Abstract
Background:
A growing body of evidence points to potential risks associated with polypharmacy (using ≥5 medications) in older adults, but most evidence is derived from studies where racial and ethnic minorities remain underrepresented among research participants.
Objective:
Investigate the association between polypharmacy and cognitive function, subjective health state, frailty, and falls in Hispanic older adults.
Methods:
Panama Aging Research Initiative–Health Disparities (PARI-HD) is a community-based cohort study of older adults free of dementia at baseline. Cognitive function was measured with a neuropsychological test battery. Frailty assessment was based on the Fried criteria. Subjective health state and falls were self-reported. Linear and multinomial logistic regression analyses were used to examine association.
Results:
Baseline evaluations of 468 individuals with a mean age of 69.9 years (SD = 6.8) were included. The median number of medications was 2 (IQR: 1–4); the rate of polypharmacy was 19.7% (95% confidence interval [CI] = 16.1–23.3). Polypharmacy was inversely associated with self-rated overall health (b =−5.89, p < 0.01). Polypharmacy users had 2.3 times higher odds of reporting two or more falls in the previous 12 months (odds ratio [OR] = 2.31, 95% CI = 1.06–5.04). Polypharmacy was independently associated with Fried’s criteria for pre-frailty (OR = 2.90, 95% CI = 1.36–5.96) and frailty (OR = 5.14, 95% CI = 1.83–14.42). Polypharmacy was not associated with cognitive impairment.
Conclusions:
These findings illustrate the potential risks associated with polypharmacy among older adults in Panama and may inform interventions to improve health outcomes in this population.
Keywords
INTRODUCTION
In recent decades, the aging population has been growing rapidly, presenting new challenges in healthcare of older adults [1, 2]. Polypharmacy, the concurrent use of multiple medications [3], is a common practice in geriatric care, driven by the need to address multiple chronic conditions simultaneously [4]. While medications are essential for managing comorbidity, the potential adverse effects resulting from polypharmacy have raised concerns regarding its impact on cognitive and functional outcomes, including falls and frailty [5]. The World Health Organization (WHO) defines polypharmacy as the routine use of five or more medications, including over-the-counter, prescription and/or traditional and complementary medicines [6]. Older adults are often polypharmacy users, which can lead to drug-drug interactions with unexpected results and potential harm.
Cognitive decline is a significant concern among aging individuals, as it can negatively impact quality of life and functional independence. Mounting evidence suggests that polypharmacy may contribute to cognitive impairment or accelerate cognitive decline in middle-aged and older adults [7–10]. Another adverse event related to polypharmacy in older adults is falls, a major health risk often resulting in severe injuries, loss of independence, and increased healthcare utilization [11, 12]. Polypharmacy has been identified as a modifiable risk factor for falls [13], as certain high-risk medications may produce side effects such as sedation, confusion, impaired balance, and falls [14–16]. Polypharmacy has also been linked to frailty [17, 18], as the cumulative burden of medications and potential adverse effects may exacerbate the physiological decline associated with frailty. Frailty, characterized by increased vulnerability to adverse health outcomes [19, 20], is prevalent among older adults and has far-reaching implications for well-being [21, 22]. The complex interactions between medications, including the potential for central nervous system effects, underscore the need to explore the relationship between polypharmacy and adverse health outcomes.
Much work has focused on the association between polypharmacy and cognitive and functional outcomes [7, 17]; however, most studies have been conducted in developed countries. Specifically, little is known about the link between polypharmacy and health outcomes in older adults in Latin America [23]. While a growing Hispanic population in the U.S. draws increasing attention in aging research, studies focused on Hispanic older adults outside the U.S. are limited, and there is a dearth of data from Central American countries. The current study aims to close this gap and describe the association of polypharmacy with cognition, falls, and frailty in a cohort of community-dwelling older adults in Panama.
METHODS
Participants
The Panama Aging Research Initiative–Health Disparities (PARI-HD) study is a community-based longitudinal study of the factors affecting older adults in Panama. All Panamanians aged 60 years and older, literate, dementia-free and community-dwelling, were eligible for inclusion in the study and were enrolled after providing informed consent. The study protocol was approved by the Institutional Bioethics Committee of the Caja del Seguro Social (P-083-16). Here we report cross-sectional data from 468 participants collected between October 2016 and March 2020. This study followed the STROBE guidelines for reporting cross-sectional observational studies (Supplementary Table 1).
