Incidence and prevalence of medication errors

Medication errors which are made during prescription, dispensing and administration [14] are common and preventable causes of patient harm [15]. Precise figure of the incidence and prevalence of MEs is difficult to obtain because the rate varies from study to study [7]. Studies showed a range of rate of serious patient injuries due to medication errors as 1 to 2% [16], 9 to 13% [2], 29% [17], and as high of 51.8% [9], and estimated 30.5% deaths per year in a survey in the United States of America (USA) were attributable to MEs [15]. A study conducted in Southern Iran with the purpose of determining the frequency of MEs in an emergency department of a teaching hospital revealed that 96.5% of patients had experienced at least one medication error, making the rate of errors 3.5 per patient [18]. A descriptive survey of 300 nurses working in hospitals affiliated to Iran University of Medical Sciences using stratified multistage sampling disclosed a mean of 19.5 medication errors that the nurses acknowledged within 3-months period however the mean of error reporting was only 1.3 of error cases [19, 20].

Factors related to medication error reporting

A focus group study on barriers to MER in Canada identified barriers as an individual, organizational and cultural [21]. According to a descriptive cross-sectional study with sample size of 799 nurses conducted in Jordan, the proportion of medication error reporting was relatively high among female nurses than male nurses [7]. Twenty-six percent of nurses in a study conducted in Israel indicated that all MEs in their wards were reported and 46% of the nurses showed self-reporting of MEs. The nurses emphasized on a personal barrier to non-reporting such as ME experiences and error reporting experiences [22].

A perception that incidence reports do not result in significant changes or benefits and errors that did not result in harm were among the factors that affected the attitude of nurses to report medication errors [23, 24]. Perceived barriers which affect attitude of nurses to report medication errors were fear of adverse consequences from reporting and being subjected to disciplinary actions, fear of being blamed, fear of reaction from the nurse manager, from peers and fear of loss of jobs [6–8, 19, 25]. Other barriers for not reporting MEs include nurses not being aware that an error had occurred, process of reporting (detailed paperwork, time constraints, not understanding incident reporting process), forgetting to make a report when the ward is busy, lack of time for reporting and lack of awareness of the importance of reporting [19, 23, 26].

Two-thirds (66.7%) of nurses involved in a study conducted in two state hospitals in Turkey who stated that they involved in medication errors in the preceding 6 months had not reported the errors. The reported reasons (social factors) for not reporting MEs included fear of consequences, fear of a culture of blame and the need to cover up for the colleague involved [27].

Modifiable barriers to MER for nurses reported in different studies as organizational factors were revealed as lack of feedback to the reporter, lack of a readily available MER system, lack of information on how to report a ME, no positive feedback for giving medication correctly, too much emphasis on ME as a quality indicator of nursing care and motivational factors (such as no encouragement by management, fear of loss of professional registration), lack of organizational leadership and support [3, 8, 19, 25, 28]. Similarly taking medical responsibility and fear of distrust from patients were barriers of medication error reporting [29].

Encouraging administrative attitudes and responses to MER were appreciated in a study to enhance nurses’ voluntary reporting [30]. It is indicated in the literature that strategies should be implemented to establish reporting mechanisms to reduce medication errors at national as well as international levels [31]. Establishing structured protocols on drug administration and adopting a non-punitive approach to reporting medication errors were shown to decrease medication errors and improve patient safety [32]. Proportion of medication error reporting by nurses might be affected by multiple factors such as socio-demographic, social, attitude of nurses, and organizational factors [7, 8, 12, 19] (Fig. 1 indicates the factors involved in medication error reporting).

Health care systems in most developing countries suffer from serious deficiencies in quality, equity, efficiency and financing [33]. The quality of care is evaluated in the light of the provider’s technical standards and clients’ expectations. Medication administration is among the routine and highly complex nursing care activities which plays a great role in patient care and outcome. Medication safety and medication errors are important concerns for health-care consumers, health-care professionals, researchers and policy makers worldwide.

