![[ Visit Client Website ]](images/banner.gif)
As harm is ultimately correlated with the amount of medication errors that occur in total, the exploration of the incidence of medication errors within a specific setting is the first step in measuring the harm caused by these errors. Internationally, the incidence of medication error was found to vary considerably from setting to setting, with an incidence as low as 1.2% of administered medications and as high as 291 errors in 168 observed intravenous doses while the median medication error rate as derived from 91 international studies was found to be 19.6% of total opportunities for error (Keers et al.,2013:237). However, this rate was proposed to be even higher in developing economies such as South Africa (Bates, 2010:174). Although South Africa has no current statistics available regarding the incidence of medication administration errors, 105 of 629 professional nurse misconduct cases between 2003 and 2008 were related to medication administration (South African Nursing Council [SANC], 2013:1).
For this reason, the purpose of this research was to determine the incidence and types of medication administration errors and medication-administration-error-related deviations from safe practice in medical and surgical units of public hospitals in the Gauteng Province of South Africa.
Methods: The incidence of medication administration errors was determined by direct observation. The observational method implemented in this study was naturalistic observation, as the researcher tried to be as inconspicuous and unobtrusive as possibly, passively recording what occurred while not modifying the behaviour occurring ordinarily in the natural setting (Gravetter & Forzano, 2012:369). The natural setting where the observations occurred was the medical and surgical units during medication administration rounds. Specific behaviour recorded was the method of administrating medications, with specific notes on the occurrence of medication administration errors and deviations from safe practice.
A check-list was adapted from the check-list used by Kim and Bates (2013:591) was used during direct observation. The original tool was structured around the five rights of medication administration (right medication, right dose, right patient, right route and right time), adherence to basic infection control principles and recording. The checklist by Kim and Bates (2013:591) consisted of positive statements, such as “label the medication immediately after preparation”. However, the researcher chose to adapt this check-list to rather reflect the errors or deviations that did occur, thus the statements were changed to the negative, example: “Medications were not labelled immediately”. This was done to prevent confusion during analysis, as the same headings could be reflected in the report. A space for indicating the rank of the medication administrator, the amount of different medications prescribed to the specific patient and the amount of interruptions occurring during the administration to the patient was added. Furthermore, omissions were added as possible error.
Eight public hospitals within the Gauteng Province that met all the inclusion criteria were selected randomly. One medical and one surgical unit from each of these selected hospitals were sampled randomly, while all the medication administrators in the selected units on the day of data-collection were included in the study. Ten parenteral and ten enteral medication administrations were observed in each unit (n = 315). All sampled medication administrators gave informed consent to be observed and tested.
Statistical analysis in the form of frequencies of errors was performed. P values (statistical significance derived from t-tests) and effect sizes (practical significance derived from Cramer’s V and correlations) of relationships between medication errors and acuity; staffing levels; occupancy; interruptions; unit type; administration route; hospital level; and the rank of medication administrator were used to obtain insight into these relationships.
Cramer’s V was calculated to determine the effect size (practical significance) between incidence of medication errors or deviations from safe practice and type of unit, level of hospital or administration route. Correlations were calculated by means of the SAS software between incidence of medication administration errors or deviations from safe practice and patient acuity, staffing levels, percentage occupancy, interruptions and rank of medication administrator, taking into account the dependency of data on individual hospitals. Odds ratios were calculated for correlations with either practical or statistical significance, taking into account the dependency of variables within hospitals.
Results: 296 errors were identified, of which wrong-time errors (n = 127, 43%) and omissions were the most common (n = 122, 41%). A further 33 patients (11%) received a bigger or smaller dose than had been prescribed, while seven out of 315 patients (2%) received the wrong medications. Medications were administered via the incorrect route to six patients (2%), while only one wrong-patient error was observed. A statistical significant correlation with medium effect was determined between interruptions and wrong dose errors (OR = 2.56; p <0.05). Patient acuity was practically and statistically correlated with wrong dose errors (OR = 10.55; p <0.05).
Conclusions: Medication administration errors pose a great threat to patient safety in public hospitals in the Gauteng Province. Both similarities with and differences to international literature were noted, which led to the need for an intervention that is developed with this specific setting in mind.
See more of: Research Sessions: Oral Paper & Posters