Potentially inappropriate medication use among elderly patients on follow-up at the chronic care clinic of a specialized teaching hospital in Ethiopia. A cross-sectional study

Abstract


Introduction
The Global proportion of elderly population (age ≥ 65years) is projected to double from 703 million in 2019 to 1.5 billion in 2050 (1).In Ethiopia, the proportion of elderly population is increasing overtime (2); in 2019 populations aged 65 years and above were 3.52% of the country's total age groups (3).This age groups are usually fragile and more susceptible to drug-related problems as a result of multi-morbidity, polypharmacy, and the physiological changes that affect the kinetics and dynamics of drugs (4)(5)(6).As a result, elderly patients are prone to PIMU, which is de ned as using a drug in which the risk of an adverse event outweighs its clinical bene t (7).Thus, medication selection in elderly patients should be made with carefulness (8).
There are multiple screening tools to assist the healthcare providers in selecting medication therapy, and reduce the exposure of the elderly to PIMU.Among them, the AGS Beers Criteria® (9) and STOPP/ START are the two most widely used criteria (10).Despite this, there is a growing evidence suggesting therapeutic decisions in older patients are frequently suboptimal or potentially inappropriate (11).
Numerous studies have been conducted to determine the magnitude and factors associated with PIMU using various screening tools.Accordingly, the reported magnitude vary across the studies due to reasons like, the type of screening tool used and others.Using Beers criteria, for instance, in a study from six European hospitals, at least one PIMU was identi ed in elderly patients ranging from 22.7-43.3%(12).While, studies from United States (13) and Brazil (8) reported PIMU in 24% and 26.9% of the elderly patients, respectively.In Middle East, several studies have reported a high prevalence of PIMU; 57.5% from Saudi Arabia (14), 62.6% and 76.0% Qatar (15,16), 59.6% from Lebanon (17), and 53.1% from Kuwait (18).In Africa, one study from Nigeria (19) reported a 31% PIMU among elderly patients, while studies from Ethiopia revealed nearly similar magnitude of PIMU; 27.72% from Gondar (20), 23% from Dessie (21), and 28.6% from Tigray (22).Using STOPP/START, a study from Kuwait (18) reported at least one PIMU in 55.7%, while in a study from Gondar (23) PIMU was identi ed in 61.5% in of elderly patients.Polypharmacy (taking more than or equal to 5 medications) (12,24), Sex (24,25), and age (25,26) were among the independent predicators of PIMU reported in studies.

Methods
The study aim, design, and setting The primary aim of this study was to determine and assess the magnitude and predictors of potentially inappropriate medications use in elderly patients on follow-up at the chronic care clinic of Jimma Medical Center.It has also addressed the magnitude of PPOs, the internal agreement, sensitivity and speci city of Beers and STOPP criteria in detecting PIMU.Hospital-based retrospective cross-sectional study design was employed.The study was conducted in JMC chronic care clinic from November 01, 2020 to December 30, 2020.JMC is the only specialized teaching hospital in the South west Ethiopia.It is located in Jimma town, 352 km south west of the capital city, Addis Ababa.JMC is the only teaching and referral hospital in the South Western part of Ethiopia with bed capacity of 620.It provides services for approximately 9000 inpatient and 80000 outpatient clients a year with a catchment population of about 15 million people.

Population Source population
The source population were elderly patients on follow-up at the chronic care clinic of JMC.

Study population
Elderly patients aged 65 years and above who had a treatment follow-up at the chronic care clinic of JMC for at least six months before the current study were included in the study.

Sample size and sampling procedure
Sample size was determined by using a single population proportion formula considering the standard normal variance (Z) = 1.96; estimated prevalence of PIMU (P) = 61.5% from Gondar study (23), and margin of error (D) =5%; total elderly patients aged 65 and above on active follow-up in the setting (N) = 543.This had resulted a nal sample of (n) =219.
The participants were selected using systematic random sampling technique.

Study variables Outcome/dependent variables
The independent variable of this study was the presence or absence of PIMU (according to Beers criteria, STOPP/START criteria) and PPOs.

Independent variables
Socio-demographic variables (age, gender, marital status, residence), clinical and medication-related variables (chronic diseases type and number, Charlson Comorbidity Index score (CCI), medications regimen, number of medications per patient).CCI score was determined using online Charlson Comorbidity Index-MDCalc (47).

Data collection tool and procedure
Data was collected using a checklist developed by extracting relevant variables from related literatures.Two professionals (Bachelor degree graduates in patient-oriented Pharmacy) were employed as data collectors.The data collectors reviewed medical charts of elderly patients, individuals aged 65 years and above as per this study (1), to extract relevant socio-demographic and clinical information, and to establish the list of all most recent medication regimen the patient received during the last visit to the chronic care clinic.

