1. Introduction
Multimorbidity is the coexistence of multiple health conditions potentially aggravating each other. A systematic review and meta-analysis of 126 studies showed that the global prevalence of multimorbidity is as high as 37.2%, and over half (51.0%) of the worldwide population aged 60 years and older has multimorbid conditions [
1]. The prevalence of polypharmacy among patients of secondary-level hospital is 98%, with 5.1% having minor polypharmacy (two to three medications), 10% having moderate polypharmacy (four to five medications), and 83% having major polypharmacy (more than five medications) [
2]. Up to 17% of older adults in Germany have at least one potential drug–drug interaction (DDI) [
3]. Over half of nursing home residents (52.7%) are exposed to at least one DDI and 25.0% to more than one DDI [
4]. Prevalence of DDIs in palliative care ranges from 31 to 75% across various health care settings [
5]. In COVID-19 patients administered with ritonavir-containing therapy in the U.S., the weighted prevalence of major to contraindicated DDIs was 29.3%. Prevalence rates of DDIs among those 60 years and older with serious heart conditions, diabetes, and moderate chronic kidney disease are as high as 60.2%, 63.4%, and 80.7%, respectively [
6].
Geriatric syndromes co-exist with acquired chronic diseases and contribute to multimorbidity. Multimorbidity predisposes a person to interactions between drugs administered for treatment of involved pathologies so that the resulting risk exceeds a simple summation of risks. Drug–drug interactions may lead to adverse drug reactions (ADRs), and medical error was reported to be the third leading cause of death in the U.S. [
7]. Polypharmacy is associated with increased emergency department transfer in older long-term care residents, with the strength of association increasing with the number of medications prescribed [
8].
One of the challenges facing healthcare today is the need for an interdisciplinary team-based approach to management of cardiovascular patients with multiple health conditions. Ideally, a cardiologist should be aware of therapies administered to patients by other medical specialists such as neurologists, endocrinologists, rheumatologists, and ophthalmologists. Administration of multiple medications is often unavoidable and, as a rule, beneficial to multimorbid patients, but the risks of potentially dangerous ADRs due to serious DDIs must be avoided. Evidence gaps exist in regard to the patterns of prescribed versus taken pharmacotherapy in older cardiovascular patients, especially during and after the COVID-19 pandemic when new drug combinations were introduced into clinical practice.
Clinical pharmacology patterns of prescribed and taken medications in older cardiovascular patients may significantly depend on patient populations where physiologic and pathologic characteristics significantly vary. Due to interindividual variability, predictors of response to pharmacological treatment comprise not only gender and chronological age, but also past and present comorbidities, co-administration of medications, liver and kidney function, smoking, exercise, weight, eating and drinking behavior [
9], genetics [
10,
11], and epigenetics [
12]. Superposition of all these factors results in a unique pharmacokinetic and pharmacodynamic fingerprint of an individual.
Studies investigating the patterns of prescribed and taken medications usually focus on patient compliance and adherence to treatment and are often limited to single-center experience and a narrow spectrum of disease entities. Little is known about the overall burden of prescribed versus taken polypharmacy and DDIs across the entire spectrum of medical care encounters including outpatient visits, home visits, and hospitalizations in older cardiovascular patients.
The present study aimed to assess the patterns of DDIs and polypharmacy in older patients with cardiovascular diseases based on electronic health records (EHRs) stored in a health information system in 2019–2022.
4. Discussion
Two primary drug lists (P-List and T-List, with “P” and “T” standing for prescribed and taken medications) were established in our study to analyze the patterns of prescribed and taken medications documented in the electronic health records in the cohort of older cardiovascular patients. A sub-list analysis enabled the assessment of the patterns of DDIs and polypragmasy at the group-based and individual levels.
There are many medical decision support systems available to assess potential DDIs while prescribing pharmacotherapy [
14,
15,
16,
17,
18]. Each of these systems has its own advantages and disadvantages, and several systems may be used for in-depth assessment of a limited number of DDIs. We selected a single medical decision support system for DDI assessment. Medscape Drug Interaction Checker [
13] was chosen among other medical decision support systems because (i) it allowed stratification of the DDIs into four classes; (ii) it was user-friendly to operate; (iii) it provided information on underlying mechanisms of DDIs; and (iv) it was previously verified to be useful in assessing DDIs in cardiovascular and comorbid patients [
19,
20,
21,
22,
23].
We analyzed DDIs on a pairwise basis because there are currently no commonly recognized resources allowing the assessment of higher-order DDIs, although such techniques are emerging and seem promising [
24]. Pairwise DDI identification enabled the provision of straightforward and comprehensible illustrations, thereby contributing to better understanding of DDI patterns. In our study, median DDI number per record often exceeded the corresponding number of drug combinations because pharmacokinetics and pharmacodynamics of one pairwise drug combination involved more than one biotransformation pathway.
