Use of potentially inappropriate medications in institutionalized elderly: prevalence and associated factors

Resumo Este trabalho teve como objetivo verificar a prevalência do uso de medicamento potencialmente inapropriado (MPI) para idosos residentes em Instituições de Longa Permanência para Idosos (ILPI), assim como os tipos de medicamentos e seus fatores associados. Trata-se de estudo transversal realizado em 10 ILPI da cidade do Natal, entre outubro e dezembro de 2013. Os medicamentos inapropriados foram classificados de acordo com os Critérios de Beers da American Geriatric Society 2015. Foram analisadas variáveis sociodemográficas, relacionadas à ILPI e às condições de saúde. Para as análises univariada e multivariada entre a variável principal (uso de MPI) e as demais foi utilizada a regressão de Poisson. A amostra foi composta por 321 idosos e, destes, 304 utilizavam medicamentos. A prevalência de uso de MPI foi de 54,6% (IC 95%: 48,9-60,2) e no modelo final esteve associada à polifarmácia e à demência. As classes terapêuticas de MPI mais identificadas foram antipsicóticos e benzodiazepínicos. O estudo revelou alta prevalência de MPI entre os idosos das ILPI, configurando a necessidade de adoção de indicadores do uso destes medicamentos e a implantação de estratégias que tornem a farmacoterapia mais segura e adequada aos idosos. Palavras-chave Idoso, Instituição de longa permanência para idoso, Prescrição inapropriada, Lista de medicamentos potencialmente inapropriados Abstract This study aimed to verify the prevalence of the use of Potentially Inappropriate Medications (PIM) for elderly living in Long-Term Care Institutions for the Elderly (LTCIE), as well as the types of medications and their associated factors. This is a cross-sectional study carried out in 10 LTCIEs in the city of Natal in the period October-December 2013. Potentially inappropriate medications were classified according to the 2015 American Geriatric Society Beers Criteria for Potentially Inappropriate Medication Use in Older Adults 2015. Sociodemographic, LTCIE-related and health-related variables were considered. Univariate and multivariate analyses were performed between the primary variable (PIM use) and the independent variables using Poisson regression. The sample consisted of 321 older people, of whom 304 used medications. The prevalence of PIM use was 54.6% (95% CI: 48.9-60.2) and was associated with polypharmacy and dementia in the final model. The most common PIMs were antipsychotics and benzodiazepines. The study revealed a high prevalence of PIM use among the elderly of the LTCIEs, evidencing the need to adopt indicators on the use of these drugs and the implementation of strategies that make drug therapy safer and more adequate for older adults.


introduction
The population aging process is a global phenomenon and is associated with many challenges concerning public health policies 1,2 .While aging is not a synonym for dependence, increased longevity is often associated with an increased number of chronic diseases, physical, cognitive and mental disabilities, as well as drug consumption 3,4 .The greater use of drugs by the elderly increases the risk of adverse health outcomes, such as adverse drug reactions, drug interactions, non-adherence to therapy, functional decline and geriatric syndromes [5][6][7] .Elderly patients are more susceptible to adverse events due to age-related physiological changes that may influence the pharmacokinetics and pharmacodynamics of medicines, with particular reference to hepatic elimination and renal excretion.Consequently, such changes will affect the choice, dose and frequency of drug administration 4,6,7 .
The risk-benefit ratio of each drug should be considered to minimize the occurrence of adverse drug reactions in the elderly.Thus, certain drugs are classified as potentially inappropriate for use in the elderly, when the risk of causing adverse events exceeds the benefit expected for the patient or when a safer, better tolerated or more effective alternative is available [8][9][10] .Lists of Potentially Inappropriate Medications (PIM) for the elderly have been developed and published to facilitate the adaptation of the drug therapy to the elderly and to help health professionals prescribe drugs more safely.Beers and collaborators pioneered the development of this systematic cataloging of PIMs.Since then, this criterion has been one of the most widely used specific methods for assessing PIM use and was revised in 1997, 2003, 2012, and the most current version was published in 2015 by the American Geriatric Society (AGS) -AGS/Beers 2015 9,10 .
Studies indicate that the use of PIMs is associated with an increased risk of falls, fractures, hospitalizations and mortality in the elderly 5,6,8,11 .Due to such adverse events and the adverse effects on quality of life, the use of PIM can be considered a public health problem, also impacting higher health costs 12,13 .
Individuals living in Long-Term Care Institutions for the Elderly (LTCIE) have a higher morbidity burden, use more drugs than community-dwelling elderly and are therefore more susceptible to the use of PIMs and adverse drug events [14][15][16] .A systematic review has estimated that 43% of older adults living in asylum institutions are exposed to PIMs 17 but estimates of PIM prevalence in LTCIEs vary widely because many factors influence PIM use, such as individual variability (morbidity of residents, prescription habits) and the differences in the specific regulations of each region or country (organization and structure of LTCIE, experience and team composition) 18 .
Some studies that aimed to measure the prevalence of PIMs were performed in Brazilian South and Southeast regions.However, most of them included only community-dwelling elderly and used pre-2015 versions of the AGS/Beers criteria [19][20][21][22][23][24] .The investigations conducted at the LTCIE included only non-profit institutions, less than 300 seniors and used the previous version of the 2015 AGS/Beers criteria [25][26][27] .
Based on these findings and the understanding that avoiding PIM is a relevant, simple and effective strategy to improve patient quality and safety 9,28,29 , it is essential to determine the size, nature and relevance of the use of PIMs in Brazilian institutions.Given this context, this study aimed to verify the prevalence of PIMs for seniors living in LTCIEs adopting the 2015 AGS/Beers Criteria, the types of medications and to check the associated factors.

