Magnitude of Medication Administration Error and Associated Factors Among Nurses Working in Madda Walabu University Goba Referral Hospital, Bale Zone Oromia Region, Southeast Ethiopia

Background: Administration of medication is the primary responsibility of nurses. Medication errors occurring during the drug administration process can be attributed to a variety of safety effects, ranging from undetected errors to prolonged hospital stays, discomfort and death. Objective: To determine the magnitude of the medication administration error and associated factors among nurses working at Madda Walabu University Goba Referral Hospital, Bale Zone Oromia Region, South East Ethiopia. Methods: A facility-based cross-sectional study was conducted at Madda Walabu University Goba Referral Hospital Inpatient Department from February to March, 2020. The study included three hundred ninety-eight medication interventions administered by 89 inpatient unit working nurses during the study period. Data were collected using a pre-tested, structured questionnaire and drug administration assessment using a checklist. Data were analyzed using SPSS version 22 and Frequency , Percentage, Means and SD were analyzed for descriptive analysis. COR and AOR were calculated to see the association of independent variables and uncontrolled hypertension at 95% CI and p-value <0.05 was considered statistically signicant. Result: The magnitude of the medication error was 248 (62.3%). The most common type of medication error was wrong documentary evidence (53.5% ) , followed by wrong time (39.2%) and wrong dosage (28.%). Variables that were substantially associated with medication administration error include work experience of nurses 0-4 years (AOR = 10.8, 95% CI (4.5-25.86), 5-9 years of service (AOR = 4.05, 95% CI (1.47-11.715), nurses 1-6 (AOR = 0.36, 95% CI (0.17-0.76) nurses 7-10 (AOR = 0.45, 95%CI (0.21-0.96) route IV of medication (AOR =0.13, 95 % CI (0.03 - 0.60) and IM route (AOR =.0.12, 95 % CI (0.02 -0.74) at p-value <0.05. Conclusion: Medication administration error was highly prevalent. Work experience, nurse to patient ratio and route of medication administration were statistically signicant effect


Background
According to the United States (US) National Coordinating Council for Medication Error Reporting and Prevention (NCCMERP) in 2017 de ned medication error as any preventable event that may cause to inappropriate medication use or patient harm while the medication is in the control of the health care professional, patient, or consumer (NCC-MERP, 2017). Medication errors have become common causes of harm to patients with up to 6.5% of patients in clinical settings affected and it causing serious public health threats (Alsulami & Zayed, 2013).
Medication administration errors are a common subtype of medication error and one of the most common types of adverse events in hospital-accepted patients and the most common cause of preventable death (WHO, 2018; Feleke, Mulatu & Yesmaw, 2015). The United States National Coordinating Council for Medication Error Reporting and Prevention (NCCMERP, 2018 ) found that 38% of the medical error occurred in the hospital during the medication administration process (NCCMERP, 2018). Patient morbidity and mortality are the major consequences of these medication administration errors (MAEs). It may also have cost implications for patients, families and health care providers due to prolonged hospital stays and psychological impact since errors undermine public con dence in health care services (Baraki et al., 2018).
Medication administration errors are usually committed by nurses because the primary responsibility of the nurse is to administer the medication, with between 16% and 40% of the time spent by nurses in MA Miscarriages of all kinds are well examined throughout much of the developed world and the outcome has shown that the magnitude of MAEs among nurses is still high (Adam, 2018;WHO, 2018). In developing countries, especially in East Africa, MAE is common and the error rate ranges from 9.4 to 80% of all medication administrations (Alsulami, Conroy & Choonara, 2013). Similarly, it has also been reported in Ethiopia that the prevalence of MAE is still high. For example, studies in 3 public hospitals in Tigray and 3 tertiary hospitals in Addis Ababa have shown 62.7% and 68.1% prevalence, respectively (Baraki et al. 2018, Adam 2018). These mean that MAE is a real concern and a real threat to patient safety. However, the issue has rarely been investigated and scarce in Ethiopia (Mekonnen et

Statement of the Problem
At every step, trained nurses have the potential to make a mistake in their everyday nursing practices, with some degree of danger to patient safety. The complexity of the medication administration process is such that errors can occur at one, several, or even at all stages between prescription and administration (Rodriguez-Gonzalez etal., 2012).
Medication errors are the most common type of medical error occurring in hospitals and ranked among the top ten causes of death worldwide . Medication errors, especially those made during the administration of drugs, were highly prevalent. Each medication process and each patient had at least one type of Medication administration (Feleke, Mulatu & Yesmaw, 2015).

