Potentially inappropriate medications according to Marc, STOPP and PRISCUS criteria in a cohort of elderly HIV+ patients. The COMMPI project

Abstract

Introduction

The objective is to determine the prevalence of potentially inappropriate drugs according to the Marc, STOPP, and PRISCUS lists in elderly HIV patients.

Patients and methods

It was an observational, retrospective, and multicenter study. People living with HIV 65 years or older who underwent chronic concomitant treatment were included. Descriptive and multivariate analyzes were performed to study the association between polypharmacy and potentially inappropriate medication compliance.

Results

A total of 55 patients were included, 81.8% men and a median age of 69 years (IQR: 67-73). The median number of comorbidities was 3 (IQR: 2-5) and the most frequent pattern of multimorbidity was cardiometabolic (62.9%). The predominant antiretroviral treatment was triple therapy (65.5%). Polypharmacy was present in 70.9% of the patients and 25.5% had major polypharmacy. The most frequent polypharmacy pattern was cardiovascular (69.2%). The percentage of potentially inappropriate medications according to the Marc, STOPP and PRISCUS lists was 65.5%, 30.9% and 14.5%, respectively (p<0.001). Adjusted for age and sex, polypharmacy was not independently associated with potentially inappropriate medication compliance in any of the lists.

Conclusion

Polypharmacy and potentially inappropriate medications have a high prevalence. There is great variability in the percentage according to the list applied. Age, sex, and presence of polypharmacy are not predisposing factors to the presence of potentially inappropriate medications.

INTRODUCTION

HIV infection has changed a deadly condition to a chronic disease due to potent antiretroviral treatments (ART) [1]. As a consequence, the life expectancy of people living with HIV (PLWH) has increased with time, and this population is aging with a high burden of age-related chronic comorbidities and noHIV prescriptions with ART concomitantly [2,3]. This fact was already predicted by the statistical model developed by Smit et al. [4], which estimated that in 2030, 73% of PLWH will be over the age of 50, 28% will have three comorbidities, and 20% of patients will receive three co-medications. The combination between antiretroviral therapy, which consists of at least two or three drugs, and medications needed to treat comorbidities leads this population to complex polypharmacy.

Some studies described that polypharmacy has been associated with a series of consequences such as drug-drug interactions, risks of falls or fractures, and a greater presence of adverse drug events and hospitalizations [5-7]. The concept of polypharmacy is related to other concepts such as pharmaco-therapeutic complexity or potentially inappropriate medication (PIM) [8,9].

In recent years, an arsenal of tools has emerged that have a purpose to optimize pharmacological treatments in older patients by detecting drugs or families of drugs that are considered inappropriate in these patients. Some of these are Beers criteria, Screening Tool of Older Persons’ Prescriptions (STOPP)/ Screeaning Tool to Alert doctors To Right Treatment (START) criteria, PRISCUS list and Marc list among others [10-13]. Available studies have reported an extensive prevalence of benzodiazepines and high anticholinergic burden drugs among the HIV population [14,15]. Furthermore, elderly people living with HIV (EPLWH) are at increased risk of having PIM and are substantially higher than HIV-uninfected people [16-18].

At present, to our knowledge, no work has been published on the identification of PIM in EPLWH based on these three lists (STOPP/START criteria, PRISCUS list and Marc list), making a comparison between the percentage of PIM according to the different tools.

The aim of this study was to determine the prevalence of PIM according to the Marc, STOPP, and PRISCUS lists in EPLWH in pharmacotherapeutic follow-up in outpatient consultations of several Pharmacy Services.

PATIENTS AND METHODS

Design and study population. An observational, retrospective, and multicenter study has been carried out in three hospitals in Spain. PLWH aged 65 years-old and older on active ART and with chronic concomitant treatment attending at Pharmaceutical Care Consultation of the Hospital Pharmacies from March 2021 until May 2021 were included. Patients in clinical trials, treated for pre and post-exposure prophylaxis, as well as with a short life expectancy, were excluded.