Clinical interview and measures
Evaluations were conducted in Spanish by clinical researchers and psychology students. The clinical interview comprised information regarding chronic conditions, incidence of one or more falls in the preceding 12 months, current or past tobacco use, self-reported sleep problems, and subjective cognitive impairment. The following chronic conditions were assessed: hypertension, ischemic heart disease, diabetes, stroke, cancer, chronic lung disease, and arthritis. Body mass index (BMI; kg/m2) was measured. For the examination of polypharmacy, participants were asked to bring their medications to each in-person visit and present the drugs or drug packages to the interviewer. Participants also reported the medical conditions for which they were taking each medication. The types of drugs and their purpose were later verified by a team of medical residents, who computed the number of medications taken by each participant and prevalence of polypharmacy in the cohort. Frequencies of high-risk drugs were also verified; these consisted of medications associated with impaired psychomotor function and falls [15, 24], including tricyclic anti-depressants, benzodiazepines, antipsychotics, first generation anti-histamines, sedative-hypnotics, and diuretics. One measure from the European Quality of Life Health Questionnaire (EQ-5D-3L) [25] was used to evaluate subjective health status. The measure includes a visual analogue scale where participants rate their health (on a scale from 0 (worse imaginable health) to 100 (best imaginable health). Independence in activities of daily living was measured with Lawton and Brody [26] and Katz Index [27]. Lawton and Brody questionnaire includes eight instrumental activities of daily living (IADL): telephoning, shopping, food preparation, housekeeping, laundering, use of transportation, and managing finances and medications. The Katz Index includes six basic activities of daily living (BADL): bathing, dressing, toileting, transferring, continence and feeding. IADL scores were adjusted in cases where an individual had never performed a task. For both IADL and BADL higher scores indicate higher function and independence. A questionnaire based on the 15-item Subjective Memory Complaints Questionnaire [28] was used to assess subjective memory complaints.
Cognitive and frailty measures
Cognition was assessed using a neuropsychological test battery from which six cognitive domains were derived (attention, executive function, verbal learning, memory, language, and visuospatial abilities). In addition, global cognition was measured by the 30-item Spanish version of the Mini-Mental State Examination (MMSE) [29], and scores were adjusted for age and education [30]. Clinical and neuropsychological data were reviewed by a consensus committee that assigned participants to groups based on core clinical criteria for mild cognitive impairment (MCI) [31] or no cognitive impairment (normal controls; NC). The Global Deterioration Scale (GDS) [32] was used to rate the clinical stages of cognitive function. Participants without cognitive impairment performed within normal limits in the neuropsychological assessment, had a MMSE score > 24, and GDS level 1 or 2. MCI diagnosis required deficits in at least one cognitive domain, independence in activities of daily living and a rating of GDS≥3 [31, 33].
Physical frailty was assessed using the five components proposed by Fried [19]: unintentional weight loss, self-reported exhaustion, weakness (grip strength), slow walking speed, and low physical activity. Weight loss was defined by self-reported unintentional weight loss greater than 4.5 kg in the 12 months before the evaluation. Exhaustion was defined through self-reported responses to the following questions: “I felt everything I did was an effort,” and “I could not get going.” This criterion was scored positive if at least one condition was present for 3 days or more during the previous week. Weakness was defined as physical performance on the grip strength test (Lafayette Hand Dynamometer, Model 78010) and according to sex and BMI. For women, BMI was categorized as ≤23.9, 24.0–26.6, 26.7–30.4, and >30.4; corresponding grip strength cutoffs were ≤20.5, ≤24.4, ≤27.6, and ≤31.7. For men, BMI was categorized as ≤24.3, 24.4–27.2, 27.3–29.4, and >29.4; corresponding grip strength cutoffs were ≤21.1, ≤24.9, ≤27.6, and ≤29.9. Participants fulfilling the criteria and those unable to take the handgrip test because of physical limitations were scored positive. Walking speed (m/s) was calculated starting in a static position and walking 5 m at a usual pace, adjusted for sex and height. The cut-off points for women were as follows: height ≤1.58 m and time ≥6.31 s (equivalent to 0.79 m/s); height >1.58 m and time ≥5.91 s (0.85 m/s); and for men: height ≤1.69 m and time ≥5.93 s (0.84 m/s); and height >1.69 m and time ≥5.91 s (0.85 m/s). Physical activity was measured through self-reported responses to the question: “Which of the following best describes your level of physical activity” and given the following options: (a) vigorous activity for at least 30 min 3 times a week; (b) moderate activity at least 3 times a week; and (c) rarely active, prefers sedentary activities. This criterion was scored positive if participants selected the last option. The frailty score comprises a possible range from 0 to 5. A person was considered frail if three or more criteria are assessed as positive, pre-frail if one or two are assessed as positive, and not frail if none are considered positive.
