Abstract
Background:
Prior work has identified disparities in the quality and outcomes of healthcare across socioeconomic subgroups. Medication use may be subject to similar disparities
Objective:
To assess the association between demographic and socioeconomic factors (gender, age, race, income, education, and rural or urban residence) and appropriateness of medication use.
Methods:
US adults aged ≥45 years (n=26,798) from the REasons for Geographic And Racial Differences in Stroke (REGARDS) study were included in the analyses, of which 13,623 participants aged ≥65 years (recruited 2003-2007). Potentially inappropriate medication (PIM) use in older adults and drug-drug interactions (DDIs) were identified through 2015 Beers Criteria and clinically significant drug interactions list by Ament et al., respectively as measures of medication appropriateness. Multivariable logistic regression was used to assess the association of disparity parameters with PIM use and DDIs. Interactions between race and other disparity variables were investigated.
Results:
Approximately 87% of the participants aged ≥65 years used at least one drug listed in the Beers Criteria, and 3.8% of all participants used two or more drugs with DDIs. Significant gender-race interaction across prescription-only drug users revealed that white females compared with white males (OR=1.33, 95% CI 1.20-1.48) and black males compared with white males (OR=1.60, 95% CI 1.41-1.82) were more likely to receive PIM. Individuals with lower income and education also were more likely to use PIM in this sub-group. Females were less likely than males (female vs. male: OR= 0.55, 95% CI 0.48-0.63) and individuals resided in small rural areas as opposed to urban areas (small rural vs. urban: OR=1.37, 95% CI 1.07-1.76) were more likely to have DDIs.
Conclusion:
Demographic and socioeconomic disparities in PIM use and DDIs exist. Future studies should seek to better understand factors contributing to the disparities in order to guide development of interventions.
Keywords: REGARDS, potentially inappropriate medication, drug-drug interaction, socioeconomic disparity
Introduction
The United States (US) spends more per capita on health care than any other nation,1 yet health-related quality indicators lag behind many other countries.2 A low healthcare quality score is partly attributed to inequalities among subpopulations within the US.3 Health disparities are defined as significant differences in the treatment, diagnosis, and prevention of certain health conditions that are more prevalent among subpopulations versus the overall population.4 The subpopulations affected can be defined by different domains of health disparities, including the following: 1) socioeconomically disadvantage; 2) location of residence (e.g., rural vs urban setting); and 3) obstacles related to race or ethnicity.4 These factors are not mutually exclusive of one another and often occur simultaneously.
Previous research has explored the relationship between health disparities and various quality indicators reflecting health care structure, process, and outcome measures. For structure measures, for example, in 2015, the Agency for Healthcare Research and Quality (AHRQ) reported that blacks and people with low income had less access to care compared to whites and high income individuals.5 Additionally, people residing in rural areas have limited access to care which is often attributed to the lack of providers.6
For process measures, for example, uninsured and low coverage individuals report poor communication with health providers compared with those with private insurance status. Uninsured adults and those with Medicaid reported that health providers did not always seek the patient’s help when making treatment decisions nor did they spend enough time clarifying explanations with patients.7 For instance, while provider-patient interactions are vital in educating patients about their overall lifestyle, less than 40% of uninsured, obese adults report ever having their health provider give advice on dietary suggestions.7
In terms of outcome measures, numerous health disparities have been documented in terms of disease-specific outcomes, morbidity, and mortality. For example, for admissions to 7 hospitals in the Washington DC area, even among those arriving to the hospital within the prescribed 3-hour treatment window, 27% of blacks and 46% of whites received tissue plasminogen activator for treatment of their stroke.8 Moreover, among adults 25 to 44 years of age, blacks and Native Americans/American Indians were found to have a higher hemorrhage mortality rate with stroke.9,10
One important but understudied quality measure revolves around the appropriateness of medication prescribing and use. While some studies have documented health disparities in terms of the appropriateness of medication prescribing and the quality of population-based medication use (e.g., which drugs are being used by specific populations and medication adherence), data are limited to support systematic assessment of the appropriateness of medication use across health disparities defined on the basis of socioeconomic factors, rural vs. urban residence, and race. In part, this gap in the literature is caused by lack of appropriate population-level data, and in part this gap is related to lack of consensus of ways to define the "appropriateness of medication use".
Potentially inappropriate medications (PIMs) are defined as utilization of medications that have an increased risk of adverse drug events (ADEs) when an alternative exists that is equally efficacious with fewer risks.11,12 Irrational use of medicine such as use of medications at a higher dose/frequency, for longer duration than clinically indicated, under use of medicine for no clinical reasons, and use of multiple medications that are known to have clinically significant drug–drug interactions (DDIs) are also considered under the umbrella of PIM use.13,14
The quality of healthcare is sometimes characterized by the frequency in which PIMs are prescribed in older adults,15 and prior work has suggested that there may be a disparity in which people are prescribed PIMs.16,17 The focus of our study is to evaluate the appropriateness of medication use across different domains of health disparities, comparing possible disparities across a variety of previously published measures that have been used to define the appropriateness of medication use.