Methods
This cross-sectional study was conducted between October and December of 2013 in LTCIEs registered under the health surveillance of the city of Natal, Rio Grande do Norte (RN), Brazil.The study population included all subjects residing in LTCIEs who were 60 years of age or older who agreed to participate in the study.Individuals who were not at the LTCIE at the time of the study due to hospitalization or who were terminally ill were excluded from the investigation.The age of 60 years was adopted, in line with the WHO recommendation that, for developing countries, individuals older than 60 should be considered as elderly 30 .
The information was retrieved from the medical records of the elderly and complemented by interviews with the LTCIE health professionals' teams.Elderly caregivers were responsible for answering questions about functional capacity and falls.A structured questionnaire tested before the investigation was used for interviews.The collection was carried out by postgraduate and undergraduate students strictly trained and supervised by the research team.The primary variable of this study was the prescription of one or more PIMs according to the 2015 AGS/Beers criteria 9 .Initially, each medication was evaluated for inclusion among those in the list of PIMs that should be avoided by most of the elderly, PIMs that should be avoided by the elderly with specific diseases or clinical syndromes and a list with clinically significant potential drug-drug interactions between drugs without anti-infective activity that should be avoided in the elderly.The inclusion of the drug in one or more of the above criteria characterizes it as PIM.
Beers' criteria are specific tools, widely used for the identification of PIMs in elderly patients.The American Society of Geriatrics updated the "Beers Criteria for Potentially Inappropriate Medication Use in Older Adults" in 2015 (AGS/ Beers 2015) 9 .The update was carried out from an extensive literature review, respecting the principles of evidence-based medicine, and was elaborated by a panel of specialists consisting of physicians, pharmacists and representatives of international organizations 9,10,31 .The independent variables used in the analysis were classified into three groups: -Sociodemographic: age, gender, level of schooling, number of children and use of private health insurance.
-Related to the institution: type of LTCIE (private for-profit or non-profit), the number of residents per caregiver and length of residence at LTCIE.
-Clinical and health status: occurrence of chronic conditions, number of chronic conditions, depressive symptoms, cognitive status, functional independence and mobility, the occurrence of falls and fractures in the 30 days before data collection, nutritional status, urinary and fecal incontinence, use of medications, polypharmacy and medications used.
Information on chronic diseases was obtained from the medical records of the elderly, considering the following conditions: hypertension, diabetes, CVA, dementias (including Alzheimer's disease), Parkinson's disease, osteoporosis, renal failure, cardiovascular disease, rheumatic disease, mental illness and dyslipidemia.Depressive symptoms were assessed by the Geriatric Depression Scale (GDS-15) 32 .The cognitive state was verified by the Pfeiffer's Short Portable Mental Status Questionnaire, which evaluates short and long-term memory, orientation, information on daily activities and mathematical ability (intact mental function, mild, moderate and severe cognitive impairment) 33 .The Barthel index was used to evaluate the functionality (independence, and mild, moderate, severe and total dependence) 34 .The evaluation of urinary incontinence (UI) or fecal incontinence was performed based on Section H, version 2012, of the Minimum Data Set 3.0 instrument, considering the definition of UI recommended and indicated by the International Continence Society (ICS) in 2002, which defined it as any involuntary loss of urine 35 .For nutritional evaluation, previously trained and calibrated nutritionist researchers performed the Mini Nutritional Assessment (MNA), which facilitates the classification of the elderly into three distinct groups: individuals with adequate nutritional status, at risk of malnutrition and with malnutrition 36 .The drugs recorded and administered daily were characterized as to the therapeutic class.We considered polypharmacy and excessive polypharmacy as the concomitant use of five or more and ten or more drugs, respectively 37 .
The descriptive analysis of the data included the calculation of absolute and relative frequencies and the means and standard deviation (SD).The univariate and multivariate analyses were based on the prevalence and respective 95% confidence intervals, estimated by the Poisson Regression model with robust variance.Poisson Regression analysis included any variables with p values less than 0.20 in the univariate analysis.A significance level of p < 0.05 was the criterion adopted to identify variables independently associated with the use of PIMs in the multivariate model.
This study is part of the research project "Human aging and health: the reality of institutionalized elderly in the city of Natal (RN)", approved by the Research Ethics Committee of the Federal University of Rio Grande do Norte.The informed consent form was signed by the resident or legal guardian, the caregiver and the LTCIE's director, following the assumptions set out in the Declaration of Helsinki.

