Abstract
Background
Polypharmacy is associated with increased risk of adverse events and may increase healthcare expenditures. However, there is a lack of study using nationally representative data to assess additional healthcare costs associated with polypharmacy. This study aimed to quantify the associations between polypharmacy and healthcare expenditures and affordability of healthcare among older adults.
Methods
We used data from the Medical Expenditure Panel Survey (MEPS) from 2018 to 2021, which provides nationally presentative data in the US. Polypharmacy was defined as using ≥ 5 medications in the survey year. We included 23,300 adult participants aged 65 years or older (13,339 with and 9,961 without polypharmacy). Propensity score matching (PSM) was employed to control for sociodemographic and an extensive list of comorbidities. We used a two-part model to examine associations between polypharmacy and healthcare expenditures and logistic regression to examine associations between polypharmacy and affordability of care. Unaffordability of care was defined as not or delay receiving healthcare because patients could not afford it in the past 12 months.
Results
After PSM, 4,925 participants with polypharmacy and 4,925 without polypharmacy were included. The mean (95% confidence interval, CI) age of study population was 73.9 (95% CI: 73.8–74.2) years, with 54.5% (95% CI: 54.0–55.0) being female. All characteristics were well-balanced (standardized mean differences < 0.1 for all). Following PSM, the average annual total expenditure was $14,691.06 for older adults with polypharmacy and $8,912.09 for those without. Polypharmacy was associated with $4,303.82 (95% CI: 3,948.56-4,659.08) higher annual total healthcare expenditures, including $1,178.67 (95% CI: 1,165.10-1,192.38) more for prescription medicines, $1,359.83 (95% CI: 885.14-1,884.52) more for office-based visits, $814.91 (95% CI: 176.77-1,192. 06) more for inpatient stays, and $886.16 (95% CI: 322.48-1,449.83) more for outpatient visits. Furthermore, polypharmacy was associated with significantly higher risk of unaffordability of prescription medications (adjusted odds ratio [95% CI] 1.70 [1.30–2.23]).
Conclusions
Using propensity score methods, we provided estimates of additional healthcare costs associated with polypharmacy. Our research highlights the potential economic burden of polypharmacy on patients, family, insurers, and society.
Supplementary Information
The online version contains supplementary material available at 10.1186/s12877-025-06545-w.
Keywords: Polypharmacy, Older adults, Healthcare expenditures, Costs
Introduction
As the global population ages and the burden of non-communicable diseases rises[1], older adults frequently require multiple medications, increasing their risk of polypharmacy [2]. Polypharmacy is typically defined as the simultaneous use of five or more medications [3]. Previous studies have found that approximately 40% of those 65 or older experienced polypharmacy [4]. While medications generally improve patient health outcomes, for some patients, the risks associated with using multiple medications may outweigh the benefits [5]. Polypharmacy has been linked to potentially inappropriate medication use and adverse outcomes such as medication-related adverse effects, hospitalizations, reduced quality of life, and mortality [6–8].
Beyond its impact on health, unnecessary polypharmacy represents a substantial waste of medical resources [9, 10]. Polypharmacy increases the costs of medications and the potential for medication-related harms, imposing significant economic burdens on individuals, families, and healthcare systems [9–12]. The World Health Organization (WHO) estimates that costs attributable to inappropriate medication use and poor prescription management account for approximately 4% of the world’s total avoidable healthcare expenditure [9]. In the US, the Agency for Healthcare Research and Quality estimates that inappropriate medication use results in $420 billion in annual expenditure[13].
A limitation in previous studies examining the associations between polypharmacy and increased healthcare costs was their insufficient control for potential confounding factors and selection bias [10–12, 14]. In particular, differences in comorbidity burden between individuals with and without polypharmacy may contribute to substantial differences in healthcare costs. Propensity score matching (PSM) offers a robust approach to account for these confounders, enhancing the validity of estimates. Utilizing nationally representative survey data from the US and applying the PSM method, this study aimed to quantify the associations between polypharmacy and healthcare expenditures and affordability of healthcare among older adults from the perspective of the healthcare system.
Methods
Data source and study population
This study utilized data from the Medical Expenditure Panel Survey (MEPS) collected between January 2018 and December 2021 [15]. The MEPS is a nationally representative survey of the US civilian, non-institutionalized population [16]. Each year, nearly 14,000 families, comprising 35,000 members, are included as a part of a complex survey sampling strategy, with the characteristics of included individuals and their families weighted to reflect their representativeness to the U.S. adult population [17]. Each year, a new household panel is sampled and interviewed five times (i.e., five rounds) over two years, which provides continuous and concurrent estimates of health care expenditures [18]. Please see detailed timing and relationship between panels, rounds, and calendar years in Additional file 1.
The current study used data from the Household Component’s Full-year Consolidated Files and Medical Provider Component, which are parts of the MEPS data files. The Household Component questionnaire files contain information related to survey administration, demographics, income, individual health conditions, health status, health insurance, and person-level medical care usage [19]. After obtaining permission from the household survey respondents, medical providers are contacted by telephone to obtain information such as dates of visits, diagnosis, procedure codes, detailed information about prescription medications, charges, and payments. To increase sample size and obtain more precise estimates, we pooled data from 2018 to 2021 together.
The study population included in our analysis consisted of individuals aged 65 years or older at the time of the survey and with complete information on sex, race/ethnicity, and prescription medications (see Additional file 2 for patient flowchart).