Methods And Materials
Study area and period The study was conducted at Madda Walabu University Goba Referral Hospital, located in the Bale Area of the Oromia Province, in the South East of Ethiopia. It is located 445 km away from the capital city of Addis Ababa in Ethiopia. MWUGRH is the only major referral and teaching hospital in South East Ethiopia. The hospital has four main departments, namely: Pharmacy, Surgery, Obstetrics and Gynacology, and Pedatrics with more than 160 beds. The hospital provides services to approximately 20,000 inpatients and 150,000 outpatients each year. There are 108 General Practitioner Physicians, 15 Specialist Doctors with different specialties, 130 nurses, 120 other health professionals of different Medication error has a signi cant impact on patients in terms of morbidity, mortality, adverse drug event, and increased length of hospital stay. It also increases costs for physicians and healthcare systems (Popescu, Currey & Botti, 2011). In the United States, for example, between 44,000 and 98,000 hospitalized patients die annually from preventable medical errors, with more than 7,000 deaths due to drug errors (Cohen, 2007). In addition, the cost of medication error in England is 98.5 million pounds per year (Elliott et al ., 2018). Similarly, Dabaghzadeh, et al. ( 2013) in a study of drug error in a large teaching hospital in Iran showed that 19% of drug administration contained at least one mistake. 44.5% of nurses made the most reported mistakes, mostly during the drug administration process (63.6%). Most of these drug-related errors, leading to poor patient recovery, decreased client satisfaction and death, and other forms of harm to patients (Redley & Botti, 2012).
Patients living in developing countries experienced as twice as many drug-related problems as those living in developed countries (WHO, 2017). According to a systematic review of adverse drug events and medication failures in African hospitals, 8.4% of patients experienced any alleged adverse drug incident at hospital admission, while 2.8% of patients admitted to hospital due to adverse drug events (Mekonnen, Alhawassi, McLachlan & Jo-anne, 2017). In general, medication administration errors attributed to being the most common cause of disability and death throughout the world (Feleke, Mulatu & Yesmaw, 2015).
However, there are only a few relevant studies on drug administration errors in developing and transitional countries, particularly in Africa. In developing countries such as Ethiopia with educational, economic and skilled labor problems, the issue is one of the least studied and neglected health problems (Adam, 2018). Therefore, assessing the magnitude and associated factors of MAE will produce information that can be used by program managers and stakeholders in the planning and interventions of medication administration errors. As far as the knowledge of the researcher, the magnitude of medication administration error and contributing factors in the study area not studied. Moreover, since the researcher have been working in Madda Walabu University Goba Referral Hospital, observing medication administration errors made by nurses in many instants and patients and families of patients also usually compiling the actuality of the error. Hence, this study was aimed assess the magnitude and contributing factors of medication administration error among nurses working in Madda Walabu University Goba Referral Hospital of Bale Zone Oromia Region, South East Ethiopia. quali cations and experiences and 486 supporting staff during the study period (Goba Referral Hospital Human Resources O ce, 2020). In addition, Madda Walabu University Goba Referral Hospital provides a number of community services, including emergency 24-hour pharmacy and delivery services. It also offers physiotherapy, counseling, antennatal treatment, TB and ART services. The study was conducted from 30 February to 12 March 2020.

Study design and population
Institutional cross-sectional research design was used. All nurses working at Madda Walabu University Goba Referral Hospital were the source population for the research. Both nurses working in the Madda Walabu University Goba Referral Hospital and engaging in the medication administration in the inpatient department during the study period were members of the study population. Nurses speci cally involved in the administration of medications and operating in the inpatient department at the time of the study were involved. Nurses that are on an annual and parental leave and/or sick at the time of study, undergoing onthe-job training and nurses who are not speci cally involved with inpatient care were excluded.
Sampling procedure and sample size A total of 398 medications which were administered by 89 nurses to patients in the inpatient department at the time of the study were the sample size of the study.