Demographic characteristics, immunological and virological laboratory data, as well as pharmacotherapeutic variables were collected. Data were obtained from the patient clinical records.

Definitions

• - Polypharmacy. It was defined as the use of six or more different drugs, including antiretroviral medications [19,20].

• - Major polypharmacy. It was restricted to the use of ≥11 different drugs.

• - Polypharmacy patterns. To describe them, we used the categorization proposed by Calderón-Larrañaga et al. [21] who classified the patterns according to the type of disease they were intended to treat cardiovascular, depression-anxiety, acute respiratory infection, chronic pulmonary disease, rhinitis-asthma, pain, and menopause. To establish the corresponding polypharmacy patterns for each patient, active drugs are classified according to the anatomical therapeutic chemical classification (ATC) [22] using only the first three levels of classification and a patient was classified into a specific pattern when at least three drugs included in the pattern were dispensed.

• - Pharmacotherapeutic complexity. It was measured by the Medication Regimen Complexity Index (MRCI). The MRCI index is a validated 65-item tool that evaluates the complexity of the treatment regimen based on the number of medications, the dosage form, the frequency of the dosage and additional or special instructions. This index score ranges from 1.5 (for someone taking a single tablet or capsule taken once a day) to an undefined maximum since the score increases with the number of medications; higher scores indicate higher complexity [23]. Furthermore, according to Morillo-Verdugo et al. [20] a cut off value of 11.25 was used for the MRCI index score to consider a complex patient.

• - Multimorbidity patterns. To describe them, we used the categorization proposed by Prados-Torres et al. [24] who classified the patterns according to the type of disease they were diagnosed (cardiometabolic, geriatric-depressive, thyroid disease, and mixed). Patients were classified according to a specific pattern when they had been diagnosed with at least two diseases included in the pattern.

• - Anticholinergic risk. To determine the anticholinergic risk, we use the Drug Burden Index (DBI) [25]. According to this scale, each drug is assigned an anticholinergic load value. The anticholinergic load will be the sum of the values of the prescribed drugs. Patients are classified into patients with low anticholinergic load when they have a value ≤0.5 and high when the value is >0.5.

Statistical analysis. Discrete variables were expressed as counts (percentage) and continuous variables as means and SD or medians and interquartile ranges (IQR) as appropriate. Differences in categorical variables were calculated using a two-sided likelihood ratio Chi square test or Fisher’s exact test, and the t test or the Mann–Whitney U test was used for continuous variables, when appropriate.

A logistic regression model was performed for each of the three lists, in order to assess the association between polypharmacy and PIM according to each list, adjusting for age and sex. The threshold for statistical significance was defined as p<0.05. Data analysis was performed with SPSS v20.0.

Ethics approval. The study was approved by the ethics committee “Comité Ético de Investigación del Sur de Sevilla” (Seville, Spain) (reference 1006-N-21).

RESULTS

A total of 55 patients were included in this study and 81.8% (n=45) were male. The median age was 69 years (IQR: 67-73). The baseline characteristics of the patients are summarized in Table 1.

Table 1.

Baseline and pharmacotherapeutics features of patients (n=55).

Characteristics	N	%	Median	IQ range
Age (years)	-	-	69	67-73
Sex Male Female	45 10	81.8 18.2	- -	- -
Undetectable viral load (<50 copies/mL)	55	100	-	-
CD4 level >200 cells/mL	48	87.3	552	368-763
CD4/CD8 ratio >0.4	47	85.5	-	-
Number of comorbidities	-	-	3	2-5
Multimorbidities patterns (35 patients)
Cardiometabolic	22	62.9	-	-
Geriatric depressive	4	11.4	-	-
Thyroid disease	1	2.8	-	-
Mixed	8	22.9	-	-
ART type
NRTI + INI	29	52.7	-	-
NRTI + NNRTI	5	9.2	-	-
NRTI + IP	2	3.6	-	-
Others	19	34.5	-	-
Polypharmacy	39	70.9	-	-
Major polypharmacy	14	25.5	-	-
Polypharmacy patterns (26 patients)
Cardiovascular	18	69.2	-	-
Anxious-depressive	4	15.4	-	-
COPD	2	7.7	-	-
Mixed	2	7.7	-	-
MRCI (points)	-	-	11	7-18
High MRCI index (>11.25 points)	24	43.6	-	-
High anticholinergic load (>0.5 points)	14	25.5	-	-

