FormalPara Key Points

The prevalence of potentially inappropriate medications (PIMs) prescription among Qatari older adult patients attending primary healthcare (PHC) centers was high.

The most prevalent classes of PIMs that should be avoided were gastrointestinal, pain, and central nervous system drugs. While diuretics, antidepressants, and antiplatelets were the most prevalent classes of PIMs that should be used with caution.

Female sex, polypharmacy, diabetes mellitus, hypertension, cardiovascular disease, asthma, gastroesophageal reflux disease, and arthritis were significant predictors of higher PIM prescription.

1 Introduction

Potentially inappropriate medications (PIMs) are classes of drugs whose risks may potentially outweigh their benefits [1]. Despite evidence of negative outcomes in older adults, healthcare practitioners continue to prescribe PIMs [2] to the extent that it has now become a public health concern because of its association with increased morbidity, hospitalization, and mortality [3,4,5,6]. The prevalence of PIM use in older adult populations in both developed and developing countries is very high. High prevalence of PIM prescription has been shown in several countries like Sweden [7], Ireland [8], and the United States [1]. The high prevalence of PIM use in Middle Eastern countries has also been a concern [9,10,11,12,13].

Research has also identified the most common medications prescribed, including anti-hypertensive drugs [12]; benzodiazepines and nonsteroidal anti-inflammatory drugs [14]; and gastrointestinal and endocrine agents [10]. Further classes of commonly prescribed PIMs include antipsychotics and selective serotonin reuptake inhibitors [11], and pain medications and antispasmodics [9]. Several factors have been associated with PIM prescription in older adults. These include increasing age [15], polypharmacy [1, 10, 16, 17], female gender [1, 9, 17, 18], dependence on instrumental activities of daily living [19], and frailty and cognitive impairments [20]. Other factors also include certain chronic comorbidities such as hypertension, diabetes mellitus, cardiovascular disease, and mental health conditions [11].

Several approaches have been used to classify PIMs. In the early 1990s, Beers and colleagues in the USA developed lists of PIMs for older adults with updates in subsequent years [21]. The Beers criteria were developed as a strategy to reduce negative health outcomes in older adults. The 2015 Beers criteria included (1) medications to avoid for many or most older adults; (2) medication classes to avoid in older adults with certain diseases and syndromes; (3) medications to be used with caution in older adults; (4) drugs for which dose adjustment is required based on kidney function; and (5) drug–drug interactions [21, 22].

The aim of this study was to determine the prevalence and predictors of PIM prescription in older adults attending primary healthcare (PHC) centers in Qatar according to the 2015 Beers criteria.

2 Analytical and Conceptual Framework

To identify variables associated with PIM prescriptions, we adopted Andersen’s Behavioral Model for health service use as a theoretical framework. In our study, we defined health service use as patients’ access to prescription medications at any of Qatar’s 23 PHC centers. This framework proposes that an individual’s health service utilization is influenced by three categories of factors: predisposing factors (e.g., age, gender), enabling factors (e.g., employment, distance to health service), and need factors (e.g., health status, diagnoses) [23]. In this study, however, our data did not capture any enabling factors.

3 Methods

3.1 Study Design and Data Source

This was a cross-sectional, retrospective study using data from April 1, 2017 to September 30, 2017 from the Electronic Medical Record (EMR) database of the Primary Health Care Corporation (PHCC) in the state of Qatar. PHCC is a non-profit organization providing primary care to Qatar’s entire population. It delivers its services through 23 PHC centers geographically situated across the country based on population densities [31]. These centers are the primary point of contact with Qatar’s healthcare system. All healthcare services at these centers are available and accessible to all nationals of Qatar free of charge. Each of these centers serve a population made up of diverse demographic backgrounds such as education, income, ethnicity, and employment representative of Qatar’s population [24].

3.2 Study Population

The study population comprised all Qatari nationals aged 65 years and over who had medication reconciliation done according to PHCC’s established policy. This policy was introduced in January 2017 as a quality assurance measure to reduce any possible medication errors such as wrong or duplicate medication orders, or over/under doses of prescriptions. A multidisciplinary team consisting of primary physicians, nurses, and pharmacists are involved in this process.

