Potential adverse drug events and its predictors among hospitalized patients at medical center in Ethiopia: a prospective observational study

Potential adverse drug event (PADE) is a medication error with the potential to cause associate degree injury however that does not cause any injury, either due to specific circumstances, chance, or as a result of the error being intercepted and corrected. This study aimed to assess the incidence, contributing factors, predictors, severity, and preventability of PADEs among hospitalized adult patients at Jimma Medical Center. A prospective observational study was conducted among hospitalized adult patients at a tertiary hospital in Ethiopia. Logistic regression was performed to identify factors predicting PADE occurrence. P-value < 0.05 was considered for statistical significance. A total of 319 patients were included. About 50.5% of them were females. The mean ± SD age of the participants was 43 ± 17.6 years. Ninety-four PADEs were identified. Number of medications (adjusted OR = 5.12; 95% CI: 2.01–13.05; p = 0.001), anticoagulants (adjusted OR = 2.51; 95% CI: 1.22–5.19; p = 0.013), anti-seizures (adjusted OR = 21.96; 95% CI: 6.57–73.39; p < 0.0001), anti-tuberculosis (adjusted OR = 2.2; 95% CI: 1.002–4.59, p = 0.049), and Elixhauser comorbidity Index ≤ 15 (adjusted OR = 6.24; 95% CI: 1.48–26.25, p = 0.013) were independent predictors of PADEs occurrence. About one-third of patients admitted to the hospital experienced PADEs.

www.nature.com/scientificreports/ medication is in the control of the health care professional and patient" 3 . Medication errors are a significant health burden causative to over half of all ADEs among hospitalized patients 4 . Globally, the price related to medication errors has been 42 billion $/year, not considering lost wages, productivity, or health care prices 5 . Potential adverse drug event (PADE) is a medication error with the potential to cause associate degree injury however that does not cause any injury, either due to specific circumstances, chance, or as a result of the error is intercepted and corrected 6 . It has been reported that PADEs constitute over 17 million emergency department visits and 8 million hospital admissions per year in the United States 7 . The incidence of PADEs was estimated to be 13.8 per hundred admissions in Saudi Hospital 8 .
Among the 52 medication errors reported in Morocco, 53.8% result in clinically significant potential harm and 46.2% result in actual patient harm. According to this report, there were 7.7 medication errors for a thousand patient-days. The preventable event occurrence was higher in the ordering (71.1%), followed by the administration (21.2%) and transcribing stage (5.7%) 9 .
The incidence of medication errors per thousand patient-days ranges from 7.7 to 40.9 9,10 . Of all medication errors, the prescribing and monitoring were the most common error stages 10 . Seventy-one percent of the potentially harmful medication error occurrence was found to be at the ordering stage of the medication-use process 11 . Non-psychiatric drugs were three times as likely to cause ADEs compared to psychiatric drugs 10 .
The total prescribing error rate was 40.9% with 1.3% significant errors, in Nigeria. Duration of treatment omission and abbreviations which can lead to serious errors was the most common 12,13 . Medication administration errors in a University Hospital in Egypt, about 5531 errors were observed with 2.67 errors per observation and the overall error rate was 37.68% 14 .
In our country, medication error incidence of 56.4% 15 , 40 per 100 orders 16 and 52.5% 17 were reported. Wrong drug combination (28.13 16 16 , 15.1% 17 ) were the common medication ordering errors. The medication administration errors were found to be 51.8%; wrong timing (30.3%) and missed doses (18.3%) were the common administration errors 18 . The errors ranged from 16.8 to 28.6% for non-intravenous medications and from 20.6 to 33.4% for intravenous medications 19 .
Hospitalized patients are more likely exposed to polypharmacy. This, in turn, is a concern for PADEs. Patients who have PADEs are likely to have a longer hospital stay, reduced quality of life, increased overall health care cost, and an increased risk of morbidity and mortality. To our knowledge, in Ethiopia, there is no prospective observational study that followed patients admitted in the ward to identity the incidence, severity, preventability of PADEs. Therefore, this study aimed to determine the incidence, contributing factors, predictors, severity, and preventability of PADEs among hospitalized adult patients at Jimma Medical Center.

Methods
Study setting and period. The study was conducted among hospitalized patients at the medical ward of Jimma Medical Center (JMC), the only medical center in the south-west part of the country with 800 active beds 20 .

Study design & population.
A prospective observational study was conducted among adult patients admitted to inpatient medical wards or units.