ART: Antiretroviral therapy; NRTI: Nucleoside reverse transcriptase inhibitors; INI: Integrase inhibitors; NNRTI: Nonnucleoside reverse transcriptase inhibitors; IP: protease inhibitor; COPD: Chronic obstructive pulmonary disease; MRCI: medication regimen complexity index.

All study participants presented undetectable viral load, 87.3% had a CD4 level >200 cells/mL (median 552; IQR: 368-763) and CD4/CD8 ratio >0.4 in 85.5%. Globally, the median number of comorbidities was 3 (IQR: 2-5). The multicomorbidity pattern was obtained in 35 patients, cardiometabolic pattern in 62.9%, mixed in 22.9%, geriatric depressive in 11.4%, and thyroid disease in 2.8%.

Overall, 70.9% (n=39) of the patients presented polypharmacy, 25.5% (n=14) had major polypharmacy, obtaining the polypharmacy pattern in 26 of them. The most common polypharmacy pattern was cardiovascular (69.2%), followed by anxiuos-depressive (15.4%) and 7.7% for both chronic obstructive pulmonary disease and mixed pattern. The median MRCI index was 11.0 (IQR: 7.0-18.0), 43.6% (n=24) had a high pharmacotherapeutic complexity. Additionally, we found a high anticolinergic load in 25.5% of patients.

The percentage of PIM according to the Marc, STOPP, and PRISCUS lists was 65.5%, 30.9% and 14.5%, respectively (p<0.001). The most identified PIM were anxiolytics (18.2%) with STOPP criteria, sedative-hypnotics (7.3%) with the PRISCUS list and beta-blockers (27.3%), antiplatelet agents (25.5%), anxiolytics (25.5%) and hypoglycemic agents (23.6%) with the Marc list. The distribution of the criteria for each list is collected in Table 2.

Table 2.

Distribution of the STOPP, Marc and PRISCUS criteria in the study population

STOPP CRITERIA	N	%
Anxiolytics lasting more than 4 weeks	10	18.2
Antiulcer drug not indicated	3	5.5
Inadequate antithrombotic	2	3.6
Antipsychotic used for more than 4 weeks	2	3.6
NSAIDs and long-term antirheumatics	2	3.6
Beta blockers	2	3.6
Oral hypoglycemic agents excluding insulin	1	1.8
Antidepressants	1	1.8
MARC LIST	N	%
THERAPEUTIC GROUPS
Beta-adrenergic blockers	15	27.3
Benzodiazepines and analogs	14	25.5
Platelet antiaggregant	14	25.5
Oral hypoglycemic agents	13	23.6
Insulins	5	9.1
Opioids	5	9.1
Loop diuretics	3	5.5
NSAIDs	3	5.5
Oral anticoagulants	2	3.6
Antipsychotics	2	3.6
Immunosuppressants	1	1.8
SPECIFIC MEDICINES
Spironolactone/eplerenone	3	5.5
PRISCUS LIST	N	(%)
DRUGS
Sedative-hypnotic	4	7.3
Anxiolytics	2	3.6
Antidepressants	1	1.8
NSAIDs and antirheumatics	1	1.8

NSAIDs: Non-steroidal anti-inflammatory drugs

A univariate analysis was performed to determine the association between the presence of PIM by any of the three listed with the following qualitative variables: polypharmacy, major polypharmacy, polypharmacy pattern, and multimorbidity pattern. The results are shown in Table 3.

Table 3.

