Inappropriate medication prescribing, polypharmacy, potential drug-drug interactions and medication regimen complexity in older adults attending three referral hospitals in Asmara, Eritrea: a cross-sectional study

Abstract

Background

Older adults often face several chronic illnesses that require them to take multiple medications. The increased number of prescribed medications has led to more complex medication regimens, putting older adults at a higher risk of potential drug-drug interactions, inappropriate medication prescribing, and adverse events. This study aimed to assess inappropriate prescribing practices, polypharmacy, medication regimen complexity, and their determinants in older adults.

Methods

A cross-sectional study was conducted among older adults (aged 65 years and above) who visited three referral hospitals in Asmara, Eritrea, between June and August, 2023. A stratified random sampling technique was used, and data were collected from patient prescriptions, medical cards, and through interviews with a questionnaire. Inappropriate medication prescribing was evaluated using STOPP (Screening Tool of Older Person's Prescriptions)/ START (Screening Tool to Alert to Right Treatment) criteria version 3. Potential drug-drug interactions (pDDIs) and medication regimen complexity (MRC) were assessed using Lexi-comp drug interaction checker and MRC index, respectively. Descriptive statistics, logistic regression, Pearson’s correlation coefficient, independent samples t-test, one-way Analysis of Variance, and paired t-test were employed using IBM SPSS (version-26.0).

Results

A total of 430 respondents, with a similar male to female ratio, were included. The prevalence of polypharmacy was 5.3% (95%CI: 3.2, 7.5). Moreover, the prevalence of clinically significant pDDI was 51% (95%CI: 46, 56). The most common medicines involved in clinically significant pDDIs were enalapril (n = 179) and acetylsalicylic acid (n = 124). The presence of chronic illness (AOR = 7.58, 95%CI: 3.73, 15.39) and the number of drugs prescribed (AOR = 2.80, 95%CI: 1.91, 4.10) were predictors of clinically significant pDDIs. The prevalence of potentially inappropriate medications (PIMs) and potential prescribing omissions (PPOs) were 27.4% (95% CI: 23.4, 31.8) and 13.3% (95% CI: 10.3, 16.7), respectively. The most common PIMs were long-acting sulfonylureas (n = 63) and aldosterone antagonists (n = 19). Besides, proton pump inhibitors (PPIs) (n = 41) and cardio-selective beta-blockers (n = 14) were the most common PPOs identified. Age (AOR: 0.95, 95% CI: 0.92, 0.98), presence of chronic illness (AOR: 1.51, 95% CI: 0.81, 2.80), and number of drugs prescribed (AOR: 2.01, 95% CI: 1.51, 2.69) were significant factors associated with PIM. MRCI score was a significant determinant of PPO (AOR: 1.25, 95% CI: 1.14, 1.38). The mean (SD) of the overall MRCI score was 9.1 (3.7), with dose frequency being the major contributor. The number of drugs prescribed was a determinant of MRCI score (r = 0.625, p < 0.001).

Conclusion

Inappropriate medication prescribing and clinically significant drug-drug interactions were common among older adults, highlighting the need for immediate attention from policymakers, program managers, and healthcare professionals.

Introduction

The world is undergoing a dramatic demographic shift, with the population aged 65 and above expanding at the quickest pace globally [1]. Forecasts indicate that by 2050, 12.2% of the total worldwide population will be over 70 years old, a substantial increase from what was 6.4% in 2022 [2].

While increased longevity is a positive development, it also presents significant healthcare challenges. Aging affects normal physiology, causing variations in pharmacokinetic and pharmacodynamic parameters of many medicines, increasing the risk of drug-related morbidity and mortality [3]. Moreover, older patients often have complex medical conditions and take multiple medications, leading to widespread problems of inappropriate prescribing and dangerous drug-drug interactions [4, 5].

The increased number of medications and multiple daily dosing in older adults may lead to errors with dosing and administration, as well as complexity in the medication regimen [6, 7]. This has led to the development of Medication Regimen Complexity Index (MRCI) as a tool to assess the complexity of medication regimens [6]. Medication regimen complexity has been associated with adverse drug events (ADEs) [8], hospital readmission [8, 9], adverse quality of life [10] and all-cause mortality [7].