Methods
Data source
Data from the REasons for Geographic And Racial Differences in Stroke (REGARDS) study was used for this study. REGARDS is a population-based study initially designed to assess health disparities in stroke treatment and outcomes in the Southeastern US (“Stroke Belt”) compared to the other parts of the US.18 Participants in this study were randomly selected via mail and telephone contacts from a commercially available list of residents purchased through Genesys Inc. To be included in the study, the sample population needed to have a name, phone number and address in the Genesys database. Potential participants were introduced to the study via mailed letter and study brochure approximately two weeks prior to the telephone contact.
Participants were then contacted by telephone, and demographic, socioeconomic and cardiovascular risk factor profile characteristics were obtained via computer-assisted telephone interviewing (CATI). An in-home visit was performed within approximately 2-weeks of the telephone contact by trained personnel of Examination Management Services, Inc. (EMSI) to obtain information regarding both prescription and nonprescription medications taken within the past two weeks through pill-bottle assessment. Participants’ written informed consents were obtained during the home visit. The REGARDS study was approved by the Institutional Review Boards (IRB) of all the participating institutions. Complete methodological details of the REGARDS study are published elsewhere.18
Study population
This study included a total of 30,239 US adults age ≥ 45 years (42% black) who were recruited from January 2003 to October 2007. Exclusion criteria included race other than blacks or whites, active treatment for cancer, residing or on a waiting list for a nursing home, or inability to communicate in English.
Measures
PIM use was coded as binary variables (yes/no) using the American Geriatrics Society 2015 Updated Beers Criteria for Potentially Inappropriate Medication Use in Older Adults (described as “Beers Criteria” from here forward) which includes a list of medications that are potentially harmful and are to be avoided in older adults (age ≥ 65 years).19 The total number of medications per participant and the total number of PIMs per participant also were calculated. The total number of medications per person was converted into quartiles. PIM use was stratified across over-the-counter (OTC) and prescription-only drugs. This was determined and coded by the trained technicians of EMSI during the in-home visit.
Another measure of PIM use is based on known DDIs which could be an indicator of inappropriate drug use. DDIs were coded as binary variables (yes/no) using a known clinically significant drug interactions list published by Ament et al.20 A complete list of commonly known interacting drugs is provided in Appendix A.
Disparity parameters for this study were defined as gender (male/female), race (black/white), annual income (stratified as less than $20,000, $20,000- $34,999, $35,000- $74,999, and ≥ $75,000), education (stratified as less than high school, high school, some college, and greater than college), and location of residence (stratified as isolated, small rural, large rural, and urban). For PIM identification with the Beers Criteria, participants were restricted to age ≥65 years since Beers Criteria is only applicable for older adults. However, for analyzing DDIs we used the full cohort since there is no such age restriction and stratified age as less than 60 years, 60-64 years, 65-75 years, and greater than 75 years.
Statistical analyses
Descriptive statistics were used to report the rate of PIM use stratified across different disparity parameters. Multivariable logistic regression were used to assess the association between PIM use /DDIs and the disparity parameters. PIM use (yes/no) and DDIs (yes/no) were considered as the response variables. Gender, race, income, education, and location of residence were considered as predictors in both the multivariable models for PIM use and DDI analyses. Age was put in the multivariable model for DDI analysis only. Interaction variables was created to test for interaction between race x gender; race x age; race x income; race x education; race x region, and race x total number of medications per person. The resultant odds ratios (OR) and 95% confidence intervals (CI) were assessed as a measure of significance of association. Similar analyses were applied for PIM use across prescription-only and OTC drugs.
Income and location of residence had around 12 % and 9 % of missing data, respectively. Missing data in covariates were replaced by multivariable multiple imputation technique using chain equations in 10 datasets with sample bootstrapping.21 Prior studies recommended multiple imputation in case of data where 10 to 60% of values were missing.22
We conducted a multivariable Poisson regression as a measure of sensitivity analysis to assess the effect of different disparity parameters on total number of medications and total number of PIM use. The resultant prevalence ratio (PR) and 95% CI were assessed as a measure of significance of association. The interaction of race with other variables were investigated. All analyses were performed using SAS, version 9.4 (SAS Institute, Inc., Cary, NC).