PIM assessment
After the completion of data collection, three Clinical Pharmacists (Masters of Clinical Pharmacy graduates) assessed PIMU using the 2019 updated AGS Beers Criteria® (9) and STOPP/START criterion's (Version 2) (10).Both criteria were used in the previous studies from Ethiopia (20,48,49).The AGS Beers Criteria® contains an explicit list of PIMs that are typically best avoided by older adults in most circumstances or under speci c situations, such as in certain diseases or conditions.The criteria is comprised of ve categories: medications that are potentially inappropriate in most older adults, those that should typically be avoided in older adults with certain conditions, drugs to use with caution, drug-drug interactions, and drug dose adjustment based on kidney function.On the other hand, the STOPP/START criteria version 2 was applied to identify a list of PIMs (STOPP criteria) and Potential Prescription Omissions (PPOs) (START criteria).STOPP/START is consists of 80 STOPP and 34 START criteria.START criteria contains medications that should be considered for people with certain conditions (PPOs).

Data quality assurance
To ensure quality of the data, a brief training was provided to the data collectors on the objective of the study, data collection tool and collection procedure.Prior to the actual data collection, pre-test was done by reviewing eleven (11) medical charts of the elderly participants to check the validity of the checklist for most of the items of the study.PIM assessors were also made more familiar with the 2019 updated AGS Beers Criteria® and STOPP/START criteria (Version 2) for assessing PIMU.

Data processing and analysis
Data entry, clearance and analysis was carried-out using SPSS Version 22.0.Frequency and percentage were calculated for categorical variables.For continuous variables, normality test was done using Shapiro-wilk test; data was considered normally distributed when the p-value of the test is not < 0.05.Then, parametric (normally distributed) data were presented using mean, whereas median was calculated for non-parametric variables.Patients' diagnosis were grouped according to the categories listed in the International Classi cation of Diseases-11 (ICD11) (50).A minimal threshold of ve medications was used to declare poly-pharmacy (51).Spearman's rho (r s ) correlation test was conducted to check the presence and strength of correlation between the number of PIMs identi ed using Beers criteria and STOPP criteria, while Cohen`s kappa (κ) test was conducted to determine the reliability between the two PIM raters used in this study (Beers criteria and STOPP criteria).The sensitivity and speci city of the two PIM raters were also checked.Using a dichotomous variable to represent the presence or absence of PIM (0 = no PIM; 1 = PIM), a binary logistic regression analysis was conducted after checking cell adequacy of each categorical variables using a Chisquare test.Variables with a p-value < 0.25 were recruited for multivariable logistic regression analysis.Hosmer and Lemeshow Test was conducted and both models of logistic regression indicated a good t (P>0.05).In all statistics, a cut-off p-value < 0.05 was considered to declare statistical signi cance of the association.

Overview of the study
This study involved ambulatory patients (n=219) aged ≥ 65 years old on follow-up at the chronic care clinic for at least 6-months.The average age of the study participants was 70 (IQR=9), and nearly two-thirds (n=143; 65.3%) of the participants were male (Table 1).

Clinical and related information
All of the participants had at least one chronic disease.Disease of the circulatory system were the most commonly class of diseases, hypertension (n=127; 58%) being the predominant of all (Table 2).

Medication-related information and PIMU
The total number prescribed medications were 902; on average each patients were prescribed with 4.0 (IQ=2.0)medications.Overall, 93.0 (42.5%) patients were on polypharmacy.PIMU was identi ed in 182 (83.1%) and 99 (45.2%)patients, according to Beers and STOPP criteria, respectively.Furthermore, 24 (10.9%)patients had at least one PPO (Table 3).According to Beers criteria, Aspirin (n=71; 24.9%) was the most commonly prescribed PIM which needs a cautious use in those aged 70 years and above followed by Hydrochlorothiazide (n=50; 17.5%) again with cautious use recommendation (Table 4).