Statistical analyses
Descriptive analyses examined sample characteristics in relation to polypharmacy (i.e., ≥5 medications). Means and standard deviations, and frequencies and percentages, were calculated, and univariate and Chi-Square analyses compared individuals with and without polypharmacy for each variable, respectively. Linear regressions examined associations between polypharmacy and mild cognitive impairment, performance across cognitive domains, and subjective health state. Multinomial logistic regressions investigated whether polypharmacy predicted falls and frailty. Analyses also examined whether polypharmacy alone or in combination with high-risk drugs predicted cognitive and functional outcomes. In all models, covariates included age, sex, years of education and number of chronic illnesses. All analyses were conducted using SPSS version 28, and cases with incomplete data were excluded from statistical analyses via listwise deletion.
RESULTS
Descriptive statistics
Demographic, clinical, and cognitive characteristics
Participants were on average 69.9 years of age (SD = 6.8) and 76.3% female, with an average of 15.9 years of formal education (SD = 4.6). Most participants completed a university degree (61.5%) and received a monthly income in the upper quartile (62.4%). Participants reported good subjective health state (M = 84.8, SD = 14.2) with moderate rates of past or current tobacco use (30.8%) and difficulties sleeping (37.2%). Many participants were overweight (42.4%) or obese (24.9%) as indicated by BMI measurements (M = 27.5 kg/m2, SD = 5.2). Most participants reported at least one chronic illness (83.8%), with hypertension being the most frequent (54.1%). In addition, 11.6% met the GDS-15 cut-off for mild or established geriatric depression (score≥6). Neuropsychological test performance indicated that a minority of participants (9.2%) were cognitively impaired, although a plurality of participants (42.1%) reported experiencing subjective memory impairment. For individual cognitive domains, univariate analyses revealed that participants with polypharmacy status scored lower on learning and long-term memory assessments compared to participants without polypharmacy status, but groups did not differ in other domains (see Table 1).
Demographic, clinical, and cognitive characteristics of study participants
Demographic, clinical, and cognitive characteristics of study participants
M, mean; SD, standard deviation; USD, U.S. Dollars; BMI, body mass index.
Polypharmacy characteristics
On average, participants were taking 2.6 medications (SD = 2.3; 95% CI = 2.40–2.81) with a median of 2 medications (IQR: 1–4). Out of 468 participants, 19.7% (95% CI = 16.1–23.3) met the criteria for polypharmacy (≥5 medications), and this subset of participants reported taking 6.2 medications on average (SD = 1.6; 95% CI = 5.88, 6.54) with a median of 6 medications (IQR: 5–6.75). Approximately one-third of participants reported at least one high-risk medication (31.1%), and the most common high-risk medications were diuretics (19.1%), followed by benzodiazepines (5.8%), anti-depressants (5.6%), and anticonvulsants (5.6%). Polypharmacy characteristics and high-risk medications are summarized in Table 2.
Polypharmacy characteristics of study participants
M, mean; SD, standard deviation; bFrequencies too low to compute Chi square; IQR, interquartile range.
Functional and frailty characteristics
Overall, participants reported a high degree of functional independence, although polypharmacy users were significantly less independent in both BADL and IADL. Most participants reported no falls (73.9%), although polypharmacy users reported twice the rate of recurrent falls as non-polypharmacy users. Aggregate assessments of weight loss, grip strength, walking speed, physical fatigue, and physical activity (according to Fried criteria) revealed significant differences between polypharmacy and non-polypharmacy groups across frailty categories (Table 3).
Activities of daily living, falls, and frailty characteristics of study participants
M, mean; SD, standard deviation; aFrailty was assessed using the five components proposed by Fried [19].