Results
3,441 out of the 30,239 participants were excluded due to missing drug inventory information. So, a total of 26,798 participants were present in the overall cohort out of which 13,623 were of age ≥ 65 years (38% black). Out of 13,623 participants aged ≥ 65 years, a total of 11,912 (87.4%) used at least one drug listed in the Beers criteria, and out of all the participants (n=26,798), 1014 (3.8%) used two or more drugs with DDIs. On an average, each participant aged 65 and over used 6.5 medications which ranged from 1 to 20. The mean number of PIM use per person aged ≥ 65 years was 2.4 which ranged from 0 to 14. The most frequently appearing PIMs include aspirin (35.7%), naproxen (3.2%), ibuprofen (3%), celecoxib (2.8%), and pioglitazone (2.7%).
Characteristics of the PIM and DDI users and non-users are shown in Table 1. Significant differences were observed between PIM users and non-users. For example, PIM users compared to the non-users were older (mean age 72.8 vs 72.3 years) and more likely whites (62.3% vs 59.6%). PIM users had significantly lower income, education, and mainly resided in rural areas compared with non-users. Similar differences were observed for DDI users and non-users. For example, DDI users compared to non-users were older (mean age 67.4 vs 65.2 years), more likely male (54.8% vs 43.4%), whites (76.5% vs 58.5%), and resided in rural areas.
Table 1.
Baseline Characteristics of the Study Population by Use of PIM or DDI.
| Characteristic | PIM use (n = 13 623) | DDI (n = 26 798) | ||||
|---|---|---|---|---|---|---|
| Yes (n = 11 912) |
No (n = 1711) |
P Value | Yes (n = 1014) |
No (n = 25 784) |
p-value | |
| Age, mean (SD), years | 72.8 ± 5.9 | 72.3 ± 5.8 | 0.0002 | 67.4 ± 9.1 | 65.2 ± 9.4 | <0.0001 |
| Female, n (%) | 6375 (53.5) | 887 (51.8) | 0.19 | 458 (45.2) | 14 603 (56.6) | <0.0001 |
| Black, n (%) | 4492 (37.7) | 692 (40.4) | 0.03 | 238 (23.5) | 10 697 (41.5) | <0.0001 |
| Income, n (%) | 0.01 | 0.47 | ||||
| Less than $20,000 | 2578 (21.6) | 319 (18.6) | 185 (18.2) | 4727 (18.3) | ||
| $20,000- $34,999 | 3404 (28.6) | 473 (27.6) | 270 (26.6) | 6240 (24.2) | ||
| $35,000- $74,999 | 3169 (26.6) | 494 (28.9) | 292 (28.8) | 7590 (29.4) | ||
| ≥ $75,000 | 1040 (8.7) | 173 (10.1) | 147 (14.5) | 4002 (15.5) | ||
| Education, n (%) | 0.0008 | 0.29 | ||||
| Less than high school | 1998 (16.8) | 246 (14.4) | 109 (10.8) | 3293 (12.8) | ||
| High school | 3231 (27.2) | 433 (25.3) | 266 (26.3) | 6734 (26.1) | ||
| Some college | 2967 (24.9) | 422 (24.5) | 278 (27.44) | 6873 (26.7) | ||
| College graduate and above | 3702 (31.1) | 609 (35.6) | 360 (35.5) | 8864 (34.4) | ||
| Location of residence, n (%) | 0.0079 | 0.0003 | ||||
| Isolated | 262 (2.4) | 27 (1.7) | 18 (2.0) | 525 (2.3) | ||
| Small rural | 656 (6.1) | 82 (5.3) | 78 (8.6) | 1359 (5.8) | ||
| Large rural | 1306 (12.1) | 155 (9.9) | 130 (14.3) | 2727 (11.7) | ||
| Urban | 8576 (79.4) | 1294 (83.1) | 684 (75.2) | 18 691 (80.2) | ||
PIM: Potentially Inappropriate Medication; for PIM use, the patients are of age >=65; DDI: Drug-Drug Interactions, for DDIs, the patients are of age >= 45 years;
For prescription-only drugs, higher rates of PIM use were observed for females compared with males, blacks compared with whites, and individuals with lower income and education (Table 2). However, for OTC drugs, we observed an opposing patterns in terms of gender and race. No consistent patterns were found in terms of income, education, and location of residence.
Table 2.