3.3 Data Extraction

Data were extracted from PHCC’s EMR database and included three categories: (1) patient information, including date of birth, gender, and body mass index (BMI); (2) clinical data, including medical diagnoses and listed comorbidities that possibly influence the number of PIM prescriptions; and (3) medications prescribed by attending physicians. The extracted data were validated and de-identified by two medical informatics specialists. Data were initially entered in a specific form designed by the lead investigator (AA), and subsequently coded and entered into the Statistical Package for the Social Sciences (SPSS) program, version 23, for statistical analysis.

3.4 Data Security

To ensure confidentiality and privacy of the subjects, all data extracted were completely anonymized by assigning unique identifier codes for each. The electronic data were stored on a secure computer with restricted access by the lead investigator.

3.5 Measures and Definitions

3.5.1 Dependent Variables

The prevalence of PIMs in older adult patients attending PHC centers in Qatar was the dependent variable. The PIMs were identified according to the updated Beers criteria of the American Geriatric Society (AGS) 2015 by applying two of its criteria: (1) medications to avoid for many or most older adults, and (ii) medications to be used with caution (see Supplementary file in the electronic supplementary material [ESM]) [21]. The prevalence of PIM prescription was grouped under these two criteria separately. The prevalence of PIM prescription was further categorized according to the number of PIM prescriptions for each patient in each of Beer’s two criteria (i.e., one, two, or three or more PIMs).

3.5.2 Independent Variables

Several independent variables were explored in this study based on the Andersen’s Behavioral Model. These variables were (1) predisposing factors including age (in years), gender, BMI (kg/m2) and polypharmacy, and (2) needs factors including diabetes mellitus, asthma, dyslipidemia, hypertension, gastrointestinal reflux disease (GERD), cardiovascular diseases (ischemic heart disease, heart failure, arrhythmia, and stroke), arthritis (osteoarthritis and rheumatoid arthritis), and mental health conditions (depression, anxiety and dementia) [25].

We defined polypharmacy as the concurrent use of five or more medications [25]. We coded clinical diagnoses based on the International Classification of Diseases, Tenth Revision, Clinical Modification codes (ICD-10-CM codes) [26]. We recorded and classified the total number of prescribed medications for each patient according to the Anatomical Therapeutic Chemical (ATC) Classification System of the World Health Organization [27]. The fifth level ATC code was used to define the number of prescribed original drug compounds. Dermatological drugs (ATC-class D) and topical products (ATC-class M02) that are less likely to cause drug-related problems (DRP) were excluded.

3.6 Statistical Analysis

Descriptive and inferential statistics were used to analyze the data. For continuous variables, means and standard deviations (SDs) were calculated. For categorical variables, frequencies and percentages were used. Bivariate analyses were done with the chi-squared test to examine for differences in characteristics between patients with and without PIM prescriptions. A multivariable logistic regression model was conducted to examine predictors associated with PIMs. Significance level was set at p value ≤ 0.05 and a 95% confidence interval (CI). All analyses were performed using SPSS.

4 Results

4.1 Sample Characteristics

A total of 5639 Qataris aged 65 years and over were identified from the EMRs of the 23 PHC centers and included in the study. The mean age of subjects was 72.8 (± 6.5) years, and 53.8% were females. About two-thirds (68.9%) of the subjects were overweight and obese (BMI ≥ 25). Furthermore, the mean number of comorbidities among the subjects was 2.62 (± 1.14). Almost three-quarters of the study subjects (75.5%) were exposed to polypharmacy. Table 1 shows the characteristics of the study population.

Table 1 Characteristics of Qatari older adults from 23 primary healthcare centers (N = 5639)
Full size table

4.2 Prevalence and Distribution of Potentially Inappropriate Medications (PIMs)

The overall prevalence of any PIM prescription in the study sample was 76.0% (4289/5639). As shown in Table 2, the prevalence of PIMs that should be avoided for many or most older adults was 60.7% (3422/5639). Most of these patients (61.1% [2091/3422]) were prescribed one PIM, 26.9% (919/3422) were prescribed two PIMs, and 12.0% (412/3422) were prescribed three or more PIMs. The prevalence of PIMs to be used with caution among older adults was 40.6% (2291/5639). The majority of these patients (72.7% [1665/2291] were prescribed one PIM, 21.8% (499/2291) were prescribed two PIMs, and 5.5% (127/2291) were prescribed three or more PIMs.

