Relationship between anticholinergic load and self-perceived health in a population aged 44 years or older

Relação entre carga anticolinérgica e autopercepção de saúde em população com mais de 44 anos

Eliz Cassieli Pereira Pinto

Universidade Estadual de Londrina, Brazil

Ana Maria Rigo Silva

Universidade Estadual de Londrina, Brazil

Marcos Aparecido Sarria Cabrera

Universidade Estadual de Londrina, Brazil

André de Oliveira Baldoni

Universidade Federal de São João del-Rei, Brazil

Daniela Frizon Alfieri

Universidade Estadual de Londrina, Brazil

Renne Rodrigues

Universidade Estadual de Londrina, Brazil

Edmarlon Girotto edmarlon78@gmail.com

Universidade Estadual de Londrina, Brazil

The increase in life expectancy in the last century is due to several sociodemographic and health factors, including access to health services and the development of modern therapies.1 Parallel to the aging of the population, care for chronic health conditions has increased, together with the use of drugs.2

If, on the one hand, drugs are essential in the recovery and maintenance of health, on the other hand, they pose a risk of adverse drug reactions (ADRs), facilitated by the physiological changes inherent to aging. Examples of such effects are decreased muscle mass, body water, and liver and kidney metabolism, which may favor the accumulation of toxic substances and thus the onset of ADRs.3 Despite the effects that organic changes resulting from aging cause in the response to drugs, pharmacological intervention is still the most common therapy in older people care.4

Inappropriate prescribing is a public health problem given its association with morbidity and mortality, with additional costs to health services resulting from ADRs.5 In many cases, the treatment of an ADR leads to the addition of new drugs to the pharmacotherapy, thus increasing the risk of polypharmacy and iatrogenic cascade.6 In the geriatric field in particular, a comprehensive evaluation of pharmacotherapy is essential.7 In this scenario, especially in recent decades, criteria and recommendations aiming for more judicious use of drugs have been developed.5,8 Recently, drugs that exert anticholinergic action have received special attention from researchers.9,10 Criteria and lists have been published in an attempt to estimate a patient’s anticholinergic load (ACL) and provide support for a more cautious use of these agents,10 since they have been associated with several clinical outcomes.9

Self-perceived health has been used as a valid marker of morbidity and mortality and of general health status.11 Self-assessment is one of the main factors that lead people to seek medical care,12 and this can increase their use of health services13 and even influence the assessment performed by professionals.12 Thus, the objective of the present study was to investigate the association between ACL and self-perceived general health in middle-aged and older adults in a medium-sized municipality in southern Brazil.

This cross-sectional and analytical study was based on data from the VigiCardio project (2015), a broader survey that examined cardiovascular diseases in the state of Paraná, including mortality, risk profile, drug therapy, and complications.14 The baseline data of the VigiCardio project were collected in 2011 using a population-based design in adults aged 40 years or older in the municipality of Cambé, northern Paraná state. Then, they were followed up in 2015.

The sample size was calculated using data from the 2007 population count performed by the Brazilian Institute of Geography and Statistics, when 92,888 people lived in the municipality – 33.10% were 40 years or older. The StatCalc® tool of the Epi Info® 3.5.3 software was used, with a margin of error of 3,00%, outcome prevalence of 50,00%, and confidence level of 95,00%, resulting in a sample size of 1066 participants. Assuming eventual losses and refusals, the sample size was increased by 25,00%, totaling 1332 people to be interviewed. Baseline data were collected between February and June 2011. To ensure greater representativeness, the spatial distribution of participants to be interviewed was arranged based on the 86 census tracts of the urban region of the municipality. After the total number of participants to be interviewed was calculated, the number of individuals to be interviewed in each census tract was determined, considering the number of residents in each tract, according to sex and age group. Based on the maps of the census divisions, the sampling route was systematically defined to ensure full coverage of the tract and random selection of households. The routes traveled by the researchers, the households visited (1 of every 2), and 1 individual per household were randomly selected to be included in the study.14