Prescribers' uncertainty regarding appropriate pharmacotherapy for older adults frequently results in inappropriate prescribing practices, which can lead to adverse drug reactions, drug-related hospitalizations, diminished quality of life, and even mortality [11]. Polypharmacy and comorbidities are closely associated with potentially inappropriate medication (PIM) use in the older adults [12]. The absence of clear guidance in clinical practice guidelines regarding prescribing for older adults with multiple conditions has led to the development of various explicit screening tools, such as the START (Screening Tool to Alert to Right Treatment) and STOPP (Screening Tool of Older Person's Prescriptions) criteria, to identify potential prescription omissions (PPOs) and PIMs, respectively [13, 14]. The START/STOPP criteria can be effectively utilized as a tool to identify patients with PIMs, thereby helping to improve the overall quality and appropriateness of pharmacotherapeutic practices [1, 15].

Prescribing medications for older adults is a complex process, as adverse effects related to drug therapy are more prevalent in this population compared to younger individuals [16]. Studies have shown a strong correlation between polypharmacy and a higher incidence of PIMs prescribed to older adults [17, 18]. Exposure to PIMs is a significant risk factor for adverse drug events (ADEs) in older patients [19, 20], and the use of PIMs and potential prescription omissions (PPOs) are associated with an elevated risk of hospital readmissions and increased mortality [21]. The use of PIMs among older adults has been found to be responsible for a significant proportion of hospital admissions, ranging from 6 to 30% of such cases [22]. A meta-analysis study has found a significant association between the use of PIMs and increased incidence of emergency room visits, adverse drug events, functional decline, reduced health-related quality of life, and hospital admissions [23].

Research has demonstrated that the STOPP criteria are more relevant than the Beers' criteria in identifying potentially inappropriate medications (PIMs) among older adults presenting to the hospital with acute illness [24,25,26]. The START criteria was the only tool among the various patient in focus listing approach (PILA) instruments analyzed that was able to effectively identify potential prescription omissions [1].

All of the above-mentioned factors contribute to the need for researches on the safety and prescription practice of medications in older adults. Inappropriate use of non-steroidal anti-inflammatory drugs and co-prescription with interacting drugs in older adults has been reported in the Eritrean healthcare system [27]. Moreover, a previous study conducted in an Eritrean community pharmacy revealed that 18.1% of prescriptions dispensed to older adults contained at least one PIM according to the 2023 American Geriatric Society (AGS) Beers Criteria [28]. The previous studies conducted in Eritrea did not consider medication regimen complexity (MRC), burden of clinically significant potential drug-drug interactions to a wide range of medications, and PPO. Therefore, this study assessed polypharmacy, inappropriate medication prescribing, medication regimen complexity, potential drug-drug interactions and their determinants among older adults attending three referral hospitals in Asmara, Eritrea.

Materials and methods

Study design and setting

A cross-sectional study was conducted in three selected referral hospitals in Asmara, Eritrea namely: Orotta National Referral and Teaching Hospital (ONRTH), Halibet National Referral Hospital (HNRH), and Hazhaz Zonal Referral Hospital (HZRH). ONRTH serves as a national referral center for pediatric, medical, surgical, and gynecology and obstetrics departments. Moreover, HNRH is a national referral center for orthopedic, medical, surgical, and dermatology. Eritrea consists of six regions, each with its own referral hospital. HZRH serves as a zonal referral center for Maekel Zone. Data were collected from June to August, 2023.

Source and study population

All older adult outpatients (aged 65 years and above) who attended outpatient pharmacies of the selected referral hospitals were the source population for the study. Older adult outpatients who were mentally capable of providing reliable information and attended the selected referral hospitals during the study period formed the study population.

Sample size determination and sampling design

The final sample size was found to be at least 377. The detailed sample size determination approach is provided in Additional file 1.

To get representative samples from each referral hospital, stratified random sampling was utilized. The three referral hospitals were considered as strata, and participants were selected using systematic random sampling. The computed sample size was proportionally allocated among the three referral hospitals. Besides, after allocating the samples proportionally into the three hospitals, every 4th patient was recruited from each hospital.