The MEPS has been reviewed and approved by the Westat Institutional Review Board and informed consent was obtained from each participant. Data used in our analyses are publicly available and individual identifiers have been removed [20, 21]. In accordance with the data use agreement, statistical analysis of these data is allowed. Because the current study only used publicly available data, additional ethnics review was not required in according to the common rules. We followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline for cross-sectional studies throughout this report (see Additional file 3).
Prescription medication use
Each year in MEPS surveys, household respondents report the names of medications and the number of times the medications were obtained or purchased over the past year [22]. All medications, regardless of duration of use, were included in the survey data. If household respondents gave permission to release their pharmacy records, pharmacy providers supplied the national drug code, strength of medicine (amount and unit), quantity (package size/amount dispensed), days supplied, total expenditure and payments by source in the past year. For those with missing payment data, values were imputed using variables including the unique identifier of drug in the US, payers, person’s age and private and public health maintenance organization enrollment, pharmacy name, pharmacy chain name, quantify, state, census division, and region [23]. A validation study showed good concordance between medication expenditures from MEPS and Medicare Part D data [24]. We defined the number of prescriptions used as the total number of distinct medications a patient was prescribed during the survey year. Polypharmacy was defined as using ≥ 5 medications in the survey year.
Healthcare expenditure
After obtaining permission from household respondents, medical providers are contacted to obtain information such as the dates of visits and payments. Total healthcare expenditures referred to the combined amount of money spent on healthcare, including office visits, hospital stays, emergency room visits, prescription medicines, dental care, home healthcare, and other medical services. A validation study suggested an overall 85% concordance between healthcare expenditure in MEPS and Medicare claims data[25].We further examined out-of-pocket (OOP) payments, as these closely reflect the financial burden on patients or families; and payments by Medicare, since Medicare is the primary payer for healthcare expenditures among older adults.
Affordability of care
The survey asked participants whether they did not receive or delayed receiving healthcare because they could not afford it in the past 12 months. These questions were split into three sections: medical, dental, and prescription medicine.
Covariates
We incorporated sociodemographic characteristics, including age, sex, race/ethnicity, marital status, education, and family income. Participants’ age was grouped into 5 categories: 65–69, 70–74, 74–79, and ≥ 80 years. Race and ethnicity were categorized as non-Hispanic White, non-Hispanic Black, Hispanic, and others (Non-Hispanic Asian only, non-Hispanic other race, or multi-race). Marital status was categorized as either living with a partner or not living with a partner. Education was categorized as < high school, high school, or college or above. Family income was categorized according to the federal poverty level (FPL) as poor (< 125% of FPL), low income (125%−200% of FPL), middle income (200%−400% of FPL), and high income (≥ 400% of FPL). Health insurance was categorized as Medicare only, Medicare with private insurance, Medicare with other public insurance, or no Medicare (not enrolled in Medicare but covered by private insurance, Medicaid, TRICARE/CHAMPVA, Veterans Administration, or other public insurance). We also included participants’ prescription drug insurance coverage (private drug insurance, Medicare Part D, or no drug insurance).
To control for potential confounding by multimorbidity, we included an extensive list of comorbidities in the analyses. A total of 60 comorbidities were included based on a clinically driven comprehensive list of chronic conditions for measuring multimorbidity in older adults [26] (Table 1 and Additional file 4). The medical conditions were self-reported by participants and then recorded by interviewers using ICD-10-CM codes with 3 digit [27]. We also calculated the Charlson Comorbidity Index (CCI) to summarize the overall burden of comorbidities [28]. In addition, we assessed participants’ perceived health status and whether they had function limitation using instrumental activities of daily living (IADL) and activities of daily living (ADL). Perceived health and perceived mental health were each categorized as excellent, very good, good, fair, and poor. We examined the variance inflation factor (VIF) for all covariates in the propensity model and all covariates had VIF < 5, indicating a low likelihood of collinearity.