A loss was deemed to have happened when the individual selected in the household was not found after at least 3 visits on alternate days and/or times or the person refused to be interviewed. There were 159 losses, 93 due to refusals and 66 due to not finding the individual after 3 attempts. Thus, the sample in 2011 was composed of 1180 individuals (88.10%). For the 2015 follow-up of the VigiCardio study (age ≥ 44 years), the same individuals were invited to participate, by telephone or home visit. The same criteria used in 2011 were adopted to account for losses. After the losses were added to deaths, a total of 885 individuals were interviewed in the follow-up.14

The final sample of the present study was composed of the individuals in the 2015 database and excluded individuals who had not taken medications in the last 15 days. For this purpose, the response to the following question was evaluated: “In the last 15 days, did you take any medications?” Those who answered “Yes” and presented the medical prescription, primary and/or secondary packaging, and/or drug package insert were included. The final definition of the sample is shown in detail in Figure 1.

FIGURE 1.
Flowchart for determining inclusion in the study sample. Cambé (Brazil), 2015.

Interviewers were trained before baseline data collection, and a pilot study was conducted to investigate possible weaknesses in the data collection instrument, selection strategy, and approach. In both years, the data were collected during a home interview using a semistructured questionnaire, and the number of questions was increased in the follow-up.

The independent variable (ACL) was measured by the Anticholinergic Drug Scale (ADS).15 The ADS has a list of 117 drugs with estimated ACLs ranging from 1 to 3. The total ACL score is the number of drugs used by the individual, assuming that a load of different drugs respects a linear additive model.15 The ADS scale was used in the present study because there are similarities between the drugs used in Brazil and the United States;16 the ADS scale is more suitable for American pharmacotherapy10 and its validation was done according to the Anticholinergic Activity Scale (AAS).

The ADS scale was applied to the drug therapy reported by the patient in the last 15 days, and the total load was obtained by summing the values of all drugs used in this period. Therapies with scores of 3 or higher were considered significant,15 while those with scores lower than 3 were considered non-significant.

The dependent variable was self-perceived health, determined with the following question: “How do you rate your health status?” The results were grouped into positive self-perception (PSP) (responses: very good and good) and negative self-perception (NSP) (responses: fair, poor, and very poor).17,18 Self-perception of health is an important indicator of morbidity and mortality12 and a relevant factor for the use of health services,12 and this simple question has been one of the most frequently employed health indicators in health research.19 Sociodemographic and health characteristics were included as control variables. The sociodemographic variables were sex (female; male), age (continuous), marital status (with a partner; without a partner), educational level (0 – 4; 5 – 8; ≥ 8 years of study), and economic status (based on the classification of the Brazilian Association of Research Companies,20 which considers the purchasing power and educational level of the head of the family, categorized as A/B and C/D/E). The health variables were polypharmacy (categorized as yes or no), defined as the use of five or more active ingredients, according to the vast majority of studies on the subject,21 depression, hypertension, and diabetes mellitus (self-report of medical diagnosis).

The data collected with paper forms were entered twice in Microsoft Office Excel® spreadsheets and then compared using the Microsoft Office Spreadsheet Compare®. In 2015, approximately one-third of the data were collected with electronic forms, which were first stored on the server and then exported to Microsoft Office Excel® spreadsheets, thus eliminating data transcription.

The associations between the variables were analyzed by Poisson logistic regression with robust variance to obtain prevalence ratios (PRs). First, univariate (crude) analyses were performed between each control variable and ACL with the outcome of self-perceived health. Next, models were constructed progressively and cumulatively including the control variables. The models consisted of the following:

• Model 1 – sex and age (continuous);

• Model 2 – model 1 plus marital status, education, and economic status;

• Model 3 – model 2 plus polypharmacy; and

• Model 4 – model 3 plus depression, hypertension, and diabetes mellitus.