Data collection instrument and approach

A structured data collection instrument [Additional file 2], developed based on a review of similar published studies [27, 29,30,31], was used to collect data. It comprised of six sections. The first section included socio-demographic and background characteristics of patients and prescriber qualification. The second section aimed to record information on prescribed medications from prescriptions. The third section was used to document information from patients’ medical cards such as indications of the prescribed medications and history of chronic illness. Furthermore, the fourth section was intended to document information related to drug-drug interactions such as the name of interacting drugs, severity, risk rating, clinical implication, and management protocols. The fifth section was used to record inappropriate medication prescribing in terms of potentially inappropriate medications (PIMs) and potential prescribing omissions (PPOs). Finally, the sixth section was intended to document information on medication regimen complexity.

Data collectors approached participants and explained the study’s objectives. Upon obtaining written informed consent, a face-to-face interview was conducted at outpatient pharmacies of the respective referral hospitals. The interview aimed to obtain information about socio-demographic and background characteristics of the participants such as age, sex, educational level, occupation, presence of chronic illness, religion, and ethnicity. Prescriber qualification was obtained from prescriptions and/or medical cards. Subsequently, data on prescribed medications were recorded and the patients’ medical cards were reviewed. Then, prescriptions containing two or more medicines were screened for potential drug-drug interactions. Finally, prescribed medicines were screened for inappropriate medication prescribing (PIMs and PPOs) and medication regimen complexity. All the obtained data were documented and no-follow-ups were conducted.

Variables

Polypharmacy, potential drug-drug interactions, PIM, PPO, and medication regimen complexity index score were considered as dependent variables, while the independent variables were age, sex, presence of chronic illness, number of drugs prescribed, prescriber qualification, and medication regimen complexity index score.

Outcome measures

Inappropriate medication prescribing in older adults was assessed using STOPP/START criteria version 3. The STOPP/START criteria has been validated by a European expert panel using the Delphi consensus process [32]. It comprised 133 STOPP criteria and 57 START criteria (i.e. 190 criteria in total). PIMs and PPOs were evaluated using STOPP and START criteria, respectively.

According to the STOPP criteria, PIMs include those related to drug indication, cardiovascular system, antiplatelet/anticoagulant drugs, central nervous system and psychotropic drugs, as well as those affecting the renal, gastrointestinal, respiratory, musculoskeletal, urogenital, and endocrine systems, in addition to drugs with a high risk of falls and analgesic and antimuscarinic drugs [32]. The START criteria identify PPOs involving the cardiovascular, respiratory, central nervous, gastrointestinal, musculoskeletal, endocrine, and urogenital systems, including analgesic drugs and vaccines. Each category has a specific code used for both PIMs and PPOs, categorized based on drug groups into codes for the hospitalization period, drugs group, discharge drugs group, and both groups [32]. Polypharmacy in this study, was defined as the simultaneous prescription of five or more medicines used at the same time [33].

Outpatient prescriptions containing multiple medicines were screened for potential drug-drug interactions (pDDIs) using the Lexi-comp® drug interaction checker, which provides information on the risk rating, severity, and clinical management of the interactions [34]. The interactions were categorized based on their risk level into five groups—A, B, C, D, and X. Category A indicates no evidence of pharmacodynamic or pharmacokinetic drug interactions, and category B denotes little to no evidence of drug interactions [34]. Categories A and B do not require any action. Category C requires therapy monitoring and dosage adjustments due to evidence of drug interactions where the benefits usually outweigh the risks. Moreover, category D necessitates therapy modification including the selection of alternative agents, aggressive monitoring, and empiric dosage changes to realize the benefits and/or minimize toxicity, and category X being contraindicated as the risks typically outweigh the benefits [34]. The clinically significant pDDIs considered in this study were those falling under categories C, D, and X.

Drug interactions are categorized based on severity as minor, moderate, and major, with major interactions being highly clinically significant and potentially life-threatening [34]. Moderate interactions being of moderate significance and potentially leading to deterioration of the patient's condition unless therapy monitoring is implemented. Besides, minor interactions being of minimal clinical relevance and not requiring any alteration in therapy [34]. Drug interactions of minor, moderate, and major severity were considered as pDDIs.

The medication regimen complexity (MRC) level was assessed using the 65-item validated MRCI (Medication Regimen Complexity Index) instrument, which accounted for the number of medications, dosage form, dosing frequency, and additional instructions (e.g., tablet crushing, timing, and relation to food/liquid) [6]. The instrument comprised three sections about the route of administration, dosing frequency, and supplementary directions, with the patient-level MRCI calculated as the sum of these three sections using a Microsoft Access 2013 electronic data capture tool [6].

