevant reviews had their reference lists hand-searched for additional studies. Conference abstracts that did not provide sufficient information to determine the prevalence and nature of MAEs were excluded, as were studies that investigated MAEs associated with 1 drug, 1 drug class, or 1 type of medical device; reported only 1 subcategory of administration error (eg, omission error) or derived error rates from simulation. Studies that provided data only on the number of clinically significant MAEs (rather than all MAEs) were also excluded, as they would not reflect the overall MAE prevalence.

Data Extraction Method

The following details were extracted from each study: year of publication, country of origin, study period, study type, setting, sampling strategy, drug administration route(s) studied, details of observers and observer training, subject details, definitions, error rate, error subcategory prevalence, severity ratings, and medications commonly associated with error. Data were extracted by RNK, SDW, and JC; any differences were resolved by consensus between all authors.

Medication Administration Error Definition

A medication administration error was defined as a deviation from the prescriber's medication order as written on the patient's chart, manufacturers' preparation/administration instructions, or relevant institutional policies.[7,26] Ward level medication dispensing/preparation errors were considered administration errors; prescribing and pharmacy dispensing errors were excluded.

Data Analysis

Studies identified for inclusion from the electronic search showed a high degree of heterogeneity; as a result, median error rates with interquartile ranges (IQRs) were calculated. Separate median error rates were calculated, using error rates including and excluding wrong-time errors (WTE). For studies reporting separate error rates for administration, ward-based preparation/dispensing and/or storage, data were amalgamated to form one overall administration error rate for comparison with studies that considered preparation/dispensing or storage but as part of the overall reported error rate.

Studies that evaluated the impact of a new intervention using a before and after method had only their preintervention data extracted for analysis. This also applied to studies that compared a new intervention with standard practice by evaluating both systems concurrently. For studies that concurrently evaluated different systems in which it could not be ascertained whether one system was newly implemented, an overall error rate was reported by combining each group's data. An overall error rate was applied to studies carried out over multiple sites but within 1 country, regardless of whether they used different systems. Articles that reported multinational data were separated by country to form multiple unique studies that were considered independently.

The TOE (defined as either the total number of doses given, whether correct or incorrect, plus omitted doses, or as the total number of doses scheduled plus any extra doses given) was the rate denominator used for comparison between studies.[7] If possible, enominators were converted to TOE as it was the most frequently used denominator. Denominators not specified as TOE were considered as such if they included omission errors and extra dose errors unless otherwise stated by the authors. Numerator data were presented as both the number of doses considered to have 1 or more errors (OME, ie, each dose can be only correct or incorrect), and the total number of errors (TNE, ie, more than 1 error per dose could be counted, potentially resulting in MAE rates above 100%). Any study reporting rate data considered to be based on the TOE denominator was also considered to use OME, according to the TOE definition,[7] unless otherwise stated. Studies that counted more than 1 opportunity for error (OE) per dose or reported drug preparation and administration as separate steps were excluded from the median calculations. Comparisons were made between studies by categorizing the data according to country of origin, demographics of patients studied, route of administration, and study setting.

Studies that evaluated error subcategories were considered in 2 groups: 1 including studies that provided timing error data and 1 that did not. For each group, the 3 most common error subtypes from each study were recorded. To facilitate comparisons between data, these error subcategories were based on the American Society of Health-System Pharmacists' (ASHP's) definition of a medication error,[27] with additional categories of wrong documentation (placed in other error category, and including ward-based labeling errors) and aseptic technique and storage errors (each placed in wrong preparation category) if the authors considered them to be MAE. Incorrect manipulation of dosage forms (eg, crushing tablets) was included as an MAE. If a study reported more than 1 definition of a timing error, ±60 minutes was used in the analysis, where possible.

Studies reporting severity outcomes were grouped according to the criteria used. Medication classes associated with MAE were categorized based on an existing framework.[28]

processing....