Univariate analysis of the presence of potentially inappropriate medication and pharmacotherapeutic variables

Pharmacotherapeutic characteristics	STOPP Compliant n (%) vs. non-compliant n (%) p-value	Marc Compliant n (%) vs. non-compliant n (%) p-value	PRISCUS Compliant n (%) vs. non-compliant n (%) p-value
Polypharmacy (n=39)	14 (82.4%) vs 25 (65.8%) p=0.067	28 (77.8%) vs 11 (57.9%) p=0.067	7 (87.5%) vs 32 (68.1%) p=0.414
Major polypharmacy (n=14)	8 (47.1%) vs 6 (15.8%) p=0.021	14 (38.9%) vs 0 (0%) p=0.001	3 (37.5%) vs 11 (23.4%) p=0.405
Polypharmacy pattern (n=26)	11 (64.7%) vs 15 (39.5%) p=0.003	24 (66.7%) vs 2 (10.5%) p=0.003	5 (62.5%) vs 21 (44.7%) p=0.510
Multimorbidity pattern (n=35)	13 (76.5%) vs 22 (57.9%) p=0.335	29 (80.6%) vs 6 (31.6%) p=0.003	7 (87.5%) vs 28 (59.6%) p=0.064

When analyzing the association between polypharmacy and STOPP list, a higher frequency is observed in those patients who comply with the list (14; 82.4% vs. 25; 65.8%; p=0.067). On the contrary, statistical significance is reached in major polypharmacy, in favor of a higher proportion of polypharmacy in patients with compliance with list (8; 47.1% vs. 6; 15.8%; p=0.021). Regarding the patterns, there is a higher frequency in patients who presented both a polypharmacy pattern (11; 64.7% vs 15; 39.5%; p=0.003) and a multimorbidity pattern (13; 76.5% vs 22; 57.9%; p=0.335), compared to those who do not meet any criteria, being only statistically significant in the case of polypharmacy pattern.

Acording to polypharmacy and compliance with the Marc list, we again observed a higher proportion of patients with polypharmacy (28; 77.85% vs 11; 57.9%; p=0.067). Nevertheless, statistically significant differences were observed in major polypharmacy (14; 38.9% vs 0; 0%; p=0.001) and both a polypharmacy pattern (24; 66,7% vs 2; 10,5%; p=0.003) and a multimorbidity pattern (29; 80,6% vs 6; 31,6%; p=0.003).

Finally, when observing the association between polypharmacy and the PRISCUS list, we again found a higher prevalence in patients who comply with the list (7; 87.5% vs 32; 68.1%; p=0.414). The same occurs with the presence of major polypharmacy (3; 37,5% vs 11; 23,4%; p=0.405). In the case of polypharmacy pattern (5; 62.5% vs 21; 44.7%; p=0.510) and multimorbidity pattern (7; 87.5% vs 28; 59; 6%; p=0.064), both are present in a greater proportion in patients with compliance with the PRISCUS list, however no statistical significance is observed in any case.

Pharmacotherapeutic complexity were also compared for the three lists. Statistical significance was only found in the Marc list, being 17.5 for the compliant versus 6.9 for the non-compliant (p=0.001).

In the multivariate analysis, we analyzed the association of PIM compliance and polypharmacy for the three lists adjusted by age and sex, and polypharmacy; it turned out to be independent in the Marc, PRISCUS, and STOPP lists, with OR=3.14 (IC95% 0.82-11.96), OR=2.93 (IC95% 0.28-30.77) and OR=0.60 (IC95% 0.10-3.57) respectively. Given the low incidence of the event and the small number of patients in the study, it was not possible to adjust for more variables.

DISCUSSION

Our study shows a high prevalence of PIM according to the lists used in the cohort analyzed, with great variability in the identification of this concept among the different tools. We found the highest proportion of PIM when using the Marc list and the lowest one with the PRISCUS list.

Our work includes patients with a high percentage of polypharmacy. The high prevalence of polypharmacy in PLWH has been widely described in international studies [26,27] and national studies [28,29]. Furthermore, some studies reported that polypharmacy is more widespread in the EPLWH population than in the general population without HIV infection [3,30].

Specifically, the multicenter study by Gimeno-Gracia et al. [28], with a slightly higher cohort of patients of the same age group, presented a polypharmacy result very similar to our study.