In the Eritrean healthcare system, prescribers include specialists, medical doctors, dentists, nurse practitioners, dental technicians, and various other lower-level healthcare providers.

Data quality assurance

To ensure face and content validity, the data collection instrument was reviewed by a panel of experts in the fields of pharmacy, medicine, public health, and epidemiology. Modifications were made based on their feedback and then subjected to a pre-test. The pre-test was conducted on 38 participants between May 29, and June 2, 2023, for checking the comprehensibility and compatibility of the data collection tool at randomly selected two hospitals. Data collection was carried out by four pharmacy professionals with prior experience. Before the pre-test, a one-day orientation session was provided to the data collectors and investigators to familiarize with the aims of the study, data collection instrument, data collection procedures, MRCI tool, drug interaction checker, and STOPP/START criteria version 3. Data reliability was confirmed through compliance to STOPP/START criteria, MRCI and drug interaction checker tools, and participation of well-oriented data collectors.

The study was reported in line with Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement [35].

Ethical approval

Ethical approval was obtained from the Ministry of Health (MOH) Research Ethics and Protocol Review Committee (reference number: 07/04/2023). Besides, permission was obtained beforehand from the medical directors of the respective hospitals. Study participants were informed about the study’s purpose and written informed consent was obtained from each respondent. All the data obtained was kept confidential and used solely for the study’s purposes. This study conforms to the principles outlined in the Declaration of Helsinki [36].

Statistical analysis

The collected data were double-entered on CSPro (version 7.3) [37] to minimize the keying errors. The entered data were exported to IBM SPSS (version 26.0) [38] for statistical analysis. Descriptive summaries of the quantitative socio-demographic variables were performed using mean (Standard Deviation (SD)) based on the normality of the variables. Frequencies and percentage were used for qualitative variables. Graphs and tables are used to present the data as appropriate. Five different outcome variables: polypharmacy, pDDI, PIM, PPO, and MRC were addressed in this study. The determinants of the four outcome variables were assessed using bivariate and multivariable logistic regression. Factors related to MRCI were assessed using Pearson’s correlation coefficient, independent samples t-test, and one-way Analysis of Variance. Comparison of the components of MRCI was conducted using paired t-test and graphically shown using error plot. Chi-square test for trend was also used to find out magnitude of association for number of drugs and its subsequent interaction respectively. Crude Odds Ratio (95%CI) and Adjusted Odds Ratio (95% CI) are reported in all logistic regression analysis. All analyses were considered significant when p < 0.05.

Results

Socio-demographics and background characteristics

A total of 430 older adult outpatients with a mean age of 74.2 (SD: 6.9) were included in the study. More than half of the respondents were female (53%), and the majority of them had not completed high school (85.2%). Most of the respondents had chronic illnesses (78.8%), with hypertension (60.5%) and diabetes mellitus (35.3%) being the most common. General practitioners attended to the majority of the older adult outpatients (95.1%) (Table 1).

Table 1 Socio-demographic and background characteristics of respondents and particular of prescribers (N = 430)

Medication utilization pattern

A total of 1124 medicines were prescribed, with a mean of 2.61 (SD: 1.05) medicines per prescription. Moreover, the minimum and maximum number of medicines per prescription were 1 and 7, respectively. The majority of the prescriptions contained two medicines (44.7%) [Additional file 3]. The most commonly prescribed medicines were enalapril (13.4%), acetylsalicylic acid (12.1%), hydrochlorothiazide (7.7%), metformin (6.1%), and glibenclamide (5.5%). Hypertension (35.3%), diabetes mellitus (20.2%), cardiac problems (6.2%), peptic ulcer disease (5.6%) and urinary tract infection (3.0%) were the most common indications for prescription.

Polypharmacy and its associated factors among older adult outpatients

The prevalence of polypharmacy was found to be 5.3% (95%CI: 3.0, 7.2). Binary logistic regression analysis indicated that age (p = 0.241), sex (p = 0.310), educational level (p = 0.858), and presence of chronic illness (p = 0.084) were not significant associates of polypharmacy [Additional file 4].