We developed an easy-to-calculate DDI index to take into consideration the differential impact of DDI categories ranging from contraindicated to minor. Procedures currently accommodated by institutions for reporting ADRs are as follows: as soon as a healthcare provider becomes aware of an ADR, this information is recorded, and medication causing the ADR is discontinued or dosage adjustment is performed. Current procedures for reporting DDIs remain at the discretion of health care workers taking care of patients. No standard procedures for reporting DDIs have been implemented yet. For the first time, we propose the use of the DDI index to take into consideration the strength of drug interactions (contraindicated, serious/dangerous, requiring close monitoring, and minor). We believe that the simple summation of DDIs without introducing the DDI index could lead to underestimation of clinically significant DDI burden. Without introducing the DDI index, the contribution of minor and contraindicated DDIs (i.e., insignificant and very dangerous, respectively) to the overall DDI burden is leveled or equalized. We believe that DDIs should be stratified quantitively both for scientific purposes and clinical application.
The proposed DDI index has not yet been validated. Upon validation, it may be implemented in clinical practice. Calculation of the proposed DDI index has been discussed with the institute running the study. The present research and introduction of the DDI index, in particular, represent the efforts aimed at the development of a strategy and measures to control clinically significant DDIs in our cardiovascular patients. We plan to further foster the concept of the DDI index in future research. Integrating quantitative systems’ pharmacology analysis with physiologically based pharmacokinetic models may lead to the development of more sophisticated scales. Multiscale modeling may predict potential pharmacodynamic DDIs, and, via clinical trial simulations, create testable hypotheses as to their potential clinical significance [
25]. It is essential to develop clinical decision support systems for data-driven prediction of ADRs triggered by DDIs [
26,
27]. However, it seems challenging to adequately measure the overlapping impact of DDIs, which is multifactorial and depends on genetic factors, ADR manifestation, and economic burden.
The frequency of occurrence of serious DDIs in our study was significantly higher in the case of prescribed medications compared with that among taken drugs. The highest median DDI index in our study was observed in patients with COVID-19. Notably, in the study by Spanakis and others [
28], clinically significant DDIs of “serious-use alternative” or “use with caution-monitor” management were found in 40.3% of cases upon admission, 21% during hospitalization, and 40.7% upon discharge, suggesting that the efforts of a medical team can successfully reduce the risks associated with dangerous DDIs during a hospital stay. It is clear that serious DDIs can hinder treatment response and complicate hospitalization in COVID-19 patients [
28].
Sex-related differences found in our study agree with data of the large-scale analysis showing that women have a 60% increased risk of DDI and a 90% increased risk of DDI leading to major ADR as compared to men [
27]. Female sex and older age also contribute to non-adherence to, in particular, statins [
29]. We agree that the potential effects of sex and gender on inappropriate prescribing and deprescribing remain poorly understood [
30]. Cognitive, behavioral, pharmacokinetic, and pharmacodynamic factors of adaptation underlying significantly higher scores in drug numbers, polypharmacy rates, and DDIs in women require further research.
Our study showed significant differences in the median numbers of serious DDIs per record (i.e., per single medical care encounter) in the lists of prescribed and taken medications. The mismatch of serious, requiring close monitoring, and minor DDIs was also found between the large cohort-based lists of prescribed and taken medications. This observation may indirectly suggest suboptimal treatment compliance and/or non-adherence of patients to prescribed therapy. Considering the significantly higher burden of serious DDIs among prescribed medications, the observed difference may be a sign of patient adaptation protecting them from exposure to serious DDIs.
Among the most commonly prescribed drug combinations associated with serious DDIs, the pairs of “aspirin + captopril” and “aspirin + enalapril” may be considered clinically insignificant due to the use of low-dose aspirin in the majority of cases. Administration of aspirin at doses less than 300 mg per day has little effect on the effectiveness of captopril and enalapril. Administration of aspirin in higher doses reduces the effectiveness of captopril and enalapril. Furthermore, captopril was often prescribed to be taken episodically when blood pressure remained high, despite intake of other antihypertensives; this corresponds to the guidelines of the Russian Medical Society on Arterial Hypertension (RMSAH) [
31], which recommend administration of relatively fast- and short-acting oral/sublingual angiotensin converting enzyme inhibitors (captopril, moxonidine, clonidine, and propranolol) for treatment of uncomplicated hypertensive crisis. It remains unclear whether the risk of taking these combinations may be completely dismissed considering the significant burden of polypharmacy and higher-order DDIs, which could potentially interfere with the pharmacokinetics of administered drugs. The combination “aspirin + lisinopril”, associated with serious DDIs, was among the most common in both lists (T- and P-Lists). Other common drug combinations associated with serious DDIs differed between the lists.