Table 1.
Baseline characteristics of older adults with and without polypharmacy before and after propensity score matching
| Before PSM Weighted % (95% CI) a |
After PSM Weighted % (95% CI) a |
|||||
|---|---|---|---|---|---|---|
| Characteristics | Without polypharmacy | With polypharmacy | SMD | Without polypharmacy | With polypharmacy | SMD |
| (N = 9,961) | (N = 13,339) | (N = 4,925) | (N = 4,925) | |||
| Age, years | 73.0 (72.8–73.2) | 74.6 (74.4–74.8) | 0.25 | 73.9 (73.7–74.2) | 74.0 (73.8–74.3) | 0.02 |
| Age,years *** | 0.25 | 0.03 | ||||
| 65–69 | 38.3 (37.0–40.0) | 27.8 (27.0–29.0) | 31.0 (29.0–33.0) | 31.5 (30.0–33.0) | ||
| 70–74 | 26.1 (25.0–27.0) | 26.5 (25.0–27.0) | 27.6 (26.0–29.0) | 26.4 (25.0–28.0) | ||
| 75–79 | 16.0 (15.0–17.0) | 19.5 (18.0–21.0) | 17.6 (16.0–19.0) | 18.3 (17.0–20.0) | ||
| 80+ | 19.6 (18.0–21.0) | 26.3 (25.0–28.0) | 23.8 (22.0–26.0) | 23.9 (22.0–26.0) | ||
| Sex *** | 0.07 | 0.01 | ||||
| Male | 47.6 (46.0–49.0) | 43.9 (43.0–45.0) | 47.8 (46.0–50.0) | 47.1 (45.0–49.0) | ||
| Female | 52.4 (51.0–54.0) | 56.1 (55.0–57.0) | 52.2 (50.0–54.0) | 52.9 (51.0–55.0) | ||
| Race ** | 0.09 | 0.02 | ||||
| Hispanic | 9.7 (8.4–11.0) | 8.2 (6.9–9.6) | 8.8 (7.4–10.0) | 8.8 (7.4–10.0) | ||
| Non-Hispanic White only | 73.4 (71.0–76.0) | 76.5 (74.0–78.0) | 74.3 (72.0–77.0) | 74.9 (72.0–77.0) | ||
| Non-Hispanic Black only | 9.3 (7.9–11.0) | 9.3 (8.1–11.0) | 10.0 (8.4–12.0) | 9.6 (8.1–11.0) | ||
| Others | 7.6 (6.2–9.2) | 6.1 (5.0–7.4.0.4) | 6.9 (5.5–8.7) | 6.7 (5.3–8.5) | ||
| Marital *** | 0.15 | 0.01 | ||||
| Living with partner | 59.3 (57.0–61.0) | 52.0 (50.0–54.0) | 56.5 (54.0–59.0) | 56.2 (54.0–58.0) | ||
| without a partner | 40.8 (39.0–43.0) | 48.0 (46.0–50.0) | 43.5 (41.0–46.0) | 43.9 (42.0–46.0) | ||
| Education *** | 0.12 | 0.01 | ||||
| < High school | 10.8 (9.7–12.0) | 14.2 (13.0–16.0) | 12.6 (11.0–14.0) | 12.5 (11.0–14.0) | ||
| High school | 27.6 (26.0–30.0) | 29.5 (28.0–31.0) | 28.3 (26.0–30.0) | 28.6 (27.0–31.0) | ||
| College or above | 61.6 (59.0–64.0) | 56.3 (54.0–58.0) | 59.1 (57.0–61.0) | 58.9 (57.0–61.0) | ||
| Insurance *** | 0.18 | 0.01 | ||||
| Medicare only | 41.2 (40.0–43.0) | 40.8 (39.0–42.0) | 41.3 (39.0–43.0) | 41.7 (39.0–44.0) | ||
| Medicare and private | 48.0 (46.0–50.0) | 46.3 (45.0–48.0) | 47.6 (45.0–50.0) | 47.4 (45.0–50.0) | ||
| Medicare and other public | 8.0 (7.2–9.0.2.0) | 11.9 (11.0–13.0) | 9.6 (8.4–11.0) | 9.5 (8.3–11.0) | ||
| No Medicare | 2.7 (2.3–3.3) | 1.0 (0.8–1.4) | 1.5 (1.1–2.0.1.0) | 1.4 (1.0–1.9.0.9) | ||
| Income *** | 0.16 | 0.03 | ||||
| <1.25 FPL | 13.0 (12.0–14.0) | 15.4 (14.0–17.0) | 14.1 (13.0–15.0) | 14.2 (13.0–16.0) | ||
| 1.25–2.0 FPL | 12.9 (12.0–14.0) | 16.5 (15.0–18.0) | 14.4 (13.0–16.0) | 15.0 (14.0–16.0) | ||
| 2.0–4.0 FPL | 27.3 (26.0–29.0) | 27.9 (27.0–29.0) | 27.8 (26.0–29.0) | 28.1 (26.0–30.0) | ||
| ≥ 4.0 FPL | 46.8 (45.0–49.0) | 40.1 (38.0–42.0) | 43.7 (41.0–46.0) | 42.7 (41.0–45.0) | ||
| Prescription insurance** | 0.14 | 0.09 | ||||
| Private | 23.4 (22.0–25.0) | 19.4 (18.0–21.0) | 22.9 (21.0–25.0) | 19.2 (18.0–21.0) | ||
| Medicare | 33.4 (32.0–35.0) | 30.1 (28.0–32.0) | 32.2 (30.0–34.0) | 31.4 (29.0–34.0) | ||
| No | 43.2 (41.0–45.0) | 50.5 (49.0–52.0) | 44.9 (43.0–47.0) | 49.5 (47.0–52.0) | ||
| IADL*** | 5.1 (4.5–5.7) | 12.0 (11.0–13.0) | 0.25 | 7.4 (6.4–8.6) | 7.6 (6.6–8.6) | 0.01 |
| ADL*** | 3.2 (2.8–3.8) | 7.5 (6.8–8.2) | 0.19 | 4.9 (4.1–5.8) | 5.0 (4.1–6.0.1.0) | 0.01 |
| Perceived health status*** | 0.58 | 0.03 | ||||
| Excellent | 26.8 (25.0–28.0) | 10.1 (9.4–11.0) | 16.4 (15.0–18.0) | 16.3 (15.0–18.0) | ||
| Very good | 36.5 (35.0–38.0) | 29.7 (29.0–31.0) | 36.5 (35.0–38.0) | 35.2 (34.0–37.0) | ||
| Good | 26.5 (25.0–28.0) | 35.5 (35.0–37.0) | 31.9 (30.0–33.0) | 32.7 (31.0–34.0) | ||
| Fair | 8.0 (7.2–8.8) | 18.8 (18.0–20.0) | 11.5 (10.0–13.0) | 12.1 (11.0–13.0) | ||
| Poor | 2.2 (1.9–2.7) | 5.8 (5.3–6.4) | 3.7 (3.1–4.4) | 3.7 (3.1–4.5) | ||
| Perceived mental health status*** | 0.28 | 0.02 | ||||
| Excellent | 36.1 (35.0–38.0) | 25.8 (25.0–27.0) | 30.5 (29.0–32.0) | 29.6 (28.0–31.0) | ||