Multivariable models
Polypharmacy and global cognition
A hierarchical multiple regression analysis was performed to examine whether total number of medications was related to global cognition (Table 4). Predictor variables were entered stepwise: covariates were entered in Step 1 [F(4, 458) = 0.28, MSE = 1.90, R2 = 0.002, p = 0.891], a binary variable assessing whether participants were taking at least 1 high-risk medication was entered in Step 2 [FΔ (1, 457) = 1.82, MSE = 1.90, R2Δ = 0.004, R2 = 0.006, p = 0.709], and total number of medications was added in Step 3 [FΔ (1, 456) = 0.48, MSE = 1.90, R2Δ = 0.001, R2 = 0.007, p = 0.81]. The omnibus tests were not significant at any step of the model, and none of the predictor variables were significantly associated with global cognition.
Hierarchical linear regression analysis of total number of medications and global cognition
Outcome variable was total score of the Mini-Mental State Examination (MMSE). At least 1 high-risk medication was coded as 0 = no, 1 = yes. Sex was coded as 0 = male, 1 = female. p < 0.05*, p < 0.01**, p < 0.001***.
Polypharmacy and subjective health state
Two hierarchical multiple regression analyses were performed to examine whether polypharmacy is related to subjective health state. For the first hierarchical multiple regression (Table 5A), predictor variables were entered stepwise: covariates were included in Step 1, the number of high-risk medications (i.e., sum) was entered in Step 2, and polypharmacy status was added in Step 3. Step 1 explained a significant portion of the variance [F (4, 458) = 6.61, MSE = 13.75, R2 = 0.055, p < 0.001] in subjective health state, and indicated significant effects for education level (b = 0.13, p = 0.006) and number of chronic illnesses (b = −0.17, p < 0.001). Step 2 explained additional variance [FΔ (1, 457) = 6.75, MSE = 13.66, R2Δ = 0.013, R2 = 0.068, p = 0.01] and revealed a negative association between the total number of high-risk medications (b = −0.12, p = 0.010) and self-rated health. Step 3 explained additional variance [FΔ (1, 456) = 9.85, MSE = 13.53, R2Δ = 0.02, R2 = 0.088, p = 0.002], indicating that polypharmacy status was inversely associated with self-rated health (b = −0.17, p = 0.002) after adjusting for covariates and the number of high-risk medications.
Hierarchical linear regression analysis of polypharmacy status and subjective health state
Outcome variable was subjective health state (EQ-5D-3 L) [25]. Sex was coded as 0 = male, 1 = female. Polypharmacy status was coded as 0 = no, 1 = yes. p < 0.05*, p < 0.01**, p < 0.001***.
The second hierarchical multiple regression was similarly comprised of three steps (Table 5B): Step 1 included covariates [F (4, 458) = 6.61, MSE = 13.75, R2 = 0.055, p < 0.001], Step 2 included a binary variable assessing whether participants were taking at least 1 high-risk medication [FΔ (1, 457) = 4.21, MSE = 13.70, R2Δ = 0.008, R2 = 0.063, p = 0.041], and the total number of medications was added in Step 3 [FΔ (1, 456) = 6.29, MSE = 13.62, R2Δ = 0.013, R2 = 0.076, p = 0.013]. The results of this analysis replicated the first, and further demonstrated the link between medication use and subjective health state; high-risk medication use (b = −0.10, p = 0.041) is uniquely and negatively associated with self-rated health when adjusting for covariates, but only the sum total of medications is associated with poorer self-rated health (b = −0.14, p = 0.013) when holding all other variables constant.
Hierarchical linear regression analysis of total number of medications and subjective health state
Outcome variable was subjective health state (EQ-5D-3L) [25]. Sex was coded as 0 = male, 1 = female. At least 1 high-risk medication was coded as 0 = no, 1 = yes. p < 0.05*, p < 0.01**, p < 0.001***.