results
Ten (71.4%) of the 14 LTCIEs registered under the health surveillance of the city of Natal (RN) participated in the investigation.Four (28.6%)LTCIEs refused to participate in the study.Of the ten participating LTCIEs, five were philanthropic institutions and five were private for-profit institutions.Of the total residents of participating LTCIEs, eight (2.4%) were excluded from the study; six (1.8%) because they were hospitalized during the collection period, one (0.3%) due to terminal illness and one was under 60 years of age.The final sample consisted of 321 individuals aged between 60 and 107 years, and most of them were female.A total of 204 (63.8%) elderly individuals resided in non-profit private LTCIEs and 118 (37%) used private health plan.Regarding the health status, most of the elderly, namely, 283 (88.2%), were carriers of some chronic disease.Hypertension, diabetes and dementia were the most prevalent chronic conditions (48.9%, 25.2% and 24.6%, respectively).Table 1 shows the characteristics of the subjects participating in the study.
Of the 321 elderly participants, 94.7% (n = 304) used at least one drug continuously.The mean number of medications used was 4.5 (range 0-14), and 47.0% (n = 143) were exposed to polypharmacy (Table 2).Concerning the group of 304 older adults, 1,440 medications were prescribed, most of them with action on the Central Nervous System (36.8%).Among the drugs, antipsychotics (n = 207) and antidepressants (n = 94) were the most prescribed for the elderly.
Among the elderly patients taking medication, 54.6% (CI 48.9-60.2) used at least one PIM, where 54.6% (166) used a PIM that should be avoided by most of the seniors, 31.3%(52) used at least one PIM that should be avoided by elderly patients with specific clinical diseases or syndromes, and in 9.2% (28) older adults, clinically significant potential drug-drug interactions were observed between drugs without antiinfective activity that should be avoided in the elderly described in the 2015 AGS/Beers criteria.Regarding the therapeutic classes of drugs identified as PIMs, it was observed that antipsychotics, benzodiazepines and sulfonylureas were the most frequent.A total of 225 PIMs that should be avoided by most of the elderly, 79 PIMs for older adults with specific diseases or syndromes due to disease-drug or drug-syndrome interactions and 28 potentially clinically significant drug-drug interactions between drugs without antiinfective activity which should be avoided in the elderly were identified.The data are shown in Table 2.
The use of at least one PIM per elderly was considered for the analysis of associations.The univariate analysis revealed significant associations between PIM use and polypharmacy, dementia, residing in for-profit LTCIE, having private health insurance, and having diabetes (Table 3).The other variables tested (age, children, marital status, elderly/caregiver relationship, hypertension, stroke, Parkinson's disease, osteoporosis, cardiovascular diseases, mental illness, dyslipid-emias, cognitive status and MNA) did not show significant associations, and only those with p < 0.20 were included in the multivariate analysis.The occurrence of collinearity was verified before the multivariate analysis, and the variables "having a health plan" and "residing in for-profit LTCIE" were collinear.We chose to include only the "Type of LTCIE" variable in the model.The final model of the analysis revealed that PIM use was significantly associated with polypharmacy and dementia (Table 3).