Table 2 Prevalence of PIM prescriptions among Qatari older adults according to 2015 Beers criteria (N = 5639)
Full size table

Table 3 shows the distribution of PIM prescriptions in the two 2015 Beers criteria: medications that should be avoided for many or most older adults and medications that should be used with caution. The most commonly prescribed PIMs that should be avoided for many or most older adults were gastrointestinal medications 84.2% (2881/3422) followed by pain medications 49.9% (1709/3422). Diuretics 83.1% (1904/2291) followed by antidepressants 25.7% (588/2291) were the most prevalent PIMs to be used with caution.

Table 3 Distribution of classes of PIM prescriptions among 5639 Qatari older adults according to 2015 Beers criteria
Full size table

4.3 Factors Associated with PIMs in Bivariate Analysis

Table 4 shows the comparison between the characteristics of patients who received prescriptions for PIMs to avoid for many or most older adults and those who didn’t receive them. The prevalence of PIMs was significantly higher in females than males (64.6% vs 56.1%; p < 0.001). Also, the prescription of PIMs was significantly higher in patients who were exposed to polypharmacy compared with those who were not (70.7% vs 28.8; p < 0.001). Moreover, patients who had chronic conditions including diabetes, hypertension, cardiovascular disease, dyslipidemia, asthma, mental disorders, GERD, and arthritis compared with those without these comorbidities were all significantly associated with PIM prescription.

Table 4 Comparison of patient characteristics between those who were prescribed PIMs to avoid for many or most older adults or not prescribed PIMs
Full size table

All factors tested in the bivariate analysis were included in the logistic regression analysis except for the ‘number of chronic health conditions’ variable because it has collinearity with the health conditions included in the model. Adjusted odds ratios (AOR) and 95% confidence intervals (CIs) for predictors of PIMs to avoid for many or most older adults are shown in Table 5. Older patients with polypharmacy were more than six times more likely to have PIM prescriptions compared with those without polypharmacy (AOR 6.54; 95% CI 5.64–7.59). In addition, PIM prescription was more likely among females compared with males (AOR 1.39; 95% CI 1.23–1.56). Finally, diabetes mellitus (AOR 1.46; 95% CI 1.26–1.70), hypertension (AOR 1.21; 95% CI 1.02–1.42), cardiovascular diseases (AOR 1.38; 95% CI 1.18–1.60), asthma (AOR 1.21; 95% CI 1.04–1.41), GERD (AOR 2.99; 95% CI 2.43–3.69), and arthritis (AOR 1.30; 95% CI 1.11–1.52) were found to be significant predictors of PIM prescription.

Table 5 Predictors of PIM prescription that should be avoided for most older adults according to Beers 2015 criteria
Full size table

5 Discussion

This was a cross-sectional, retrospective study of older adult outpatients attending PHC centers in Qatar to determine the prevalence and predictors of PIM prescription. The dependent variable in this study was PIM prescription and the independent variables were predisposing and need factors according to the Andersen Behavior Model. To the best of our knowledge, this is the first study in Qatar that investigated PIM prescriptions in older adult outpatients according to the 2015 Beers criteria. A previous study in Qatar by Alhmoud et al. [11] had used the 2012 Beers criteria and involved a small sample of patients (N = 501) who received specialized home health care services in comparison with our study population.

Results in our study showed a high prevalence of PIM prescription (76.0%). Other studies in the Middle East have reported varying prevalence rates: in Qatar 38.2% [11] and in Jordan 62.5% [9]. Our study showed a very high prevalence of prescribed PIMs with 60.7% (n = 3422) in the should be avoided and 40.6% (n = 2291) in the should be used with caution categories. The prevalence rates reported in other studies varied. For instance, in the study in Qatar by Alhmoud and associates [11], the prevalence of PIMs that should be avoided was 35.0% and those that should be used with caution was 56.0%. In another cross-sectional retrospective study in Saudi Arabia of 4073 older adults in an ambulatory care setting of a large tertiary hospital, Alhawassi et al. [10] showed the prevalence of PIMs that should be avoided was 57.5% and those that should be used with caution was 37.5%. Also, in another cross-sectional study in Jordan involving 4622 older adult outpatients, Al-Azayzih et al. [9] reported 62.5% (n = 2891) of patients had at least one PIM prescribed. Of these outpatients, 69.0% (n = 1995) had prescriptions of PIMs to be used with caution [9]. Finally, in a very large population study of 523,811 older adults in Korea, You-Seon et al. [17] applied the Beers 2012 criteria (medications to be avoided for most older adults and medications to be avoided for certain diseases or syndromes) and reported that 80.96% of their subjects were prescribed at least one PIM regardless of their diagnosis or health condition [17].