Polypharmacy and frailty
Two multinomial logistic regressions assessed associations between polypharmacy and frailty category (i.e., pre-frailty and frailty). The first multinomial regression investigated whether polypharmacy characteristics (i.e., number of high-risk medications and polypharmacy status) predict frailty while adjusting for covariates. The omnibus test was significant (χ2 = 77.02, p < 0.001, Nagelkerke R2 = 0.18) and showed acceptable classification accuracy (60%). Pre-frailty was significantly associated with sex, age, and polypharmacy status, but not the total number of high-risk medications. Participants were 1.1 times more likely to be pre-frail for every unit increase in age, 2.7 times more likely to be pre-frail if female, and 2.9 times more likely to be pre-frail if they were taking 5 or more medications. Similarly, frailty was associated with age, sex, and polypharmacy status, but not the total number of high-risk medications. Participants were 1.1 times more likely to be frail for every unit increase in age, 3.1 times more likely to be frail if female, and 5.1 times more likely to be frail if they were taking ≥5 medications. These results are summarized in Table 6A.
Multinomial logistic regression analysis of polypharmacy status and frailty
Outcome variable was frailty category with ‘not frail’ as the reference category. Sex was coded as 0 = male, 1 = female. Polypharmacy status was coded as 0 = no polypharmacy, 1 = polypharmacy. p < 0.05*, p < 0.01**, p < 0.001***.
The second multinomial regression model was performed using the same covariates as the previous analysis, but instead included a binary variable assessing whether taking at least 1 high-risk medication (0 = no, 1 = yes) and the total number of medications predict frailty (see Table 6B). The omnibus test of the model showed significant fit (χ2 = 70.17, p < 0.001, Nagelkerke R2 = 0.17) and similar classification accuracy (60%). In line with the previous analysis, both pre-frailty and frailty categories were significantly associated with sex, age, and the total number of medications, but not with whether participants were taking at least 1 high-risk medication. Females were 2.6 times more likely to be pre-frail and 3 times more likely to be frail than males. Participants were 1.1 times more likely to be pre-frail or frail for every unit increase in age. Results further indicated that participants were 1.2 times more likely to be pre-frail (b = 0.15, p = 0.022) and 1.3 times more likely to be frail (b = 0.26, p = 0.01) for every unit increase in the total number of medications, but taking at least one high-risk medication was not associated with any frailty category.
Multinomial logistic regression analysis of total number of medications and frailty
Outcome variable was frailty category with ‘not frail’ as the reference category. Sex was coded as 0 = male, 1 = female. At least 1 high-risk medication was coded as 0 = no, 1 = yes. p < 0.05*, p < 0.01**, p < 0.001***.
Two multinomial logistic regressions investigated whether polypharmacy was associated with falls (i.e., one fall, two or more falls) during the preceding year. The first multinomial regression assessed demographic factors (e.g., age, sex, education), physical comorbidities (i.e., number of chronic illnesses), and polypharmacy characteristics (i.e., number of high-risk medications and polypharmacy status) to predict falling. The omnibus was significant (χ2 = 23.78, p = 0.022, Nagelkerke R2 = 0.06) and showed good classification accuracy (74.1%). Reporting one fall was significantly associated with female sex only. Specifically, female participants were 2.6 times more likely to have fallen once in the past year. In contrast, recurrent (2 or more) falls were associated with polypharmacy status alone, such that participants were 2.3 times more likely to have fallen 2 or more times if they were taking ≥5 medications (see Table 7A).
Multinomial logistic regression analyses of polypharmacy status and falls
Outcome variable was number of falls with ‘no falls’ as the reference category. Sex was coded as 0 = male, 1 = female. Polypharmacy status was coded as 0 = no polypharmacy, 1 = polypharmacy. p < 0.05*, p < 0.01**, p < 0.001***.
Following the first analysis, the second multinomial regression used the same demographic and health variables, but included a binary variable assessing whether participants were taking at least 1 high-risk medication (0 = no, 1 = yes) and the total number of medications to predict falling. The classification accuracy of the model improved slightly (74.3%) and yielded a significant omnibus test (χ2 = 21.05, p = 0.05, Nagelkerke R2 = 0.06), however, results showed only a significant association between participant sex and reporting one fall. Specifically, female participants were 2.8 times more likely to report falling once in the past year, whereas taking at least one high-risk medication and number of medications were not associated with falling. Table 7B summarizes these results.
Multinomial logistic regression analyses of total number of medications and falls
Outcome variable was number of falls with ‘no falls’ as the reference category. Sex was coded as 0 = male, 1 = female. At least 1 high-risk medication was coded as 0 = no, 1 = yes. p < 0.05*, p < 0.01**, p < 0.001***.