Clinically significant potential drug-drug interactions and its determinants

Out of 1132 drug pairs screened for potential drug-drug interaction (pDDI), 44.2% resulted in pDDI. Majority of the pDDIs were moderate in severity (n = 400) and risk category C (n = 401) (Fig. 1). Besides, out of the 390 prescriptions that contained more than one medicine, the prevalence of clinically significant pDDI among prescriptions was 51.5% (95%CI: 47.0, 57.0). A total of 416 clinically significant pDDIs with a mean number of 1.07 (SD: 1.43) per prescription was detected.

Fig. 1

Distribution of potential drug-drug interaction by risk category and severity (n(B) = 84), n(C) = 401, and n(D) = 16)

The most common medicines involved in clinically significant pDDIs were enalapril (n = 179), acetylsalicylic acid (n = 124), metformin (n = 109), glibenclamide (n = 60) and human insulin lente (n = 31). The most common clinically significant pDDIs along with their risk rating, severity, and clinical implication are displayed in Table 2.

Table 2 Most common clinically significant potential drug interactions with risk rating, severity, and clinical implication

Binary logistic regression analysis showed that presence of chronic illness (p < 0.001), MRCI score (p < 0.001), and number of drugs prescribed (p < 0.001) were significantly associated with occurrence of clinically significant pDDI (Table 3).

Table 3 Determinants of clinically significant potential drug-drug interactions across the categories of socio-demographic and other background characteristics in Asmara, Eritrea, 2023

Moreover, multivariable logistic regression indicated that respondents who had a chronic illness were approximately eight times more prone to clinically significant pDDI than those who did not have (AOR = 7.58, 95%CI: 3.73, 15.39) (Table 3). As the number of drugs prescribed increased by one unit, the odds of clinically significant pDDIs increased by 2.80 units (AOR = 2.80, 95% CI: 1.91, 4.10).

Moreover, chi-square for trend analysis indicated that there was a significant increase in drug interaction with increase in the number of chronic illnesses (χ2 = 26.52, p < 0.001). As the number of chronic illness increases by one unit, the odds of clinically significant potential drug interaction increases by 2.80 unit (COR: 2.80; 95%CI: 1.84, 4.25) [Additional file 5].

Potentially inappropriate medications and its determinants

The prevalence of PIM among the older adults was 27.4% (95% CI: 23.4, 31.8). According to the STOPP criteria version-3, a total of 121 PIMs were detected. The most common PIMs were related to the endocrine system (52.1%), cardiovascular system (27.3%), and central nervous system (14.0%) (Table 4). Sulfonylureas with a long half-life in type 2 diabetes mellitus patients (J01, n = 63), aldosterone antagonists with concurrent potassium-conserving drugs without monitoring of serum potassium (B13, n = 19), and first-generation antihistamines for allergy or pruritis were the major PIMs detected among the older adults. Table 8 displays a detailed description PIMs as per STOPP criteria version 3.

Table 4 Descriptions of potentially inappropriate medications as per STOPP criteria (n = 121)

Binary logistic regression analysis showed that age (p = 0.028), presence of chronic illness (p = 0.029), number of drugs prescribed (p < 0.001), and MRCI score (p = 0.002) were significant associates of PIM (Table 5).

Table 5 Determinants of potentially inappropriate medications vis-à-vis STOPP criteria version 3 across the categories of socio-demographic and other background characteristics in Asmara, Eritrea, 2023

Besides, multivariable logistic regression indicated that a year increase in age resulted to a 5% lower-odds of a PIM (AOR: 0.95, 95% CI: 0.92, 0.98). Older adults who had chronic illnesses were 1.51 times more likely to encounter a PIM that their counterparts (AOR: 1.51, 95% CI: 0.81, 2.80). Moreover, as the number of drugs prescribed increased by one unit, the odds of PIM increased by 2.01 units (AOR: 2.01, 95% CI: 1.51, 2.69) (Table 5).

Potential prescribing omissions and its determinants

The prevalence of PPO among older adults was 13.3% (95% CI: 10.3, 16.7). According to the START criteria version 3, a total of 57 PPOs were identified, with the most common related to gastrointestinal system (71.9%). Proton pump inhibitors with non-steroidal anti-inflammatory use (F03, n = 41) and cardio-selective beta-blocker for stable heart failure with reduced ejection fraction (B06, n = 14) were the predominant PPOs (Table 6).