Significant associations were those with p < 0.05, considering the 95% confidence interval (95%CI). To analyze the effect of each ACL level (continuous) on self-perceived health, a regression model with generalized estimating equations was adopted, with the following parameters: normal distribution of the outcome, log link, permutation matrix, and robust variance. The analyses were performed within the Statistical Package for the Social Sciences (SPSS), version 20.0.

The participants were informed about the content of the study and signed an informed consent form. In addition, all research procedures were approved by the Human Research Ethics Committee of State University of Londrina (Universidade Estadual de Londrina) with certificate no. 0192.0.268.000-10 (2011) and certificate no. 39595614.4.0000.5231 (2015).

The prevalence of NSP among the 662 individuals evaluated in the present study was 50.91%. The population was predominantly female (61.78%), aged between 50 and 59 years (35.20%), of economic status C/D/E (66.16%), had a low educational level (53.32%), and lived with a partner (68.08%). Also, about one-quarter of the participants underwent polypharmacy treatment and reported having a diagnosis of depression. Approximately 60,00% and 20,00% of the participants reported a medical diagnosis of hypertension and diabetes mellitus, respectively (Table 1).

TABLE 1.
Frequency of sociodemographic, economic, and health variables, according to self-perceived health, in individuals aged 44 years or older.

*Economic status: considers the purchasing power and educational level of the head of the family,20 categorized as A/B/C and D/E; ^tUse of five or more active ingredients.21

Of the 662 respondents who used some type of medication, 137 had significant ACL (20.69%). In those individuals, NSP was more prevalent (61.31%) than in those whose ACL was not significant (48.19%). Participants of female sex, older age, lower economic status, lower education, or no partner had a higher frequency of NSP. Those who underwent polypharmacy treatment and reported a diagnosis of depression, hypertension, and diabetes mellitus also had a higher prevalence of NSP (Table 1).

Fluoxetine, orphenadrine, and amitriptyline were the most common medications with an ACL among middle-aged adults (44 to 59 years). The top three medications for older adults (60 years or older) were furosemide, clonazepam, and amitriptyline.

As shown in Table 2, when the crude analysis was performed, significant ACL, age (in years), lower economic status, lower education, polypharmacy, and reported diagnosis of depression and hypertension were significantly associated with a higher prevalence of NSP.

TABLE 2.
Prevalence ratio (PR) of socioeconomic and health variables according to self-perceived health in adults aged 44 years or older (n = 662).

*Poisson logistic regression with robust variance; ^†Economic status: considers the purchasing power and educational level of the head of the family,20 categorized as A/B and C/D/E; ^‡Use of five or more active ingredients.21

In the adjusted analysis, a significant association between ACL and NSP was found in all models (Table 3). Linear regression adjusted for sex, age (continuous), economic status, educational level, marital status, polypharmacy, and reported diagnosis of depression, hypertension, and diabetes mellitus showed that, with each additional ACL level, the chance of worse self-perceived health was 6.10% higher (95%CI 2.70 – 9.50; p < 0.001).

TABLE 3.
Prevalence ratio for the association between anticholinergic load and negative self-perceived health, according to different statistical models, in adults aged 44 years or older (n = 662).

* Prevalence ratio obtained by Poisson regression with robust variance. ACL: anticholinergic load. Model 1: adjusted for the variables sex and age (continuous); Model 2: adjusted for model 1 plus economic status, marital status, and education; Model 3: adjusted for model 2 plus polypharmacy; Model 4: adjusted for model 3 plus reported diagnosis of depression, hypertension and diabetes mellitus.