Participant's eligibility and inclusion.
Participant's eligibility and inclusion were performed according to previous study report 20 . PADEs during/before admission were not included in the calculated incidence.
Sample size and sampling technique. The sample size equals 319 was calculated based on the assumption detailed in the previous article 20 . The proportion of PADE occurrence (P) = 0.525 was taken from a study done in JMC 18 . Data collection instrument, procedures, and quality assurance. A semi-structured questionnaire was designed by reviewing different literature for important variables 3,21-23 . Patient medical chart review, patient interview, and direct observation was performed to obtain the data 20 . PADEs were identified on the conditions that medication errors that can cause clinically serious harm in advance 3 . Drug-drug interaction was assessed as per Lexicomp drug interaction classification since Lexicomp Interactions scored highest in scope and completeness compared to seven drug information resources 24,25 . Drug-drug interaction with major severity levels; contra-indicated (avoid combination) and consider therapy modification were considered. The severity of PADEs was classified according to the National coordinating council for medication error reporting and prevention (NCCMERP) severity category modified definition 21 and according to the stage in the medication use stages, they have occurred as prescribing (ordering), dispensing, administering, transcribing and monitoring. The training was given to data collectors on the data collection procedure and research objectives. Before exporting to SPSS, data was checked and cleared in EpiData to exclude ambiguous, incomplete, and erroneous data. Outcome measures and validating methods. In the current study, methods used for detecting PADEs include a chart review, patient interview, and direct observation 20 . The patient's medical chart and documents such as the progress note, laboratory result, prescriber's orders, and drug administration chart were assessed 20 .

Data processing and analysis. Statistical Package for Social Sciences (SPSS) version 24 and Microsoft
Excel (2010) were used for analysis. Multivariate logistic regression was performed to identify independent predictors of PADE occurrence. A p-value of < 0.05 was considered to be statistically significant. The outcome of the study was reported as PADEs incidence per 100 admissions, per 1000 patient-days, and per 100 medication orders; severity of PADEs; the percentage of PADEs in stages of medication use (ordering/ prescribing, transcribing, dispensing, administering, or monitoring).
• "PADEs incidence per 100 admissions: The total number of PADEs identified, divided by the total number of admissions; multiplied by 100" • "PADEs incidence per 1000 patient-days: The total number of PADEs identified, divided by the total number of patient-days multiplied by 1,000" • "PADEs incidence per 100 medication orders: The total number of PADEs identified, divided by the sum of medications ordered multiplied by 100" Ethical approval and consent to participate. Ethical clearance & approval was obtained from the institutional review board (IRB) of Jimma University with the reference number of IHRPGD/550/19. It was based on the 1964 Helsinki declaration and its later amendments or comparable ethical standards. Before the start of the study, written informed consent was requested and received from the patient. Informed consent was obtained from all individual participants included in the study.

Operational definitions and definition of terms.
• Medication errors: "Any preventable event that may cause or lead to inappropriate medication use or patient harm while the medication is in the control of the healthcare professional and patient" 6,26 . • PADE: "A medication error with the potential to cause an injury but which does not actually cause any injury, either because of specific circumstances, chance, or because the error is intercepted and corrected" 3 . • Educated: Participants who had primary, secondary, or tertiary education.

Results
Socio-demographic characteristics of the study participants. From a total of 319 participants, 158 (49.5%) of them were males. The mean ± SD age of the participants was 43 ± 17.6 years. Most of the participants, 225 (70.5%) were from a rural area. About 27.3% of study participants drunk alcohol and 14 (4.4%) patients had used traditional medicine. The mean ± SD and the total length of hospital stay of the patients were 17.8 ± 14.5 days and 5667 patient-days respectively. Comorbidities were determined by weighted Elixhauser Comorbidity Index (ECI) and the mean ± SD of ECI was 5.7 ± 5.8. The mean ± SD number of medications prescribed for the study participants was 4.4 ± 2 ( Table 1).
Diagnosis of study participants. The diagnoses of the patients were categorized according to the international classification of disease (ICD)-10 codes. Most of the patients were diagnosed with diseases of the circulatory system (53%), infectious and parasitic diseases (34.5%), and diseases of the genitourinary system (28.5%). The diagnosis category most commonly associated with PADEs were diseases of the circulatory system (51.1%) and infectious and parasitic diseases (46.8%) ( Table 2).
Among the patients involved in the study, 171 (53.6%) had a previous medical condition. Diseases of the circulatory system 88 (51.46%), infectious and parasitic diseases 48 (28.07%), and endocrine, nutritional and metabolic diseases 25 (14.62%) were the common previous medical condition of the patients (Table 3).