Table 6 Descriptions of potential prescribing omissions as per START criteria (n = 57)

Binary logistic regression analysis indicated that presence of chronic illness (p = 0.041), number of drugs prescribed (p = 0.012), and MRCI score (p < 0.001) were significant associates of PPOs (Table 7). MRCI score was identified as a determinant of PPO, in which a unit increase in MRCI score resulted in a 1.25 units increase in the odds of PPO (AOR: 1.25, 95% CI: 1.14, 1.38) (Table 7).

Table 7 Determinants of potential prescribing omission vis-à-vis START criteria version 3 across the categories of socio-demographic and other background characteristics in Asmara, Eritrea, 2023

Medication regimen complexity and its determinants

The mean (SD) of the overall MRCI score was 9.1 (3.7) with a minimum score of 2 and maximum score of 21 [Additional file 6].

Number of drugs prescribed was the only significant determinant of MRCI score. MRCI score significantly increased with an increase in the number of drugs prescribed (r = 0.625, p < 0.001) (Table 8).

Table 8 Determinants of medication regimen complexity index across the categories of socio-demographic and other background characteristics in Asmara, Eritrea, 2023

Comparison of MRCI categories

Paired comparison of the categories of MRCI revealed that mean MRCI scores in descending order for dose frequency, drug administration, and dosage form were 4.3 (SD: 2.1), 3.0 (SD: 1.9), and 1.8 (SD: 1.5), respectively (Fig. 2).

Fig. 2

Comparison of medication regimen complexity index (MRCI) categories

Discussion

In this study, one in twenty of the older adults (5.3%) were prone to polypharmacy. This finding is much lower than studies conducted in Ireland (31.5%) [39], Portugal (72.09%) [40], Qatar (75.5%) [41], Somaliland (71%) [42], South Africa (97.6%) [43], China (91.3%) [44], and Togo (22.7%) [45]. The variation in results could be explained by the difference in study design, study setting, disease pattern, study population, and disease-specific treatment guidelines. Moreover, the current study focused on prescribed medications rather than other over-the-counter medications and food supplements that older adults may be taking at home. This may help explain the low prevalence of polypharmacy observed. Neither of the selected socio-demographic and background characteristics were significantly associated with polypharmacy. This is inconsistent with findings of similar studies where they reported that age, sex, and presence of comorbidity as significant determinants of polypharmacy [39,40,41,42, 45].

Half of the older adults (51.5%) had at least one clinically significant potential drug-drug interaction. The fact that majority of the older adults in this study are encountered with chronic illness might be attributable to the high occurrence of clinically significant pDDIs. This finding is higher than studies conducted in China (9.2%) [46], Portugal (10.6%) [40], Italy (16%) [47], Thailand (36.8%) [48], and Uganda (37.6%) [49]. However, it is lower than a study conducted in Ethiopia (68.6%) [50]. This variation in results can be explained by the difference in study design, study setting, study population, drug utilization pattern, changes in prescribing practices, significance of drug-drug interactions, and drug interaction checker used across the studies.

Majority of the documented potential DDIs were risk category ‘C’ and of moderate severity. In terms of risk category, this finding is comparable with similar studies conducted in India [51] and Uganda [49]. Moreover, in terms of severity, finding from this study is comparable to similar studies conducted in Ethiopia [50], South Africa [43], and Uganda [49]. This urges healthcare professionals to carefully assess such drug interactions, consider disease state of patients, and closely monitor therapy.

The most common medicines involved in clinically significant pDDIs were enalapril and acetylsalicylic acid. This finding is consistent with a similar study conducted in Italy [47] where they reported cardiovascular medications including enalapril as the most common medicines involved in pDDIs. To mitigate risks associated with these medications, prescribers and dispensers should be vigilant in monitoring and managing prescriptions containing such medications.

In this study, the most common interacting pairs were acetylsalicylic acid-enalapril and glibenclamide-metformin. This finding is in contrast to studies conducted in Uganda (metformin-quinine) [49] and Portugal (angiotensin-converting enzyme inhibitors (ACEI)-angiotensin-receptor blockers) [40]. This variation in results could be due to the differences in study setting, study population, disease pattern, drug prescribing practice, national standard treatment guideline and formulary, and drug interaction checker. The implication is that the interacting pairs of acetylsalicylic acid-enalapril and glibenclamide-metformin would result in increased nephrotoxic effect of ACEI and increased hypoglycemic effect, respectively [34]. Both drug interaction pairs were moderate in severity requiring close monitoring and proper disease management by prescribers. Besides, non-steroidal anti-inflammatory drug (NSAID) therapeutic duplication were detected among the older adults. Whenever possible, healthcare professionals should avoid therapeutic duplication to minimize risks associated with increased toxic adverse effects.