Discussion
The prevalence of PIMs use was high in the LTCIEs investigated in this study, evidencing a positive association with polypharmacy and dementia.The therapeutic classes that determined the high prevalence of PIMs were antipsychotics, benzodiazepines and sulfonylureas, in agreement with previous studies.Brazilian studies have a wide variety of PIM-related prevalence levels, ranging from 24.6% to 82.6% [19][20][21][22][23][24][25][26][27] .These studies collected data in different contexts, such as domiciles 20,22,23,38 , basic care services 21,24 , and they vary in sample size, which justifies this wide prevalence variation.Similarly, Brazilian studies conducted specifically in LTCIEs also show a wide discrepancy among prevalence levels (29.2 to 82.6%) [25][26][27]39 . In hese cases, the reduced sample size 26,27,39 and the implementation of the study in only one institution 26,39 decreased the generalization potential.However, in addition to these factors, we considered the adoption of different versions and the partial use of the Beers Criteria lists as the primary source of bias for studies of this nature.
Many changes are implemented as a result of the updates between the versions of the Beers criteria, such as the inclusion, removal or reclassification of medicines in the various lists.Thus, the same population may have different prevalence levels depending on the version used.The tool is built with specific criteria that may not be used in full to characterize the use of PIMs.Varallo et al. 39 and Nascimento et al. 26 used the Beers-Fick criterion published in 2003, whereas the studies conducted by Vieira Lima et al. 25 and Alves-Conceição et al. 27 used the Criteria de Beers published in 2012 and considered for the assessment of the PIMs all the three lists of the document (PIMs that should be avoided by most of the elderly, PIMs due to drug-disease or drug-syndrome interactions that may deteriorate the disease or syndrome and medicines that should be used with caution).
Our research adopted the 2015 AGS/Beers criteria, which is the most updated version and one of the tools most used by health professionals 10,40 .This version has been revised from the principles of evidence-based medicine with extensive literature review; it includes the analysis of drug interactions and evidence of new drugs and conditions not available at the time of the 2012 version.Also, it is more specific and is organized by medical specialty and pathology, facilitating a quick and practical use 9,10,31,38 .
In our study, we chose to carry out a comprehensive evaluation and to apply three lists of the 2015 AGS/Beers criterion, since we had data that are generally not available in other works and are essential for PIM assessment, such as functional capacity, nutritional status, occurrence of falls, urinary incontinence and cognitive status 20,23,26 .Thus, our prevalence of 54.6% was obtained through the broader and more detailed use of the 2015 AGS/Beers criteria.It is important to note that in a systematic review involving 26 American and European studies (227,534 elderly), we estimated a PIM prevalence in LTCIEs of 43% (95% CI: 37.3-49.1) 17.
Thus, we also call attention to the consideration of drug interactions as a PIM criterion, which in our study aimed to widely disseminate medications and interactions cited in the 2015 AGS/Beers criterion, thus making this essential educational tool known by professionals, managers and health services, so that they can be considered in the evaluation of the elderly who require medicines 41 .As a result, we chose to show the result with the drug interactions that were included in the 2015 update.
In our investigation, most of the elderly residing in private for-profit LTCIEs (88.8%) used health insurance plans, whereas the non-profit LTCIE elderly residents essentially used the SUS network.The univariate analysis showed that these two variables have a statistically significant association between them, to the point of being considered collinear.Therefore, only one of them was included in the multivariate analysis of the association with PIM use.No difference was observed between the LTCIE's administrative profile (for-profit/non-profit) regarding the use of PIMs following the multivariate analysis.
Regarding the factors associated with the use of PIMs, polypharmacy stands out and has been identified as the most frequently associated factor to increase the probability of receiving PIMs.Our result is consistent with other studies that found the same association 18,25,26,42,43 .One of the possible explanations for such a finding is that polypharmacy is triggered by the so-called prescription cascade, and several classes of drugs associated with the phenomenon of the prescription cascade include agents that are often considered PIMs, such as psychotropics 43 .In the lit-erature, polypharmacy has been associated with an increased risk of adverse events, as it entails a greater complexity of drug therapy and favors the occurrence of medication errors 26,27 .