Nurses are more prone to making medication errors because of the increasing demands and pressures placed on them. Critical incidents must be detected and reported and turned into positive situations, from which lessons are learned and used to design better patient care practices and systems. So far nurses are at the front line of defense to intercept and report medication errors, yet the errors are severely under-detected and under-reported in practice.

Medication errors affect the quality of health care delivery. Improving patient safety and learning from errors relies on voluntary error reporting which gives the complete picture of medication errors. Thus, exploration of the proportion of nurses reporting medication errors and associated factors is important to inform service providers, program implementers and policy makers to improve the quality of the healthcare service.

Sample size determination

Where: n = minimum sample size required for the study.

z = score for 95% confidence interval (z = 1.96).

p = prevalence of medication error reporting (50%).

d = tolerable error (d = 5%).

For possible non-response, the sample size was increased by 10%, thus comprising 38.416 respondents the final sample size became 423.

Sampling procedure

Stratified sampling technique was used to allocate the sample proportionally to each FMoH level governmental hospital and then the study participants who had work experience of six months and above were selected from the list of all nurses using simple random sampling technique; a lottery method (Fig. 2).

Data collection tool and procedure

Data were collected by three nurses who were diploma holders using structured self-administered English-version questionnaire. The data collection tool was adopted from two studies conducted in Australia by Evans [3] and in Jordan by Mrayyan et al. [7] with the authors’ permissions and face and content validity were established by two expert nurse advisors. The tool was composed of five parts. The first part contained socio-demographic characteristics of nurses. The second part of the questionnaire was related to error incidence. The third part related to medication error reporting was used to estimate the proportion of medication error reporting. The fourth and fifth parts were used to collect data regarding the attitudes of nurses on medication error reporting and perceived organizational culture and reality of dealing with errors respectively (Additional file 1).

Data quality control

Data quality was assured by conducting a pretest among 22 (5%) nurses at Zewditu hospital and appropriate modifications were made after analyzing the pre-test result before the actual data collection. The questionnaire was coded before data collection and cross-checked for consistency and completeness every day.

Data management and analysis

The returned questionnaires were checked for completeness, cleaned and the data were entered into EPI Info-7 and then exported to SPSS (Statistical Package for the Social Sciences) version 20.0 for analysis. Frequencies and cross tabulations were used to summarize descriptive statistics and tables were used for data presentation. Binary logistic regression was used to identify factors associated with medication error reporting and then the variables were checked for significant association using p-value, odds ratio and 95% confidence interval.

Socio-demographic characteristics

Proportion of medication error reporting

The proportion of medication error reporting among nurses in this study was found to be 57.4% (n = 288). Out of the total participants (n = 397), relatively high (74.5%) proportion of medication error reporting was found among female nurses as compared to male nurses (42.7%). A high proportion (70.8%) of medication error reporting was disclosed by married participants than participants who are single (52.6%). On the other hand, 277 (69.8%) of the participants perceived that medication errors should be reported as they occur. Moreover, (70.7%) of the nurses who did not make errors themselves reported medication errors.

Factors associated with medication error reporting

The proportion of medication error reporting was high among female nurses as compared to male nurses. Female nurses were 72.7% times more likely to report medication errors than male nurses (AOR = 0.273; 95% CI = 0.165–0.450). Similarly, marital status was an important predictor of medication error reporting. Nurses who are married were 54.6% times less likely to report medication errors as compared to those who are single (AOR = 0.454; 95% CI = 0.251–0.821).

Having a medication error experience was found to be another important determinant of medication error reporting. Nurses who had no medication error experience were 55.5% times more likely to report medication errors than those who had medication error experiences (AOR = 0.445; 95% CI = 0.274–0.722). The odds of medication error reporting among nurses who have not made errors themselves was 57.4% times higher as compared to those nurses who had previously made medication errors (AOR = 0.426; 95% CI = 0.230–0.789).

Limitation

Causes of medication errors were not studied in this research. Knowing the causes of medication errors gives a complete picture of medication errors. Therefore, it would be better if the study was triangulated. Response bias and recall bias were not reduced.

Recommendation