Study variables
The dependent variable for the stdy was an error in the administration of medication.
The independent variables were Socio-demographic characteristics (sex, age, marital status, educational status, and where nurses attend their previous educational), work-related factors (work experience, on job refreshment training, period in the clinical unit, route of medication), institutional factors (availability of reporting systems of medication errors, availability of written guidance for medication administration, availability of marked medication shelf, availability of medication preparation room) and environmental factor (interruption while preparing or administering medication, change of medication administration).

Conceptual de nition of terms
In order to have common understanding the following de nition of terms were used Wrong medication: medication administered to the patients that were not on the patient's medication chart (Feleke et al., 2015).
Wrong dose: when the medication dose, strength or quantity given is different from that of prescribed (Baraki et al., 2018).
Wrong time: administration of medications 30 minutes earlier or later from its scheduled administration time (Feleke et al., 2015).
Wrong route: the actual route of medication administration differs from the recommended route of medication administration (Feleke et al., 2015).
Wrong patient: when a medication of one patient is wrongly given to another patient (Baraki et al., 2018).
Documentation error: medication that is administered to the patient but not documented in medication administration record sheet (Adam, 2018). Or medication that is administered to the patient incorrectly or incompletely documented in medication administration record sheet.
Unauthorized drug error: Medication administered was not authorized by the prescriber (Feleke et al., 2015).
Missed drug error: failures to administer a prescribed medication while the drug available at the patient bedside (Feleke et al., 2015).

Data collection instruments
In order to achieve the study objective and to gather the relevant data, an observational checklist and a structured self-administered questionnaire were used. The instruments were adapted from already existing studies ( The observational checklist and self-administration questionnaire were pre-tested on 20 nurses and recti ed on 40 doses prior to the main data collection at Robe Regional Hospital, which is not included in the analysis and the constraining factors. Cronbach's alpha coe cient of the data collection instrument was 5-007 = 0.83.

Data collection procedures
In data collection, six data collectors (BSc nurses) with prior experience in data collection and two supervisors were recruited. Prior to the actual work, intensive training was given to data collectors and supervisors for two days on the study objective, questionnaire and checklist style, evaluation procedures and reporting methods to supervisors and principal investigators. Written documentation has been prepared and delivered to data collectors and supervisors to ensure that everyone provides the same direction and information. Data collection took place from 30 February to 12 March 2020.
Data on the administration of drugs were collected through a self-administered standardized questionnaire and observation checklist by personally observing nurses in charge of the administration of drugs. In addition, after observation, the data on the recorded observation compared to the physicians are ordered by reference to the patient chart. After ensuring their willingness to participate in the study, the self-administered questionnaire was distributed among the study population. The anonymity of the participants was maintained by informing them not to enter their names on the questionnaire.

Data Quality Assurance
To ensure the quality of the data, the questionnaire was adapted from the previous study and four experts (two clinical nurses and two nurse academics) revised the tools to check their validity. In addition, the questionnaire and observation checklist were pre-tested at Robe General Hospital prior to the actual data collection. Data collectors and supervisors have been recruited on the basis of their experience in data collection and research. In addition , the main investigator was trained for two days on the objective of the study, instrument and data collection procedures.
Health professionals (in rmaries) who were observed during the administration of drugs to each patient were informed of the work prior to the start of data collection, but the entire purpose of the study was not disclosed to ensure that the ndings were impartial. The data collectors checked the completeness of the questionnaires and the checklists, and nally the supervisors and the main investigators checked the questionnaire. During the data collection period, the supervisors and the main investigators followed up on a continuous basis.

Data analysis
After the questionnaires and completed observation checklists were returned, reviewed for completeness and marked, SPSS version 22.0 was used to enter, clean and analysed data. Descriptive statistics such as frequencies, percentages, mean and standard deviations have been calculated and de ned. To see the relationship of dependent and independent variables, a bivariate and multivariate logistic regression analysis has been carried out. The odds ratio with 95% CI and p-value at 0.05 is used to declare statistically signi cant association. The results of the study were presented using statistics, tables, graphs and texts.