Medication history.
Based on documented and available data, 166 (52%) patients had a history of medication use in the 3 months before the study period. One hundred eight patients were on medication during admission. Most of the patients were on cardiovascular medicines 79 (73.15%), antibiotics 28 (25.93%), and antiviral agents 28 (25.93%) (   (Fig. 2). The clinical pharmacists and clinical pharmacy postgraduate students working in the ward intervened and prevented the PADEs from causing harm.

Discussion
Medication errors may occur at any medication use stages (prescribing, dispensing, administration, and monitoring) and can result in severe harm, disability, and even death which are avoidable harm 27 . Health care systems should design specific programs of action for improving patient safety in each of four medication use stages, developing strategies, plans, and tools to ensure that the medication process has the safety of patients, monitoring medication-related harm, and producing a strategy for setting out research priorities 2 . In present study, the incidence of PADEs were 29  Key: "(B) An event occurred but the medication did not reach the patient. (C) An event occurred that reached the patient but did not cause harm. (D) An event occurred that reached the patient and required monitoring to confirm that it resulted in no harm to the patient and/or required intervention to preclude harm". www.nature.com/scientificreports/ to 23.5) per 1000 person-days. The higher incidence rate was observed in the current study compared to 5.5 PADEs per 100 admissions reported by Bates and colleagues 29 .
Multivariate analysis indicated that the number of medications the patient was receiving, ECI, anti-seizures, anti-TB agents, and anticoagulants were independent predictors of PADE occurrence. The ability of these variables to predict PADE occurrence was assessed using AUROC, which is 80.2% (95% CI: 74.9-85.4%); thus the model demonstrated excellent performance.
Patients who received greater than or equal to 7 medications had higher odds of experiencing PADEs among the study participants. In line with this, Diaz and colleagues 30 reported an increased number of prescribed medications were significantly associated with all adverse events. Using multiple drugs concurrently, ADEs result from alterations of the pharmacokinetics parameters 31 .
Anticoagulants were independently associated with the occurrence of PADEs. In anticoagulant therapy, the thrombotic and hemorrhagic risk is easily affected by factors such as age, co-morbidities, and concomitant medications. PADEs of anticoagulants are influenced by the types of anticoagulant agents, therapeutic versus prophylactic therapy, and duration of treatment 32 . Besides, anticoagulants have a narrow therapeutic index, and pharmacokinetics or pharmacodynamics interactions with other drugs may result in PADEs 33 .
Anti-seizures were significantly associated with the occurrence of PADEs. When other drugs combined with anti-seizures to treat intercurrent illness, there is a probability of PADEs, because anti-seizures are commonly given for prolonged time, have a narrow therapeutic window, and little alterations in their pharmacokinetics can result in toxic effects. Carbamazepine, phenytoin, valproic acid, and phenobarbital greatly alter liver enzymes and can affect the metabolism of other combined medications 34 .
Anti-TB was also found to have a significant association with the occurrence of PADEs. Rifampin, isoniazid, and pyrazinamide are hepatotoxic and their interaction with other drugs will increase the risk of PADEs. Genetic causes, advanced age, malnutrition, high dosage, and multiple comorbidities are predisposing factors for PADEs of anti-TB agents 35 .
Patients who were receiving antiviral agents were more likely to experience PADEs than patients who were not receiving these agents. Mok and colleagues 36 noted a significant number of PADEs of antiviral agents, leading to severe PADEs. Anwikar and colleagues 37 observed a highly significant association between the use of zidovudine and anemia.

Conclusion
The incidence of PADEs was 29.47 per 100 admissions, 16.59 per 1000 person-days, and 6.74 per 100 medication orders. The most common stage of the medication use process at which PADEs occurred was at the prescribing stage. The number of medications, ECI, anti-seizures, anti-TB agents, and anticoagulants were independent predictors of the occurrence of PADEs.
Key: "(B) An event occurred but the medication did not reach the patient. (C) An event occurred that reached the patient but did not cause harm. (D) An event occurred that reached the patient and required monitoring to confirm that it resulted in no harm to the patient and/or required intervention to preclude harm".   www.nature.com/scientificreports/