| Very good | 32.9 (32.0–34.0) | 31.4 (30.0–32.0) | 33.5 (31.0–36.0) | 34.3 (33.0–36.0) | ||
| Good | 23.6 (22.0–25.0) | 30.6 (29.0–32.0) | 26.6 (25.0–28.0) | 26.6 (25.0–28.0) | ||
| Fair | 5.6 (4.9–6.3) | 9.5 (8.8–10.0) | 7.1 (6.2–8.3) | 7.2 (6.3–8.3) | ||
| Poor | 1.8 (1.5–2.3) | 2.8 (2.4–3.2) | 2.3 (1.8–3.0.8.0) | 2.2 (1.8–2.8) | ||
| CCI*** | 0.6 | 0.07 | ||||
| ≤2 | 31.3 (30.0–33.0) | 10.8 (10.0–12.0) | 20.3 (19.0–22.0) | 17.9 (16.0–19.0) | ||
| 3–5 | 64.9 (63.0–66.0) | 75.3 (74.0–76.0) | 72.8 (71.0–75.0) | 75.6 (74.0–77.0) | ||
| >5 | 3.9 (3.3–4.5) | 13.9 (13.0–15.0) | 6.9 (6.0–7.9.0.9) | 6.5 (5.6–7.6) | ||
| Hypertension*** | 48.5 (47.0–50.0) | 77.8 (77.0–79.0) | 0.64 | 67.3 (65.0–69.0) | 68.0 (66.0–70.0) | 0.02 |
| High Cholesterol*** | 45.2 (44.0–47.0) | 71.7 (70.0–73.0) | 0.56 | 61.0 (59.0–63.0) | 62.8 (61.0–65.0) | 0.04 |
| Arthritis*** | 43.8 (42.0–46.0) | 66.5 (65.0–68.0) | 0.47 | 55.3 (53.0–57.0) | 56.1 (54.0–58.0) | 0.02 |
| Cancer*** | 25.3 (24.0–27.0) | 33.5 (32.0–35.0) | 0.18 | 29.6 (28.0–32.0) | 29.7 (28.0–32.0) | 0.01 |
| Joint Pain*** | 22.3 (21.0–24.0) | 16.4 (15.0–17.0) | 0.15 | 20.0 (18.0–22.0) | 19.6 (18.0–21.0) | 0.01 |
| Other kind of heart disease*** | 15.2 (14.0–16.0) | 33.0 (31.0–34.0) | 0.42 | 22.7 (21.0–25.0) | 23.1 (21.0–25.0) | 0.01 |
| Other musculoskeletal and joint*** | 14.5 (14.0–16.0) | 29.3 (28.0–30.0) | 0.36 | 20.7 (19.0–22.0) | 21.0 (20.0–22.0) | 0.01 |
| Diabetes*** | 10.4 (9.4–11.0) | 34.6 (33.0–36.0) | 0.61 | 19.7 (18.0–22.0) | 21.3 (20.0–23.0) | 0.04 |
| Thyroid diseases*** | 9.6 (8.7–11.0) | 21.7 (21.0–23.0) | 0.34 | 14.6 (13.0–16.0) | 15.7 (14.0–17.0) | 0.03 |
| Osteoarthritis and other degenerative joint*** | 9.2 (8.3–10.0) | 23.7 (23.0–25.0) | 0.4 | 15.0 (14.0–17.0) | 15.8 (15.0–17.0) | 0.02 |
a Data presented were weighted to be nationally representative
ADL the activities of daily living, CCI the modified Charlson Comorbidity Index, FPL the federal poverty level, IADL the instrumental activities of daily living, PSM, propensity score matching, SMD standardized mean difference
*p<0.05, **p<0.01，***p<0.001 before propensity score matching
Statistical analysis
Continuous variables were presented as mean with corresponding 95% confidence intervals (CI), and categorical variables were presented as percentage with corresponding 95% CI. Descriptive analyses using the chi-square test were conducted to examine differences between older adults with and without polypharmacy.
To control for potential differences between older adults with and without polypharmacy, we employed PSM to construct a matched sample of patients with polypharmacy and those without. We used logistic regression to derive propensity scores for polypharmacy, including the aforementioned comorbidities and sociodemographic characteristics. Subsequently, propensity scores were calculated for each participant. Participants with and without polypharmacy were matched 1:1 without replacement using a nearest-neighbor approach with a caliper of 0.1 standard deviation. Pair-wise standardized mean differences (SMD) of characteristics before and after PSM were used to test balance in covariates. SMD < 0.1 was considered good balance [29]. Sample weights were taken into account during propensity score construction and matching.
We used a two-part model to estimate the associations between polypharmacy and healthcare expenditures. A probit model was first used to estimate the probability of zero versus positive healthcare expenditures. After confirming data distribution, a generalized linear model (GLM) with a gamma distribution and log-link function was used in the second part to assess the association between polypharmacy and increased expenditures, conditional on having a positive expenditure. After PSM, 54 (1.1%) patients with polypharmacy and 385 (7.81%) without polypharmacy had 0 OOP expenditures, while 115 (2.3%) patients without polypharmacy had 0 total healthcare expenditures. The model addressed the zero concentration and positive skewness of expenditures and allowed us to calculate incremental effects and standard errors from the two parts of the model [30]. Expenditures were adjusted to 2021 US dollars according to the Consumer Price Index [31]. We used logistic regression to assess the association between polypharmacy and the affordability of care.
To account for the association of pandemic on healthcare system, in sensitivity analysis, we stratified our analyses into pre-pandemic (2018–2019) and post-pandemic period (2020–2021) and replicated the analyses.