The main objective of this study was to investigate the association between ACL and self-perceived health. A significant association was found between these variables, regardless of confounding factors. Considering the multidimensionality of self-perceived health, in addition to the conditions investigated in this analysis, it is plausible to assume that many factors influence self-perceived health status. For this reason, an analytic model for confounding factors including polypharmacy and depression was adopted, as these variables are associated with NSP.22

In the present study, the prevalence of NSP was 50.90%. This finding is similar to those of other Brazilian studies in the older population.19,23 A similar prevalence was also found in Serbia,24 but a significantly lower rate was reported in Canada (NSP < 20,00%) in a study with a similar age range and design.25 Because self-perceived health is a subjective, multifactorial concept that includes individual, collective, and socioenvironmental aspects,26 the discrepancies may be due to socioeconomic and cultural differences between countries. Therefore, comparisons of this dimension between populations with such divergent characteristics should be made with caution.

Regarding the use of drugs, those exposed to significant ACL had a higher prevalence of NSP. Self-perceived health is a widely used dimension and is described as an indicator of morbidity and mortality. Thus, variables that are associated with self-perceived health may also be indirectly associated with increased morbidity and mortality.11 In fact, a meta-analysis showed that a higher ACL is associated with an increased risk of mortality.27

Self-perceived health is recognized as an indicator of both the use of health services and the medical assessment of health conditions.12,13 Although our hypothesis assumes that NSP is a consequence of high ACL, the bidirectionality of these factors cannot be ignored. Thus, indirectly, the presence of significant ACL might lead to a worse self-perception of health, and this perception, in turn, might lead the patients to demand more health services and even influence their assessment of their medical condition.14 Moreover, a cyclical phenomenon may be observed, as the demand for health services has been a risk factor for the use of anticholinergic drugs.28

According to López-Alvarez et al.,9 some recommendations are essential to guide health professionals in safer prescription of anticholinergic drugs. Those authors say that if prescribing an anticholinergic medication is necessary, the drug with lower anticholinergic potential and greater affinity for its site of action should be prioritized.9 Whenever a drug is added and, in some way, affects functions sensitive to anticholinergic actions, a global assessment of the pharmacotherapy should be performed, along with the avoidable or unavoidable ACL, and nonpharmacological approaches should be considered.9 Finally, given the potential risk of irreversible cognitive decline from the prolonged use of high-anticholinergic-potential drugs, treatment should not exceed 3 months without a reassessment.9

Among other aspects to be considered, the perception that patients have about their care should be a point of attention during clinical evaluation.29 Shared care and person-centered therapy are some of the topics that encompass the perceptions and expectations of patients and professionals. Therefore, such topics should be a common subject in the training of health professionals to ensure that, once provided with information, they can use the available tools and resources, including medications, in a rational manner.

Our findings are another indicator of morbidity and mortality related to the use of anticholinergic drugs. In addition to the results found in this population, we sought to reflect on the complexity of drug use and its impact on self-perceived health.

The limitations of this study are those inherent to a cross-sectional design — although associations between the variables can be observed, no temporal relationship can be defined. In this sense, although the hypothesis that the side effects caused by anticholinergics lead to a negative self-perception of health, which in turn leads to the use of other drugs, is supported, this relationship may be bidirectional, and the opposite directionality cannot be excluded. Additionally, the study was conducted in only 1 city in southern Brazil, which limits the extrapolation of the results to the entire country.

Conversely, the systematic process adopted from sampling to data analysis contributes to the quality of the information presented. The study of adults enabled us to investigate the action of anticholinergic drugs beyond the older population, a demographic known to be polymedicated and more susceptible to side effects, and may have provided evidence of anticholinergic action before older age. In addition, the results were controlled for possible confounding factors, including age, which gives us greater confidence in the associations found. Additionally, it should be highlighted that the literature supports the use of polypharmacy with a multimorbidity proxy because there is a systematic correlation with the existence of chronic conditions (pattern of multimorbidity and polypharmacy).

Individuals taking drugs with anticholinergic activity have a higher prevalence of negative self-perceived health, with a load-dependent effect. That is, the higher the ACL, the greater the chances of worse self-perceived health. Longitudinal studies are suggested to investigate the causality of the links between the associated variables in order to elucidate the possible bidirectionality of this relationship.