The determinants of clinically significant pDDI in this study were presence of chronic illness and number of drugs prescribed. This finding is in line with similar studies conducted in Ethiopia [50], India [51], Uganda [49], and Thailand [48]. Older adults with chronic illness were approximately eight times more likely to encounter a clinically significant pDDIs that their counterparts. Besides, in this study, the occurrence of clinically significant pDDIs increased by 2.8 units with a unit increase in the number of chronic illness. The fact that older adults with chronic illnesses require multiple medications could contribute to the occurrence of clinically significant pDDIs. As the number of drugs prescribed increased by one unit, the odds of clinically significant pDDIs increased by 2.8 units. Hence, prescribers should be extra-cautious while treating older adults with chronic illnesses and multiple medications.

PIMs were detected in nearly one in three of the older adults (27.4%). This finding is consistent with a study conducted in China (26.5%) [44]. However, it is lower than studies conducted in Greek (61%) [52] and Spain (73.2%) [53], and higher than a study conducted in Ethiopia (18.5%) [54]. The variation in results could be due to the differences in study setting, study design, and the versions of STOPP/START used.

The most common PIMs in this study were sulfonylureas with a long half-life with type 2 diabetes mellitus. This finding is in line with a similar study conducted in Ethiopia [54]. Sulfonylurea formulations with long half-life would put older adults at high risk of severe and prolonged hypoglycemia [55] with glibenclamide among the commonly prescribed medications in this study. The adverse effects of long-acting sulfonylureas can be intensified in older adults with hepatic and renal insufficiency and those with poor oral intake [56]. To mitigate risks associated with long-acting sulfonylureas, use of metformin, as first-line therapy, for older adults with type 2 diabetes mellitus (T2DM) is highly recommended due to a low-risk of hypoglycemia and reduced risk of cardiovascular events [55]. Whenever metformin is contraindicated or not tolerated, short-acting sulfonylureas over long-acting sulfonylureas should be preferred in older diabetic patients [57].

The second most common PIM was aldosterone antagonists with potassium-conserving drugs without serum potassium monitoring (Spironolactone/Enalapril combination in this study). This finding aligns with a similar study conducted in Bulgaria [29]. The combined use of these medications can increase the risk of hyperkalemia, leading to cardiac arrhythmias including sinus bradycardia, sinus arrest, slow idioventricular rhythms, ventricular tachycardia, ventricular fibrillation, and asystole [58]. To mitigate these risks, patients are advised to avoid potassium-rich foods and salt substitutes, and close monitoring of serum potassium when the combination is inevitable [59, 60].

Age, presence of chronic illness, and number of drugs prescribed were significantly associated with PIMs. Age was a protective factor, with each year increase in age resulting in a 5% reduction in the occurrence of PIM. Despite differences in PIM assessment tools and study setting, finding from this current study is consistent with a previous Eritrean study using AGS Beers Criteria [28]. This could be attributed to a regimen optimization with prolonged intake of the same medication over time.

The presence of multimorbidity could result in increased intake of medications which further put older adults at higher risk of PIM than their counterparts which is in line with studies conducted in Ethiopia [54], China [44], and Greek [52]. This mandates healthcare professionals to strictly monitor and manage medications prescribed to such patients to reduce risks associated with adverse events.

PPOs were detected in approximately one in eight older adults (13.3%). This figure is much lower than studies conducted in China (64.1%) [44], Greek (78%) [52], and Spain (35.5%) [53]. Variations in results could be due to differences in study design, study setting, disease pattern, and versions of STOPP/START criteria used.

The most common PPO in this study was missed use of proton pump inhibitors (PPIs) with NSAID use consistent with studies conducted in Bulgaria [29] and Spain [53]. PPIs use in older adults reduce the risks associated with NSAID-related GI adverse effects. International guidelines recommend the use of PPI in older adults taking NSAIDs [61, 62]. Moreover, the second most PPO was missed use of cardio selective beta-blockers for stable heart failure with reduced ejection fraction. Several studies and guidelines recommend the use of beta-blockers for all stable patients with current or previous symptoms of heart failure and reduced left ventricular ejection fraction as they have a mortality benefit [63].