All analyses incorporated the complex survey design of MEPS and survey weights [32]. The Taylor series (linearization) method was utilized to derive standard error estimates and corresponding confidence intervals. Statistical analyses were performed using Stata, version 17 (StataCorp, College Station, TX) and R (www.R-project.org/). A two-sided p-value < 0.05 was considered statistically significant.
Result
Characteristics of the study population
The study included a total of 23,300 participants aged 65 or older (see Additional file 2). The mean age of the study population was 73.9 (95% CI: 73.8–74.2) years, with 54.5% (95% CI: 54.0–55.0) being female and 75.1% (95% CI: 73.0–77.0) being non-Hispanic White. Among the participants, 13,339 (57.0%) were categorized as polypharmacy users. Compared to those without polypharmacy, participants with polypharmacy were older, more likely to live without a partner, more likely to have function limitation, and reported lower perceived health and perceived mental health (p < 0.05 for all, Table 1). Additionally, participants with polypharmacy had a greater burden of comorbidities (Table 1 and Additional file 4).
After PSM, 4,925 participants with polypharmacy were matched with 4,925 without polypharmacy. A good balance was achieved for all covariates, including each comorbidity, with SMD < 0.1 for all variables.
Annual healthcare expenditures among older adults with and without polypharmacy
Before PSM, the total annual healthcare expenditures were significantly higher among participants with polypharmacy compared to those without ($18,492.44 [95% CI: 17,758.71-19.71,226.17] vs. $6,515.69 [95% CI: 6,077.56-6.56,953.82]) for all healthcare service, with the largest differences observed in prescription medications ($4,641.15 vs. $899.00) and office-based visits ($4,264.10 vs. $1,781.57) (Additional file 5). Similarly, out-of-pocket and Medicare expenditures were also higher among participants with polypharmacy compared to those without (p < 0.05 for both).
After PSM, the differences in healthcare expenditures between older adults with and without polypharmacy decreased but remained significant. The average annual total expenditure was $14,691.06 (95% CI: 13,658.55-15.55,723.59) for older adults with polypharmacy and $8,912.09 (95% CI: 8,232.78-9.78,591.41) for those without (p < 0.001, Fig. 1A). The largest difference was observed in prescription medication expenditure ($3,506.00 vs. $1,175.29), followed by office-based visits. Similarly, out-of-pocket costs and Medicare costs were higher for older adults with polypharmacy ($1,842.05 vs. $1,331.11 for out-of-pocket costs; $8,763.59 vs. $5,210.89 for Medicare costs, Fig. 1B and C), with the largest difference observed in prescription medication expenditures.
Fig. 1.
Healthcare Expenditures in older adults with and without polypharmacy after propensity score matching. (A) Total payments, (B) Out-of-Pocket payment, and (C) Medicare payments. Note: TOT, all health services expenses; OBV, total office-based visits expenses; RX, total prescription medicines expenses; IPT, total inpatient stays expenses; OPT, total outpatient visits expenses; ERT, total emergency room visits expenses
Estimated difference in healthcare expenditures from two-part models
After PSM, older adults with polypharmacy were more likely to have higher non-zero expenditures across various categories, including overall health services, office-based visits, outpatients visits, and prescription medicines (Table 2). On average, older adults with polypharmacy had $4,303.82 (95% CI: 3,948.56-4.56,659.08) higher annual total healthcare expenditure compared to those without polypharmacy (Table 2). This included $1,178.67 (95% CI: 1,165.10-1.10,192. 38) more for prescription medicines, $1,359.83 (95% CI: 885.14-1.14,884.52) more for office-based visits, $814.91 (95% CI: 176.77-1.77,453.06) more for inpatient stays, and $886.16 (95% CI: 322.48-1.48,449.83) more for outpatient visits.
Table 2.
Incremental healthcare expenditures associated with polypharmacy estimated by two-part regression model after propensity score matching
| Odds Ratio (95% CI) | Incremental annual cost (95%CI), $ | |
|---|---|---|
| Total expenditure | ||
| All Health Services | 1.62 (1.46-1.80) | 4,303.82 (3,948.56-4,659.08) |
| Office Based Visits | 1.48 (1.27-1.72) | 1,359.83 (835.14-1,884.52) |
| Outpatient Visits | 1.55 (1.10-2.18) | 886.16 (322.48-1,449.83) |
| Emergency Room Visits | 1.02 (0.83-1.24) | 110.55 (53.91-167.19) |
| Inpatient Stays | 0.90 (0.73-1.11) | 814.91 (176.77-1,453.06) |
| Prescription Medicines | 2.73 (2.19-3.39) | 1,178.67 (1,165.10-1,192.38) |
| Out-of-pocket | ||
| All Health Services | 1.31 (1.01-1.69) | 529.49 (163.04-896.95) |
| Office Based Visits | 1.48 (1.09-2.02) | 184.57 (34.19-335.96) |
| Outpatient Visits | 1.46 (0.84-2.53) | 36.33 (5.43-78.10) |
| Emergency Room Visits | 1.12 (0.73-1.71) | 7.38 (0.55-14.20) |
| Inpatient Stays | 0.89 (0.45-1.79) | 11.88 (0.00-46.84) |
| Prescription Medicines | 2.23 (1.93-2.58) | 270.25 (229.03-311.48) |
| Medicare | ||
| All Health Services | 1.60 (1.41-1.82) | 3,624.90 (2,702.18-4,547.63) |
| Office Based Visits | 1.38 (1.21-1.57) | 670.11 (449.87-890.36) |
| Outpatient Visits | 1.44 (1.07-1.92) | 475.03 (178.97-771.10) |
| Emergency Room Visits | 1.21 (1.02-1.43) | 98.74 (63.06-134.42) |
| Inpatient Stays | 0.89 (0.70-1.14) | 622.68 (85.36-1,160.00) |
| Prescription Medicines | 2.39 (1.85-3.09) | 1,672.46 (1,216.37-2,128.55) |
For out-of-pocket expenditures, older adults with polypharmacy had $529.49 (95% CI: 163.04–895.95.04.95) higher overall healthcare costs, including $270.25 (95% CI: 229.03–311.48.03.48) more for prescription medicines. Similarly, Medicare expenditures were substantially higher for older adults with polypharmacy compared to those without.