DISCUSSION
In this cohort of community-dwelling older adults, polypharmacy was associated with poorer self-rated health, and greater likelihood of recurrent falls and frailty. Older adults with polypharmacy were 2.3 times more likely to have reported falling two or more times in the previous 12 months than those without polypharmacy, and the likelihood of recurrent falls increased with each additional medication. We also found evidence of associations between polypharmacy and frailty, indicating that polypharmacy status as well as any increase in the number of medications increased the likelihood of meeting criteria for pre-frailty and frailty. Although studies have shown that polypharmacy in older adults is associated with poorer cognitive outcomes [7, 34], we did not find associations between polypharmacy and cognitive impairment or individual cognitive domains. There are a few possible reasons for these inconclusive results. One possibility is that prolonged rather than incident polypharmacy may affect cognitive capability. There is increasing evidence that cumulative exposure to polypharmacy is associated with risk for cognitive impairment [7, 10] and longitudinal data will allow us to examine this further. Second, associations between polypharmacy and cognitive impairment may not yet be evident in our relatively healthy cohort, where only 9.2% was cognitively impaired at baseline. PARI-HD enrolls relatively young and cognitively healthy participants, which may have limited the range in possible cognitive scores and the ability to identify an association with polypharmacy. In most studies, polypharmacy is associated with impairment in older, more vulnerable, and less cognitively healthy adults; in others, cognitive change is reported but not cognitive impairment based on diagnostic criteria [7–10]. At least one study showed associations between polypharmacy and cognitive performance in dementia-free older adults [35]. Third, Panama is an ethnically diverse country, and studies have shown that pharmacokinetics and pharmacodynamics differ between population groups [36]. Race and ethnicity can influence drug efficacy and may impact the interaction between polypharmacy and cognition. Ongoing studies aim to elucidate factors associated with polypharmacy and cognition.
We examined also whether associations between polypharmacy and negative functional outcomes were due to polypharmacy itself or to a higher probability of taking risk-increasing medications that have been shown to produce confusion, impaired balance and falls in older adults. Although most evidence links risk-increasing drugs to falling [37, 38], we examined whether specific drugs influenced subjective health, cognition and frailty as well. We found that taking at least one high-risk medication did not affect cognitive function, frailty or falls, nor did the number of high-risk medications influence these outcomes. At least one study [14] found that middle-aged and older adults who reported polypharmacy also had a greater likelihood of taking a medication associated with fall risk, and that polypharmacy was a risk for falling only if it included one of these individual medications. In our cohort, polypharmacy itself was the factor associated with falls and frailty.
Many studies have provided evidence that falls and frailty are intricately related [39, 40]. Both are important age-related health issues associated with common adverse events. Falls have been shown to be the result of many factors [41], one of which is polypharmacy, in addition to muscle weakness and gait alterations, which are part of the frailty criteria that we employed in this study. In our cohort, roughly one in five polypharmacy users reported recurrent falls compared to one in ten non-polypharmacy users. Frailty, on the other hand, is a dynamic process where different conditions interact to produce various stages of frailty [19, 43] as individuals age. Several measures have been proposed to operationalize and measure frailty, and we employed the Fried criteria, which has been widely used in studies with varied populations [44–46]. Briefly, these studies have shown that being pre-frail is associated with various poor health outcomes, such as falls, hospitalizations, disability, and mortality [19, 20], and frailty has been shown to be a strong predictor of all-cause mortality [47, 48]. In this context, falls can be considered a manifestation of the frail state. Thus, it is unlikely that polypharmacy by itself is a cause of falling or frailty, but rather, that polypharmacy and the risk for frailty and falls are not only correlated but associated with other geriatric conditions which over time could predict severe outcomes.