The most common classes of PIM prescription in our study were gastrointestinal (84.2%) followed by pain medications (49.9%), compared with the Saudi Arabia study where 35.6% of the PIMs were gastrointestinal agents followed by 34.3% endocrine agents [10]. In the home care population study in Qatar, the two leading classes of PIMs were antipsychotics (27.4%) and selective serotonin reuptake inhibitors (16.0%) [11]. Other studies have reported different classes of PIM drugs such as nonsteroidal anti-inflammatory drugs [2, 28] and antidepressants, antipsychotics, and benzodiazepines [28,29,30]. The differences in the reported PIM prescription prevalence rates and medication classes may be related to different study populations in different settings, study designs, and versions and criteria of the Beers tool used.

Consistent with other studies, the multivariable logistic regression showed female gender to be significantly associated with PIM prescription [9, 17]. A possible reason for this may be that females are at a higher risk of developing chronic conditions [31, 32], which may lead to more inappropriate prescribing. Other significant predictors of PIM prescription our study identified were the presence of polypharmacy, diabetes mellitus, hypertension, cardiovascular diseases, asthma, GERD, and arthritis. Other studies have also shown similar predictors [6, 9,10,11]. A surprising finding was that age was not a significant predictor. As the prevalence of chronic morbidity increases with age, so does the risk of multiple comorbidities [31]. Multiple comorbidities potentially increase PIM prescriptions, which may lead to adverse events in older adults [33, 34]. However, the association of age with PIM prescriptions has been inconsistently reported across studies [1].

5.1 Strengths and Limitations

Our study had several strengths. We used a very large sample of older adults attending PHC centers in Qatar and highlighted the high prevalence of PIM prescription, which is associated with potentially serious negative outcomes in this population. While the study subjects were not a representative sample, our study allowed for the depiction of the magnitude of the practice of PIM prescription by physicians. Furthermore, using EMR records facilitated the retrieval of accurate data and avoided potential recall and selection biases.

While our study had several strengths, it also had a few limitations that need to be highlighted. For one, our study used a cross-sectional, retrospective design, which prevents the establishment of causal relationships between the dependent and independent variables. Second, the study design and the availability of EMR data restricted us to the use of only two of the 2015 Beers criteria. Application of the five criteria would have given us a more comprehensive profile of the PIM prescriptions. Third, while we captured the prevalence of PIM prescriptions, we did not confirm their actual consumption by patients. Non-adherence to prescribed medications is a common occurrence in clinical practice [35].

Regarding the Andersen’s Behavioral Model, our data had age, gender, and BMI as predisposing characteristics and several need characteristics, but no enabling ones. We recommend that in future studies in Qatar, a more fulsome spectrum of data is collected to better explain findings in accordance with the Andersen’s Behavioral Model.

5.2 Implications for Clinical Practice, Policy, and Research

The high prevalence of PIMs and their demonstrated negative outcomes in older adults highlights several important implications for clinicians, researchers, and policy makers. An essential practice should be communication and information transfer between clinicians, especially in primary care settings where patients are seen by any number of physicians. Another implication for clinicians should be the periodic critical review of prescribed medications and their continued necessity in the therapeutic management of patients. Policy makers and the leadership in PHC centers should develop best practice guidelines for clinicians for prescribing and critical medication reconciliation. There should also be corresponding mechanisms for auditing and reviewing clinicians’ prescribing practices.

In our study we focused on prevalence of PIMs, types of medications, and patient factors. It would be essential to also study clinician factors. Research has shown several factors that influence clinicians’ prescribing practices [36]. Such research in Qatar’s PHCs would provide useful data to inform policy and best practice guidelines development.

6 Conclusions

Our study showed that in older adult patients attending Qatar’s 23 PHC centers, a large number of PIMs are prescribed that should be avoided or used with caution. Logistic regression showed female gender, polypharmacy, and certain comorbidities were significantly associated with PIM prescription. The most prevalent classes of PIM prescription that should be avoided included gastrointestinal, pain, central nervous system, and cardiovascular disease drugs, while the should be used with caution category included diuretics, antidepressants, antiplatelets, and antipsychotics. With growth in the proportion of older adults in Qatar and the high risk of PIM prescription, the practice of medication reconciliation by primary care physicians should be strengthened and reinforced. Future studies should study PIM prescription in prospective cohorts to explore the occurrence of negative outcomes in this vulnerable population of older adults.