When examined by subgroup, the incremental costs associated with polypharmacy were significantly greater for those with functional limitations ($10,935.72 for those with IADL vs. $4,258.21 for those without, p-for-interaction = 0.02) and those with poorer perceived health status ($8,092.68 for poor vs. $2,657.10 for excellent perceived health status, p-for-interaction = 0.005, Additional file 6). Although not statistically significant, the incremental costs associated with polypharmacy were greater among the oldest old ($5,226.73 for ≥ 80 years vs. $3,871.41 for 65–69 years old), those with Medicare plus public insurance, those with poorer perceived mental health, and those with greater comorbidity burden.
In sensitivity analyses, the differences in most healthcare expenditures between patients with and without polypharmacy were comparable in pre- and post-pandemic periods. However, compared to pre-pandemic period, patients with polypharmacy had higher total expenditures for outpatient visit and lower total expenditures for inpatient stays than those without polypharmacy in post-pandemic period (incremental annual cost of polypharmacy for outpatient visit: $529.98 pre-pandemic vs. $1,249.03 post-pandemic; for inpatient stays: $1,201.02 pre-pandemic vs. $451.07 post-pandemic, Additional file 7). No substantial differences between pre- and post-pandemic estimates were observed for other costs.
Polypharmacy and affordability of care
Approximately 2.9% (95% CI: 2.6–3.2) of older adults reported unaffordability of prescription medications, 8.6% (95% CI: 8.0–9.2.0.2) reported unaffordability of dental care, and 2.3% (95% CI: 2.0–2.5.0.5) reported unaffordability of medical care. After PSM, the prevalence of unaffordability of prescription medication was significantly higher in older adults with polypharmacy compared to those without (prevalence 2.46% vs. 4.11%, OR = 1.70 [95% CI: 1.30–2.23], Table 3). In contrast, polypharmacy was not associated with higher unaffordability of dental care (OR = 1.12 [95% CI: 0.97–1.29]) or general medical treatment (OR = 1.16 [95% CI: 0.93–1.45]).
Table 3.
Prevalence of health care unaffordability among older adults
| Prevalence, % (95% CI) | Odds ratio (95% CI) for unaffordability to care associated with polypharmacy a | ||
|---|---|---|---|
| Without polypharmacy | With polypharmacy | ||
| Prescription Medicines | 2.46 (1.95–2.96) | 4.11 (3.43–4.78) | 1.70 (1.30–2.23)*** |
| Dental Care | 11.61 (10.32–12.90) | 12.79 (11.58–13.99) | 1.12 (0.97–1.29) |
| Medical Care | 3.83 (3.15–4.52) | 4.41 (3.73–5.09) | 1.16 (0.93–1.45) |
a The associations were estimated after propensity score matching
CI Confidence Interval
***p<0.001
Discussion
Using nationally representative data and propensity score matching, we quantified the associations between polypharmacy and additional healthcare expenditures as well as prescription medication affordability. Our results underscore the potential economic association of polypharmacy.
The use of a greater number of medications inevitably leads to higher prescription medication costs. Moreover, polypharmacy is associated with an increased risk of adverse outcomes, such as drug-drug interactions[6–8], which can further elevate non-medication healthcare expenditures. By employing rigorous propensity score matching and carefully balancing an extensive list of comorbidities and sociodemographic characteristics, we demonstrated that polypharmacy was associated with an additional $1,178.67 in annual prescription medication costs and $4,303.82 in total annual healthcare costs. These findings were consistent with a previous study among older adults with cardiovascular disease [12]. We showed that the increase in healthcare expenditure were borne by both out-of-pocket payments and Medicare, highlighting the potential economic burden of polypharmacy on patients, family, insurers, and society as a whole.
We also found that the prevalence of medication unaffordability was higher among older adults with polypharmacy compared to those without. Previous studies have shown that polypharmacy is associated with lower medication adherence[33], and our results suggest that cost may be a contributor to this nonadherence. Our findings underscore the potential economic burden of prescription medication in older adults with polypharmacy and emphasize the urgency of better implementation of interventions to reduce unnecessary polypharmacy.
For unnecessary polypharmacy, deprescribing is a promising intervention strategy [34]. Previous studies indicate that deprescribing interventions can reduce the number of medications, decrease the use of potentially inappropriate medication use, improve medication adherence, and lower medication costs [35]. Economic evaluations suggest that the majority of deprescribing interventions are cost-effective, making it a promising approach from an economic perspectives [36].