In addition, many authors have described which concomitant medications were prescribed most frequently prescribed; cardiovascular drugs and central disorders of the nervous system are the most prevalent [31,32]. This fact is reflected in the present study. The most observed polypharmacy pattern in our cohort is the cardiovascular pattern, followed by the depressive-anxious pattern. Consistent with this, the most identified multimorbidity pattern is cardiometabolic.

Concomitant medication is known to be the cause of complexity pharmacotherapeutic in EPLWH. ART contributes to a lesser extent due to the simplification of dosage guidelines in recent years [33]. In our study, almost half of the patients present a high complexity index, with a median of 11.0 points. There are other published studies, such as the one by Contreras et al. [34], where a lower mean complexity index (6.9 points), with 20.6% of patients with a high complexity index. This difference may be due to the fact that the study was carried out in HIV patients older than 18 years. Younger patients usually present a lower pharmacotherapeutic complexity than older people due to the absence of comorbidities.

As already mentioned, some authors have described a high prevalence of PIM in PLWH. Among these PIM they found benzodiazepines and drugs with anticholinergic properties [15]. In this study, there is also evidence of a widespread prescription of benzodiazepines; it is the group of drugs most identified as PIM by all three tools. However, due to the low value that benzodiazepines contributes to the anticholinergic load, we only observed 25.5% of patients with a high anticholinergic load. This result is not insignificant, since there are other studies of a similar population with lower percentages of high anticholinergic load, 15% [15].

Polypharmacy, major polypharmacy and polypharmacy and multimorbidity patterns were found to be more common in patients whose compliance with any of the three lists was high. It seems logical to think that presenting different drugs for different comorbidities will allow them to be included within the same pattern of polypharmacy and diseases within the pattern of multimorbidity. Additionally, this means that the specific clinical characteristics that define any of the criteria collected by the three tools are more likely to be met.

On the other hand, multivariate analysis shows that this compliance is not affected by the variables age, sex and poly-pharmacy.

Regarding the variability in the identification of PIM by the different lists, we might think that the criteria included in them do not quite fit the PLWH. Similarly, other authors have evaluated the concordance of criteria for four lists for the identification of PIM in the older population, obtaining great variability of them [35].

It is important to note that the characteristics of each of the tools used in this study differ from each other, and maybe because of this, we have found variability in the identification of PIM. In the case of the Marc list, it only takes into account the prescription of drugs belonging to different groups of therapeutics that are considered high-risk for chronic patients. This list does not describe the clinical situation in which that drug is used or the duration of treatment. In contrast, the STOPP criteria and PRISCUS lists are more specific and demanding lists. These lists detail the use of PIM in specific clinical situations or mention the duration of treatment, making it more difficult for the patient to meet some criteria. Despite this, Aguiar et al have described the limitations of the STOPP criteria [36].

We recognize some limitations. First, the underestimation of PIM prescription, not including medications not considered chronic (less than 90 days of prescription) or OTC (over-the-counter medication) or recommended in the community pharmacy for minor illness or minor symptoms. However, they are considered to not have a relevant effect because they are used sporadically or for less than three months. Another possible limitation could be the small number of patients who meet the inclusion criteria. This can lead to inconsistent results and limits us in the multivariate analysis that adjusts only for sex, age, and polypharmacy. However, this limitation could be overcome with increasing sample size. Thus, it does not invalidate the results obtained, which can serve as hypotheses for future lines of research.

Given the design of the study that does not contemplate patient follow-up, it is unknown whether the PIMs detected may have had any negative effect on the study population.

The results obtained in this study can provide guidance on which tool may be more sensitive to detect the use of PIM. Additionally, seeing the variability of the data, another possible line of research would be the development and validation of a PIM list for the elderly HIV population.

There is great variability in the prevalence of potentially inappropriate medications depending on the tool used. This finding should lead to the design of a tool more sensitive to detect PIM in the elderly HIV-infected population.

FUNDING

None to declare.

CONFLICT OF INTEREST

The authors declare no conflict of interest.