Discussion
This was a retrospective cross-sectional study conducted involving 219 elderly patients on follow-up at the chronic care clinic of a specialized teaching medical center in Ethiopia.The main objective of this study was to determine the magnitude and factors associated with PIMU based on Beers and STOPP criteria.Accordingly, 83.2% and 45.2% of the patients had at least one PIM based on Beers and STOPP criteria, respectively.Additionally, 24 (10.9%)patients had at least one PPO.
In the present study, the magnitude of PIMU based on Beers criteria was higher than some previous studies.The magnitude of PIMU was 50.0% in a study from USA (52), 26.9% from Brazil (8), and 30.5% from Irish (24).In India, studies had reported PIMU prevalence of 23.5% (53), 24.6% (26), and 61.9% (25) (20), and Dessie (21) reported PIMU in 28.6% 27.7%, and 23% of the elderly patients, respectively.The discrepancy in the magnitude of PIMU could be due to many factors.For instance, Beers criteria are the commonly used guidelines to manage and improve the care of individuals aged 65 years and older in healthcare settings (9).Contrary to this, our study setting lack the privilege of agging potentially inappropriate medication list for extra caution which will made prescribers to comfortably rely on the same medication for years without the concern of safety.Additionally, adopting different version of Beers criteria in the previous studies (AGS Beers Criteria 2012 and 2015) as compared to the present study (AGS Beers Criteria 2019) might also explain the difference in the magnitude of PIMU.Furthermore, the difference in the data collection method (chart review versus prospective) employed across those studies might have also contributed to the variation in the magnitude of PIMU.
Based on STOPP criteria, at least one PIMU was identi ed in 45.2% of the elderly patients in our study.This indicates nearly half of our participants were taking medication which could be harmful to their health.Using the same criteria, studies from Gondar (23) and Kuwait (18) reported at least one PIMU in 61.5% and 55.7% of elderly out patients, respectively.These magnitudes are higher than our study nding.The Gondar study was a prospective study which is a better design to track all medication used by the patient, and the Kuwait study was also a prospective study and the investigators employed both medical electronic and non-electronic records to exhaustively access the patientsp rescribed medications and other information.In our case there is only non-electronic records (patient medical chart) to access prescribed medications and other information which might have some incomplete medication list.Besides, limited availability of some medications in Ethiopia could have contributed to the less magnitude of PIM identi ed in our study.
In the present study, as the age of the patient increased, the risk of Beers PIMU was also observed to increase [p<0.001].Based on either Beers or STOPP criteria, hypertension and taking polypharmacy were signi cantly increased the probability of PIMU.Taking polypharmacy had increased the risk of PIMU by more than fourteen [p < 0.001] and six times [p < 0.001] based on Beers and STOPP criteria, respectively.Being hypertensive increased the likelihood of PIMU by more than four times [p < 0.01] and two times [p < 0.04] based on Beers and STOPP criteria, respectively.
Similarly in the previous studies, taking polypharmacy (14,55,56), advanced age (25,56), and hypertension (57) were reported as a signi cant predicting factors for PIMU.As the age advance, metabolic changes and decreased drug clearance, and increased drug-drug interactions are expected (58).On the other hand, simultaneous use of multiple medications probably increases the risk of drug-drug, drug-disease interactions as well as diverting clinician's attention to provide quality care, which in turn increases the likelihood of prescribing PIMs.Contrary to our expectations, in the current study, patients with ischemic stroke [p= 0.01] and asthma [p < 0.001] were associated with lower Beer's PIMU.In our study, the proportion of the patients with these disease conditions were small which could be a possible justi cation.
According to STOPP criteria, increased likelihood of PIMU was also observed in patients with ischemic heart disease [p < 0.04], diabetes mellitus [p < 0.04], and peripheral neuropathy [p < 0.001].Other studies had also reported similar predictors (14)(15)(16).Surprisingly, age was not a signi cant predictor of PIMU based on STOPP criteria.As chronic morbidities are expected to increases with age, so does the risk of multiple comorbidities and multiple medication use.

Conclusion
In the present study, PIMU was identi ed in large proportion of the participants.Multiple medication use and certain comorbidities had increased the probability of PIMU.We recommend the use of screening tools for reviewing medications prescribed for each hospitalized elderly patients to reduce the adverse consequences related to PIMU.

Declarations Ethical approval and consent to participate
The study was ethically approved by the ethics committee of Jimma University school of Pharmacy (Ref.no:SP/200/2013).Subsequently, permission was granted from JMC to access medical records.Raw data extracted from the patient medical chart were held con dential during the data collection and afterwards.Informed consent requirement was waived by the ethics committee of Jimma University school of Pharmacy as the study was done through chart review.All methods were carried out in accordance with the approved protocol.

Table 1 :
Sociodemographic information of the participants.

Table 3 :
Medication-related information and the magnitude of PIMU identi ed in the study.
PIMs-Potentially inappropriate medications, PPO-Potential Prescription Omissions, STOPP-Screening Tool of Older People's Potentially Inappropriate Prescriptions, START-Screening Tool to Alert Doctors to Right Treatment.

Table 5 :
Speci c PIMs and PPOs according to STOPP/START criteria.it had a sensitivity of 52.20% and speci city of 89.19%, whereas taking Beers criteria as a test result, the sensitivity and speci city of Beers criteria was 96.0% and 27.5%, respectively.

Table 2 :
Clinical and related information of the study participants.Human immunodeficiency virus disease, viral hepatitis, Neurosyphilis, Pneumonia, Pulmonary tuberculosis.

Table 4 :
Specific Beers PIMs prescribed in the elderly patients involved in the study.

Table 7 :
Logistic regressions analysis for identifying predictors of PIMU based on STOPP criteria.