Contrary to several studies that suggested that cognitive impairment and dementia were associated with a lower probability of receiving PIMs 17,44,45 , our investigation found that PIMs were more likely to occur among seniors with dementia.Most patients who have dementia show behavioral and psychological symptoms, many of which are particularly detrimental to the patient and their caregivers, such as agitation, aggression, oppositional behavior, delusions or hallucinations.Antipsychotics are commonly used for the treatment of psychotic symptoms and aggressiveness, and together with benzodiazepines were the most prescribed PIMs in the study, although there is only modest evidence of its efficacy and lack of regulatory approval for use in dementia 46 .The use of antipsychotics is strongly associated with adverse outcomes such as mortality and stroke, especially among individuals with dementia 46,47 .Because of this, many regulatory agencies have published, as of 2004, alerts on the risks of using antipsychotics by older adults with dementia 46 , and this may be a justification by which, in international studies, no association was found between PIM use and dementia.Dementia is a non-modifiable factor but serves as a warning about the need for a continuous drug therapy review in this type of patient and greater precaution in the prescription of new drugs, to minimize the inclusion of PIMs and polypharmacy whenever possible.
Another class of PIMs much prescribed in our study, that of benzodiazepines has been reported in the literature as the most commonly identified as PIM.This finding is of concern, since benzodiazepines, especially long-acting ones, are more likely to cause residual sedative effects and increased risk of falls, cognitive impairment, dependence, and delirium 48 .Such occurrences become more noticeable due to age-related physiological changes that modify pharmacokinetics and pharmacodynamics, prolonging the half-lives of benzodiazepines and increasing the risk of adverse events 6,13,48 .The use of benzodiazepines in the elderly is only recommended if safer alternatives are not available 9,48 .
In order to reduce the inappropriate use of psychotropic drugs in LTCIEs, international studies recommend that institutions provide the elderly 28,49 with behavioral therapy programs, meaningful activities and physical activity pro-grams.The use of psychotropics and the occurrence of polypharmacy are suggested as indicators of the quality of care in LTCIEs 50,51 .
The systematic review of drug therapy has been suggested as a standard practice in LTCIEs, with the implementation of deprescription programs and the use of quality indicators to monitor actions 50,51 .
The results of this study should be interpreted taking into account some limitations.First, we highlight the cross-sectional design that may have underestimated the PIM estimate by not allowing the evaluation of some drugs that are considered PIMs when used for an extended period, such as proton pump inhibitors, which are considered inappropriate when used for more than eight weeks.
Another limitation that may have led to the lower prevalence estimate was the unavailability of some clinical information of the elderly.Such information is necessary for a more accurate evaluation of the drug-disease interaction and, therefore, of the assessment of the use as potentially inappropriate by the 2015 AGS/Beers criteria (e.g., the absence of creatinine clearance).Finally, the Pfeiffer test is not yet validated in Brazil, but it is less demanding than other instruments such as Mini-Mental State Examination, which could have caused a "ground" effect due to the high overall sample weakness 15 .
The main strengths of the study are the sample size, which included 71.4% of all LTCIEs in Natal (RN) and the inclusion of for-profit and non-profit institutions, reducing selection bias.The use of the 2015 AGS/Beers Criteria 9 is also a strength because it has been developed according to updated scientific evidence.

conclusion
The study has shown that the use of PIMs is high among residents of LTCIEs, mainly of long-acting antipsychotics, benzodiazepines and sulfonylureas.Polypharmacy and dementia were the characteristics most associated with the use of PIMs, regardless of the type of institution (for-profit or non-profit).The prevalence of PIMs of this magnitude reveals the need to improve the quality of the drug therapy of the elderly living in LTCIEs and requires the actions of professionals and managers.

collaborations
FSM Moreira, J Jerez-Roig, LMBM Ferreira, APQM Dantas, KC Lima and MAF Ferreira contributed to the conception, design, analysis and interpretation of data, the writing of the article and the revision of the version to be published.

table 1 .
Distribution of the frequencies of sociodemographic and health variables related to institutionalized elderly.Natal, RN, Brazil, 2018.

table 1 .
Distribution of the frequencies of sociodemographic and health variables related to institutionalized elderly.Natal, RN, Brazil, 2018.

table 3 .
Univariate and multivariate analysis of the association between sociodemographic characteristics, health conditions and use of potentially inappropriate medications in institutionalized elderly.Natal (RN), Brazil, 2018.Barthel Index.** Poisson Regression significant when p<0.05.***Prevalence ration (95% CI) estimated by the adjusted Poisson Regression.