Ethical considerations
Ethical clearance was obtained from Madda Walabu University Goba Referral Hospital School of Health Sciences Department of Nursing Ethics Committee. The o cial letter was sent to the hospital. A letter of approval was then received from the Chief Clinical O cer of the Hospital. The study was conducted after permission was granted by the hospital administrators. In addition, all research participants have been advised of the intent of the research and their right to participate and have the right to terminate at any time if they do not wish to continue the study without violating any of the bene ts they may receive. Respondents were assured of con dentiality of the information obtained. The respondents' verbal consent was gained by asking whether or not they would like to participate. Respondents were not required to mention their name and other personal identi ers, such as telephone numbers.

Results
Socio-demographic characteristics of nurses observed A total of 89 nurses were reported for this study while administering medications. The average age of the nurses was 30.9 years with ± 6.5 SDs. More than one-third of them (39.3%) were aged 25-29 years. More than half of the 54 (60.7%) nurses were male. In terms of educational quali cation, 62 (69.7%) of them had a Bachelor of Science degree in nursing. Eighty-one (91%) of the nurses surveyed had graduated from government teaching institutions. Approximately half of them (48.3%) had work experience of less than ve years (Table 1).   (Table 3). Types of the observed drugs and their administration More than two thirds of 278 (69.8%) of the 398 drug administrations observed were Antibiotics. As far as the route of drug administration is concerned, 339 (85.2%) of the drugs observed were administered through the Intravenous Route followed by the Intramuscular Route 33 (8.3%). More than half of the 237 (59.5%) drugs were administered twice daily with respect to the frequency of drug administration (Table 4). Magnitude and type of medication administration error Eight types of drug error were observed in this study, namely: wrong time, wrong patient, wrong dose, wrong route, unauthorized drug, missed drug, wrong route, and incorrect documentation that occurred at the time of the nurse's administration of the medication to the patient was used to identify a drug error.
Medication administration error was recorded when a single or combination of the above mentioned errors occurred while administering the medication(s) to the patient.
The prevalence of drug error from a total of 398 drugs administered was 248 (62.3%) (Fig. 1). Out of 248 medication errors committed, 61 (25.4%) had only one type of error, 171 (68.9%) had 2 types of error, and 16 (6.5%) had 3 and more than 3 types of drug error. The most widely perpetuated form of drug error was incorrect documentation 213 (53.5%) followed by incorrect time 156 (39.2%). Incorrect dose administration is the third common 112 (28.1%) drug administration and error (Table 5; Fig. 2). These errors were committed by 72 (80.8%) nurses from 89 nurses who participated in the study. Of these, more than half of the 38 (52.8%) nurses made a mistake at least three times during the stated period.  Table 6 Types of medication administration errors observed in the inpatient departments of Madda Walabu University Goba Referral Hospital, South East Ethiopia, 2020.

S. No
Some observed examples 1 The observer observed while Insulin given intradermal instead of subcutaneous route.

2
Most medications in the morning were being given after 7:00 am instead of 6:00 am 3 The observer observed a nurse while she was giving Plasil IV instead of Tramadol IV 4 Most of the nurses did not document after administration of the drug.

5
The observer observed while Ampicillin was given to a 28 years old although the order sheet reads as Cloxacillin. 6 Metronidazole IV medication mostly missed at 2 PM (especially during the weekends) 7 Instead of administering 2 g of Ceftriaxone, the nurse administered 1 g.
Usually at paediatric ward, the nurse did not calculate the exact doses of medication. 8 The observer observed while Ceftriaxone was given to an 43 years old male even after the physician ordered to discontinue the medication