Drug interactions observed in some patients could directly result from the official clinical recommendations regarding the treatment of certain disease entities. British researchers from seven medical centers performed a systematic study focusing on the recommendations given in twelve national clinical guidelines [
32]. The analysis of drug–disease interactions and DDIs between medications recommended by national guidelines showed that following the guidelines may result in serious DDIs in multimorbid patients. The number of potentially serious (dangerous) DDIs reaches 30 in patients with most common multimorbidities, which poses a significant risk of serious ADRs including neurotoxicity, abnormal renal function, bleeding, and cardiovascular reactions [
32]. Randomized clinical trials underlying guidelines produce high-quality data regarding the benefits rather than the risks of taking medication in real clinical settings where people are usually frailer and multimorbid and take multiple drugs for treatment of conditions distinct from those in clinical trial populations. Studying real groups of patients improves understanding of the heterogeneity of patient populations, and allows the development of measures that provide pure benefits without the harm posed by potentially serious DDIs [
33]. Furthermore, paper versions of guidelines are challenging to integrate for people with geriatric syndromes and multimorbidity due to the overwhelming body of knowledge produced in recent years and the vast array of factors to be taken into consideration. Solving this problem would require going beyond the clinical recommendations in this category of patients.
The Working Group on Cardiovascular Pharmacotherapy of the European Society of Cardiology encourages implementation of a multidisciplinary team approach and consideration of age-related changes in the pharmacokinetics and pharmacodynamics of cardiovascular drugs to address the issues of polypharmacy [
34]. The working group considers that adherence to pharmacotherapy is a key question. It is vital to thoroughly understand the most common ADRs, practices of deprescribing [
34,
35], problems of omissions, and potentially inappropriate medications, which may require going beyond the guidelines while implementing binary or multicore team-based approaches to care for vulnerable patients [
36].
Genetic variations markedly increase or ameliorate the severity of potential DDIs and should be considered while prescribing pharmacotherapy to patients with polypharmacy. Most current guidelines on DDIs neither consider the potential effect of genetic polymorphisms in the strength of the interaction nor do they account for the complex interaction caused by the combination of DDIs and DGIs (drug–gene interactions) when there are multiple biotransformation pathways, which are referred to as DGGIs (drug–gene–gene interactions) [
10]. The increasing availability of real-world drug outcome data linked to genetic technologies and resources is likely to enable the discovery of previously unrecognized clinically significant drug–drug–gene interactions to develop clinically useful models to reduce adverse DDIs and improve drug outcomes in the setting of increasing multimorbidity and polypharmacy [
10,
11].
Our study has some limitations. First, we studied the lists of prescribed and taken medications documented in the EHRs. These lists could differ from the drugs taken by patients in reality, especially in the case of medical records documenting patient visits to outpatient facilities, as it is possible that older individuals could misreport some of the medications they take. On the contrary, records of administrated medications during hospitalizations may be more precise. Furthermore, some community-dwelling older patients could practice self-administration of over-the-counter drugs, food supplements, and herbal medications without reporting it to their health care providers. Due to the retrospective nature of our study, there was no way to verify with certainty whether the drugs documented as those patients were taking had indeed been taken by patients. Future studies involving patient surveys may contribute to solving this issue. The second limitation is caused by the fact that the health information system covers only a portion of healthcare institutions in Tomsk and Tomsk Region, and this coverage will expand in the future. Therefore, the data in this paper represent only a portion of the patient population triaged to particular healthcare providers. The third limitation of the study is due to the significant disruption in routine healthcare during the COVID-19 pandemic, which provides a rationale for continuous monitoring of the situation with pharmacotherapy patterns in the vulnerable cohort of older cardiovascular patients. The fourth limitation is the absence of data on some medications, in particular, umifenovir and favipiravir in the Medscape Drug Interaction Checker at the time of investigation. However, these drugs constituted less than 1% of the entire pool of medications administered to our cohort, so they would not make a significant difference to the overall picture. Finally, we did not study the associations between DDIs and potential ADRs in our cohort, which require further independent research.
We propose the following solutions to the problem of high DDI burden: (i) building a better structure of EHRs; (ii) patient engagement in medication diaries and ADR documentation using specially built portals linked to EHRs [
37]; (iii) identifying patients with clinically significant polymorphisms of genes involved in drug metabolism [
10,
11]; (iv) developing electronic decision-making support system for control over DDIs; and (v) an interdisciplinary approach to team building [
36]. Data such as those obtained in our study should urge the medical expert community to develop consensus guidelines for the pharmacotherapy of geriatric patients with multimorbidity.