Our study has several strengths, including the use of nationally representative data and propensity score matching methods to carefully balance potential measured confounders between older adults with and without polypharmacy. However, our study also had limitations. First, our study was cross-sectional and did not differentiate the appropriateness of medications, thus we were not able to determine whether polypharmacy was necessary or unnecessary. Future studies are needed to further investigate the appropriateness of medication use. Second, because MEPS only includes a noninstitutionalized adult population, our results were only generalizable to adults living in the community and not those living in nursing homes. Older adults living in nursing homes are likely to have higher burden of medication use and higher healthcare expenditures. Third, we did not capture over-the-counter medication use, which may have led to an underestimation of the total medication burden. Fourth, the number of prescription medications was based on self-report, which could result in possible underreporting. Fifth, while PSM was able to balance measured differences between those with and without polypharmacy, there may still be unmeasured confounders, such as disease severity. These unmeasured confounders may result in overestimation of the associations between polypharmacy and healthcare expenditures. Sixth, due to differences in healthcare system, our results might not generalize to other countries.
In summary, utilizing nationally representative data and propensity score methods, we quantified the associations between polypharmacy and healthcare expenditures as well as prescription medication affordability among older adults in the US. The results suggest potential significant economic burden of polypharmacy and call for future studies to further examine the appropriateness of medication use and if needed, identify potential interventions to reduce unnecessary polypharmacy among older adults.
Supplementary Information
Acknowledgements
Not applicable.
Authors’ contributions
S.Pan and B.Lyu were involved in the conception, design, conduct of the study, and the analysis. All authors were involved in interpretation of the results. S.Pan and B.Lyu. wrote the first draft of the manuscript, and all authors edited, reviewed, and approved the final version of the manuscript. S.Pan and B.Lyu are the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Funding
This study was supported by the National Science Foundation of China (72404006 and 72504066).
Data availability
The data are publicaly available on MEPS website.
Declarations
Ethics approval and consent to participate
The MEPS has been reviewed and approved by the Westat Institutional Review Board and informed consent was obtained from each participant. The current study used publicly available data and individual identifiers have been removed. In accordance with the data use agreement, statistical analysis of these data is allowed. Because the current study only used publicly available data, additional ethnics review was not required in according to the common rules.
Consent for publication
Not applicable.
Competing interests
The authors declare no competing interests.
Footnotes
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Contributor Information
Hongyan Wu, Email: hongyanwu@gmc.edu.cn.
Beini Lyu, Email: blyu@pku.edu.cn.
References
- 1.Wise J. Number of older people with four or more diseases will double by 2035, study warns. BMJ. 2018;360:k371. 10.1136/bmj.k371. [DOI] [PubMed] [Google Scholar]
- 2.Ye L, Yang-Huang J, Franse CB, et al. Factors associated with polypharmacy and the high risk of medication-related problems among older community-dwelling adults in European countries: a longitudinal study. BMC Geriatr. 2022;22(1):841. 10.1186/s12877-022-03536-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Masnoon N, Shakib S, Kalisch-Ellett L, Caughey GE. What is polypharmacy? A systematic review of definitions. BMC Geriatr. 2017;17(1):230. 10.1186/s12877-017-0621-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Pan S, Li S, Jiang S et al. Trends in Number and Appropriateness of Prescription Medication Utilization Among Community-Dwelling Older Adults in the United States: 2011–2020. Magaziner J, ed. J Gerontol A: Biol Sci Med Sci. 2024;79(7):glae108. 10.1093/gerona/glae108 [DOI] [PubMed]
- 5.Murawski M, Bentley JP. Pharmaceutical therapy-related quality of life: conceptual development. J Soc Adm Pharm. 2001;18:2–14. [Google Scholar]
- 6.Hajjar ER, Cafiero AC, Hanlon JT. Polypharmacy in elderly patients. Am J Geriatr Pharmacother. 2007;5(4):345–51. 10.1016/j.amjopharm.2007.12.002. [DOI] [PubMed] [Google Scholar]
- 7.Nordin Olsson I, Runnamo R, Engfeldt P. Medication quality and quality of life in the elderly, a cohort study. Health Qual Life Outcomes. 2011;9(1):95. 10.1186/1477-7525-9-95. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Muhlack DC, Hoppe LK, Weberpals J, Brenner H, Schöttker B. The association of potentially inappropriate medication at older age with cardiovascular events and overall mortality: A systematic review and Meta-Analysis of cohort studies. J Am Med Dir Assoc. 2017;18(3):211–20. 10.1016/j.jamda.2016.11.025. [DOI] [PubMed] [Google Scholar]
- 9.WHO-Medication safety in polypharmacy: technical report. Accessed June 28. 2024. https://www.who.int/publications/i/item/WHO-UHC-SDS-2019.11
- 10.Elliott RA, Camacho E, Jankovic D, Sculpher MJ, Faria R. Economic analysis of the prevalence and clinical and economic burden of medication error in England. Bmj Qual Saf. 2021;30(2):96–105. 10.1136/bmjqs-2019-010206. [DOI] [PubMed] [Google Scholar]