Factors Associated With Medication Administration Errors
Bivariable logistic regression was used to determine the correlation of each independent variable to the dependent variable. In this study, the variables that were signi cantly associated with the dependent variable in Bivariable logistic regression were the work experience of nurses, period in a speci c unit of nurses per month, medication preparation, nurse-to -patient ratio, availability of medication guideline and route of medication administration.
One of the variables that demonstrated association with MAE in bivariate logistic regression was the nurse's work experience. Nurses with work experience < 5 years and 5-9 years were 23.6 times COR = 23.6, 95% CI (11.60-48.052) and 5.9 times COR = 5.9, 95% CI (2.59-13.68) were more likely to make a medication administration mistake compared to those with work experience > = 15. Nurses who worked in a speci c ward for a period of 0-6 months is 6.7-fold COR = 6.7, 95% CI (3.37-13.37) more likely to make a medication administration error compared to those who worked in a speci c ward for a span of more  To control the effect of the confounding variables, the variables which showed association in bivariate logistic regression at p-value < 0.05 were taken into multivariate logistic regression. Multivariable logistic regression analysis identi ed the work experience as positively associated, while nurses identi ed the patient ratio and route of administration as negatively associated with the dependent variable.
Nurses with work experience of 0-4 years and 5-9 years were 6.0 times (AOR = 6.0, 95% CI (4.5-15.08) and 4.05 times (AOR = 4.05, 95% CI (1.47-11.715) higher than those with work experience of more than 15 years. Another factor that showed association with MAE was the ratio of nurse to patient. Nurses with a nurse-to -patient ratio of 1-6 and 7-10 were 64% (AOR = 0.36, 95% CI (0.17-0.76) and 55% (AOR = 0.45, 95% CI (0.21-0.68) less likely to make an error compared to nurse-to -patient ratios of > 10. Route of drug administration was also found to be one of the predictors of MAE. Medications administered via IV route and IM route were 87% and 88% less likely to have MAE than those administered via SC route with (AOR = 0.13, 95% CI (0.03-0.60) and (AOR = .0.12, 95% CI (0.02-0.74) respectively (Table 8). Thus, the possibility of making an error minimized with better work experience because they may get much familiar with different types of medications.
This study also showed that nurse to patient ratio was signi cantly associated with medication administrations errors. Nurses working in situation where the nurse to patient ratio of 1-6 and 7-10 were less likely to make an error as compared to nurses working in situation where the nurse to patient ratio greater than 10. This nding agrees with the study conducted in Ghana tertiary care hospital that shows the occurrence of MAE was signi cantly associated with number of patients under the nurse's care (Franklin Acheampong ARTaBPA, 2015). Similarly, a study carried out in Felege Hiwot Referral Hospital revealed that nurse to patient ratio of 7-10 and nurse to patient ratio greater than 10 was found to be signi cantly associated with medication administration error (Feleke, Mulatu & Yesmaw, 2015). This can be explained by the fact that as workload or number of medication administered by nurses increases nurses are usually exhausted and the possibility of making an error would be high.
Furthermore, in this study route of medication administration was found to be one of the predictors of MAE. Medications, which were administered through IV route and IM route, were less likely to have MAE than a medication which was administered through SC route. This nding is partly supported by studies conducted in France that showed medication errors were signi cantly associated with administration route (error was higher for administrations by injection) (Berdot et al., 2012).

Conclusion
The magnitude of medication administration error in Madda Walabu University Goba referral hospital was highly prevalent. This show that Nurses most often make medication administration error during administering medications in the patient care unit, Documentation error followed by wrong time and wrong dose was the most common types of medication administration errors. Nurse's work experience, nurse to patient ratio and route of medication administration were signi cantly associated with medication administration. Nurses' capacity building and supportive supervision were required to avert the medication administration errors.
Limitation of the study Potential care providers bias during data collection using observational method (social desirability bias). And as the study was cross-sectional it does not allow to conclude cause and effect relationship between the variables. Then a letter of permission was secured from the Chief Clinical Director O ce of the hospital. The study was conducted after the administrators of the hospital gave their permission. Moreover, all participants of the study were informed about the purpose of the study and their right to participate and have the right to terminate at any time if they don't like to continue in the study without infringement of any of the bene t they could get. Respondents were assured about the con dentiality of information collected. Verbal consent of the respondents was obtained by asking whether they want to participate or not. The respondents were not required to mention their name and other personal identi ers such as phone number.

Consent for publication
All the information used for this study was collected with the consent of participates of the study. And all the authors have read the manuscript and have consented to publish it this journal.
Data availability