Polypharmacy was associated with poor self-reported health in our cohort, as has been previously reported [49, 50]. Taking high-risk medications was also associated with poor self-reported health, however, when polypharmacy or the total number of medications was considered, there was no longer an effect of high-risk drugs. In individuals with polypharmacy, poor self-rated health may be due to many factors such as a greater number of comorbid conditions or unmet health or psychological needs, possibilities that require further research. Much more is known about how polypharmacy is linked to multimorbidity than its relationship to self-rated health. In our cohort, polypharmacy users had poorer self-rated health, as well as significantly greater frequencies of all chronic illnesses surveyed and greater overall comorbidity than non-polypharmacy users. Moreover, although 57.9% of participants had at least two chronic conditions, the prevalence increased to 88% for those with polypharmacy. While reviews of polypharmacy in older adults have concluded that there are adverse cognitive and functional outcomes associated with taking multiple medications, the full range of outcomes potentially associated with polypharmacy is not well understood. In addition, estimates of polypharmacy vary widely across studies. In our cohort, the prevalence of polypharmacy was 19.7%, similar to estimates in adults aged 65 years and older in Brazil [51] (18.1%), but much lower than estimates in Chile [52] (37.6% in ages 65 years and older) and Colombia [53] (27.4% in adults 60 years and older). In a nationally representative sample of adults 50 years and older in Brazil, a lower prevalence of 13.5% was reported, and was found to vary according to sex, race, and geographic location [23]. In the global north, a meta-analysis of studies in North America, Europe and Asia [3], found a pooled prevalence of polypharmacy in adults aged 65 years and older of 45%. In sum, our observation of roughly one in five community-dwelling adults taking five or more medications at the same time is within the lower end of the range of estimates reported in other studies.
An important limitation of the present study is the cross-sectional nature of the data, which prevents us from drawing causal inferences about polypharmacy and changes in cognition, falls, and frailty over time. Also, because the study was observational, the issue of confounding is pertinent, because individuals who take more medications are also likely to have poorer health. However, after adjustment for comorbid conditions, polypharmacy (i.e. the number of drugs) remained a significant risk factor for falling and frailty. Moreover, drug interactions potentially can play a role in health outcomes, but the methodology of our study was not suitable to address that issue. Study strengths include the use of detailed clinical interviews and several objectively measured variables, including types and number of medications, as well as objective measures of cognitive function and frailty.
This work adds to the literature on the health of older adults in Latin America, an understudied population, and demonstrates how polypharmacy contributes to poor health outcomes in older adults in Panama. Ultimately, longitudinal follow-up and comprehensive medication assessments are needed to provide conclusive evidence that addressing excess medication can influence cognitive and physical health in this population. These studies could determine the points in time when reducing polypharmacy would have the greatest benefit for health outcomes. Nonetheless, the findings of this study are a first step toward identifying the potential risks associated with polypharmacy and may inform the development of targeted interventions to optimize medication management and improve health outcomes in this population.
AUTHOR CONTRIBUTIONS
Gabrielle Britton (Conceptualization; Investigation; Methodology; Project administration; Supervision; Writing – original draft; Writing – review & editing); Ivonne Torres-Atencio (Conceptualization; Data curation; Writing – original draft; Writing – review & editing); Maria B. Carreira (Writing – original draft; Writing – review & editing); Alondra Mendez (Data curation; Resources; Validation); Maryonelly Quintero (Data curation; Resources; Validation); Adriana Broce (Data curation; Resources; Validation); Diana Oviedo (Investigation; Writing – review & editing); Giselle Rangel (Investigation; Writing – review & editing); Alcibiades Villarreal (Investigation; Writing – review & editing); Adam Tratner (Data curation; Formal analysis; Methodology; Writing – review & editing); Sofia Rodriguez-Araña (Project administration; Supervision).
Footnotes
ACKNOWLEDGMENTS
We thank PARI-HD study participants for their contribution to research, and past and present PARI-HD staff who assisted in data collection.
FUNDING
This research was funded by Sistema Nacional de Investigación (SNI) de Panamá, grant numbers [SNI-063-2023; SNI-044-2023; SNI-64-2021; SNI-66-2021; SNI-074-2022; SNI-040-2023], Secretaría Nacional de Ciencia, Tecnología e Innovación (SENACYT), grant number [FID-22-092] and Instituto de Investigaciones Científicas y Servicios de Alta Tecnología (INDICASAT AIP), grant numbers [IGI-2021-002; IGI-2021-006].
CONFLICT OF INTEREST
G.B.B. is an Editorial Board Member of this journal but was not involved in the peer-review process of this article nor had access to any information regarding its peer-review.
The remaining authors have no conflict of interest to report.
DATA AVAILABILITY
The data supporting the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.