- 11.Sum G, Hone T, Atun R, et al. Multimorbidity and out-of-pocket expenditure on medicines: a systematic review. BMJ Glob Health. 2018;3(1):e000505. 10.1136/bmjgh-2017-000505. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Kwak MJ, Chang M, Chiadika S, et al. Healthcare expenditure associated with polypharmacy in older adults with cardiovascular diseases. Am J Cardiol. 2022;169:156–8. 10.1016/j.amjcard.2022.01.012. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.WHO-Medication Without Harm. Accessed November 10. 2024. https://www.who.int/initiatives/medication-without-harm
- 14.Black CD, Thavorn K, Coyle D, Bjerre LM. The health system costs of potentially inappropriate prescribing: A Population-Based, retrospective cohort study using linked health administrative databases in Ontario, Canada. PharmacoEconomics - Open. 2020;4(1):27–36. 10.1007/s41669-019-0143-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Agency for Healthcare Research and Quality (AHCRa). Medical Expenditure Panel Survey Public Use File Search Results. Accessed June 28. 2024. https://meps.ahrq.gov/mepsweb/data_stats/download_data_files.jsp
- 16.Medical Expenditure Panel Survey Background. Accessed November 8. 2024. https://meps.ahrq.gov/mepsweb/about_meps/survey_back.jsp
- 17.Cohen JW, Cohen SB, Banthin JS. The medical expenditure panel survey: a National information resource to support healthcare cost research and inform policy and practice. Med Care. 2009;47(7Supplement1):S44–50. 10.1097/MLR.0b013e3181a23e3a. [DOI] [PubMed] [Google Scholar]
- 18.Medical Expenditure Panel Survey Content Summary of the Household Interview. Accessed November 8. 2024. https://meps.ahrq.gov/mepsweb/survey_comp/hc_data_collection.jsp
- 19.Medical Expenditure Panel Survey: Full Year Consolidated Data File. Accessed November 8. 2024. https://meps.ahrq.gov/data_stats/download_data/pufs/h233/h233doc.shtml
- 20.Medical Expenditure Panel Survey Download Data Files. Accessed November 8. 2024. https://meps.ahrq.gov/mepsweb/data_stats/download_data_files.jsp
- 21.Protections (OHRP) O for HR. US Department of Health and Human Services. Protection of human subjects. 45 CFR § 46. February 16, 2016. Accessed November 8. 2024. https://www.hhs.gov/ohrp/regulations-and-policy/regulations/45-cfr-46/index.html
- 22.Medical Expenditure Panel Survey: Prescribed Medicines. Accessed November 8. 2024. https://meps.ahrq.gov/data_stats/download_data/pufs/h206a/h206adoc.shtml
- 23.Methodology, Report. Outpatient Prescription Drugs: Data Collection and Editing in the 2011 Medical Expenditure Panel Survey. Accessed November 8, 2024. https://www.meps.ahrq.gov/data_files/publications/mr29/mr29.shtml#1335ImputingPP
- 24.Hill SC, Zuvekas SH, Zodet MW. Implications of the accuracy of MEPS prescription drug data for health services research. Inquiry. 2011;48(3):242–59. 10.5034/inquiryjrnl_48.03.04. [DOI] [PubMed] [Google Scholar]
- 25.Zuvekas SH, Olin GL. Accuracy of medicare expenditures in the medical expenditure panel survey. INQ: J Health Care Organ Provis Financ. 2009;46(1):92–108. 10.5034/inquiryjrnl_46.01.92. [DOI] [PubMed] [Google Scholar]
- 26.Calderón-Larrañaga A, Vetrano DL, Onder G, et al. Assessing and measuring chronic Multimorbidity in the older population: A proposal for its operationalization. J Gerontol A: Biol Sci Med Sci. 2016;72(10):glw233. 10.1093/gerona/glw233. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Agency for Healthcare Research and Quality (AHCRa). Medical Conditions. Accessed July 1. 2024. https://meps.ahrq.gov/data_stats/download_data/pufs/h207/h207doc.shtml
- 28.Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis. 1987;40(5):373–83. 10.1016/0021-9681(87)90171-8. [DOI] [PubMed] [Google Scholar]
- 29.Li F, Morgan KL, Zaslavsky AM. Balancing covariates via propensity score weighting. J Am Stat Assoc. 2018;113(521):390–400. 10.1080/01621459.2016.1260466. [Google Scholar]
- 30.Belotti F, Deb P, Manning WG, Norton EC, Twopm. Two-Part models. Stata J: Promot Commun Stat Stata. 2015;15(1):3–20. 10.1177/1536867X1501500102. [Google Scholar]
- 31.CPI Inflation Calculator, Washington DC. U.S. Burea of Labor Statistics (BLS). Bureau of Labor Statistics. Accessed June 28, 2024. https://www.bls.gov/cpi/factsheets/
- 32.Agency for Healthcare Research and Quality (AHCRa). Medical Expenditure Panel Survey.1996–2021 Pooled Linkage Variance Estimation File. Accessed June 28. 2024. https://meps.ahrq.gov/data_stats/download_data/pufs/h036/h36u21doc.shtml
- 33.Franchi C, Ardoino I, Ludergnani M, Cukay G, Merlino L, Nobili A. Medication adherence in community-dwelling older people exposed to chronic polypharmacy. J Epidemiol Community Health. 2021;75(9):854–9. 10.1136/jech-2020-214238. [DOI] [PubMed] [Google Scholar]
- 34.Scott IA, Hilmer SN, Reeve E, et al. Reducing inappropriate polypharmacy: the process of deprescribing. JAMA Intern Med. 2015;175(5):827. 10.1001/jamainternmed.2015.0324. [DOI] [PubMed] [Google Scholar]
- 35.Bloomfield HE, Greer N, Linsky AM, et al. Deprescribing for Community-Dwelling older adults: a systematic review and Meta-analysis. J Gen Intern Med. 2020;35(11):3323–32. 10.1007/s11606-020-06089-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Romano S, Figueira D, Teixeira I, Perelman J. Deprescribing interventions among Community-Dwelling older adults: A systematic review of economic evaluations. PharmacoEconomics. 2022;40(3):269–95. 10.1007/s40273-021-01120-8. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The data are publicaly available on MEPS website.