The MRCI score was found to be a determinant of PPO where a unit increase in MRCI score resulted in 1.25-unit increase in the odds of PPO. This association can be explained by the involvement of multiple medications in a complex regimen leading to potential interactions. This requires careful attention in identifying and managing these interactions while dealing with a large number of medications and omissions can occur if interactions are overlooked or not addressed appropriately.

In this study, the mean (SD) and median (IQR) of MRCI score were 9.1 (3.7) and 8.8 (5.0), respectively. These figures were much lower than studies conducted in Indonesia (M: 17.2, SD: 11.2) [31], Portugal (M: 18.2, SD: 11.2) [30], and Turkey (Md: 11.0, IQR: 8.0) [64]. The reason for the relatively low MRCI score could be due to the difference in study area, disease burden, and national list of essential medicine across several countries. Moreover, a low MRCI can be a positive indicator of effective medication management, but it's crucial to consider the context of prescription and assess whether it reflects optimal care or PPO. However, negative outcomes associated with high MRCI can include reduced medication adherence and patient satisfaction [65,66,67]. Thus, healthcare providers should strive to ensure that patients receive the appropriate medications for their needs, regardless of the complexity of their regimens.

The number of drugs prescribed was a determinant of MRCI score, with a significant direct relationship. This finding is comparable to studies conducted in Turkey [64] and Indonesia [31]. Dose frequency was the main contributor to the overall MRCI score which is in line to similar studies conducted in Portugal [30] and USA [68]. However, it is in contrast to a study conducted in Indonesia [31] where they reported dosage form as the main contributor to the MRCI score. The difference in results could be due to difference in study setting where the current study was conducted in an outpatient setting whereas the Indonesian study was conducted in an emergency setting.

To further simplify the medication regimen complexity, where possible, healthcare professionals should avoid the prescription of multiple medications, medications with greater dose frequency, enhancing medication reviews are highly recommended. Moreover, to increase the use and applicability of MRCI tool in clinical setting, it could be automated and integrated within an electronic medical records so as to easily notify prescribers and dispensers to manage older adults’ prescriptions promoting rational use of medicine.

Older adults often face multimorbidity, requiring multiple medications to manage their health conditions. This can lead to significant drug-drug interactions, adverse drug reactions, inappropriate medication prescribing, and complexity in medication regimens [69]. Medication review and optimizations are crucial in minimizing medication-related adverse health outcomes in older adults. The significance of this study and its findings are possibly of a particular interest to countries with similar socio-economic and cultural characteristics as Eritrea.

Limitations of the study

Due to the cross-sectional nature of the study, cause-effect relationship cannot be established. The drug-drug interactions, PIMs, PPOs, and medication regimen complexity documented in this study are theoretical, their clinical effect on the ground were not assessed. Moreover, medications prescribed by other specialists or healthcare professionals that patients were still using during the study period were not considered. The lack of access to data on prescribers' behavior could hinders the identification of causal relationships between prescribing practices and outcomes such as drug-drug interactions, inappropriate medication prescribing, and polypharmacy. Older adults who filled their prescriptions outside of hospital pharmacies, lack of access to data on prescribers’ behavior may have influenced the findings. Furthermore, the findings cannot be generalized to all older adults in Eritrea. Thus, the authors recommend further broader study that assesses the rational use of medicine in the older adults treated in various clinical settings to obtain a more comprehensive understanding of the subject.

Conclusion and recommendations

The main problems identified in this study were potential drug-drug interactions, PIM, and PPO. A considerable number of older adults were encountered with polypharmacy and a low overall MRCI score was detected. Moreover, presence of chronic illness and number of drugs prescribed were significant associates of clinically significant pDDIs and PIM. Age was also a significant factor associated with PIM. Besides, MRCI score was a determinant of PPO, and the number of drugs prescribed was a determinant of the MRCI score. Dose frequency was the most significant contributor to the overall MRCI score.

To improve medication prescribing in older adults, continuous education programs targeting healthcare professionals, adherence to standard treatment guidelines and protocols, the development of rational prescribing guidelines for the older adults, introduction of electronic medical records and the integration of inappropriate prescribing and MRCI tools, as well as drug interaction checker software within the healthcare system, are highly recommended.