Descriptive statistics of PIM use across prescription vs. OTC drugs in different subgroups of the study population age >=65
| Characteristic | PIM use across prescription-only drugs*, n (%)** |
PIM use across OTC drugs*, n (%)** |
|---|---|---|
| Gender | ||
| Male | 4719 (79.4) | 2709 (63.0) |
| Female | 5815 (84.3) | 2677 (52.2) |
| Race | ||
| White | 6416 (81.3) | 3708 (58.1) |
| Black | 4118 (83.3) | 1678 (55.1) |
| Income | ||
| ≥$75,000 | 860 (76.4) | 528 (60.5) |
| $35,000 - $74,999 | 2754 (79.8) | 1493 (57.3) |
| $20,000 - 34,999 | 3033 (83.0) | 1515 (56.2) |
| < $20,000 | 2347 (85.6) | 1060 (56.5) |
| Education | ||
| >College | 3156 (78.6) | 1774 (57.0) |
| Some college | 2641 (82.8) | 1319 (55.7) |
| High school | 2897 (83.2) | 1462 (57.5) |
| <High school | 1827 (85.5) | 824 (59.2) |
| Location of residence | ||
| Urban | 7640 (81.9) | 3739 (56.4) |
| Large rural | 1123 (82.0) | 657 (60.9) |
| Small rural | 559 (81.1) | 345 (62.4) |
| Isolated | 226 (81.9) | 122 (52.6) |
The same person may have both prescription and over-the-counter (OTC) drugs and thus, may be counted more than once.
% calculated as of the total corresponding population
Gender-race and the total number of medications-race interactions were found to be significant in our multivariable logistic regression models for PIM use across all the drugs and prescription-only drugs (Table 3). White males demonstrated a significantly higher odds of having PIMs compared with white females (female vs. male among whites: OR= 0.84, 95% CI 0.74 – 0.96). Higher number of medication use was a significant predictor of PIM use across all the drug types. For PIM use across prescription-only drugs, income and education were independent predictors of PIM use whereby individuals with lower income and education had significantly higher odds of having a PIM prescription For example, the odds of PIM prescription among individuals with an annual income of less than $20,000 was 26% more than those with annual income ≥ $75,000. Similarly, participants with education level less than high school had 31% more odds of having a PIM prescription than the individuals with college degrees or above. A significant gender-race interaction indicated that black males compared with white males (OR= 1.60, 95% CI 1.41 – 1.82) and white females compared with white males (OR= 1.33, 95% CI 1.20 – 1.48) had higher odds of PIM prescription.
Table 3.
Disparities across PIM use and DDIs
| Characteristic | PIM use across all drugs, OR (95% CI) |
PIM use across prescription-only drugs, OR (95% CI) |
PIM use across OTC drugs, OR (95% CI) |
DDIs, OR (95% CI) |
|---|---|---|---|---|
| Gender | ||||
| Male | Ref | Ref | Ref | Ref |
| Female | a | a | 0.61 (0.56 - 0.67) | 0.55 (0.48 - 0.63) |
| Race | ||||
| White | Ref | Ref | Ref | Ref |
| Black | a, b | a, b | c | d |
| Income | ||||
| ≥ $75,000 | Ref | Ref | Ref | Ref |
| Less than $20,000 | 1.16 (0.86 - 1.44) | 1.26 (1.01 - 1.57) | c | 1.22 (0.93 - 1.60) |
| $20,000- $34,999 | 1.06 (0.85 - 1.34) | 1.24 (1.02 - 1.51) | 1.11 (0.87 - 1.41) | |
| $35,000- $74,999 | 0.96 (0.77 - 1.20) | 1.11 (0.92 - 1.33) | 1.02 (0.81 - 1.28) | |
| Education | ||||
| College graduate and above | Ref | Ref | Ref | Ref |
| Less than high school | 1.15 (0.94 - 1.41) | 1.31 (1.09 - 1.56) | 1.16 (1.01 - 1.33) | 0.86 (0.66 - 1.11) |
| High school | 1.08 (0.92 - 1.28) | 1.17 (1.01 - 1.34) | 1.12 (0.99 - 1.26) | 0.96 (0.79 - 1.16) |
| Some college | 1.09 (0.92 - 1.27) | 1.19 (1.04 - 1.37) | 1.00 (0.89 - 1.12) | 0.97 (0.81 - 1.16) |
| Location of residence | ||||
| Urban | Ref | Ref | Ref | Ref |
| Isolated | 1.45 (0.94 - 2.24) | 1.09 (0.77 - 1.54) | 0.84 (0.64 - 1.09) | 0.76 (0.47 - 1.24) |
| Small rural | 1.24 (0.96 - 1.62) | 0.93 (0.75 - 1.15) | 1.29 (1.07 - 1.54) | 1.37 (1.07 - 1.76) |
| Large rural | 1.20 (0.99 - 1.46) | 0.96 (0.82 - 1.13) | 1.18 (1.03 - 1.35) | 1.07 (0.88 - 1.31) |
| Total number of medications | ||||
| First quartile | Ref | Ref | Ref | Ref |
| Second quartile | b | b | 1.42 (1.20 - 1.68) | 9.22 (4.28 - 19.84) |
| Third quartile | 1.73 (1.46 - 2.04) | 28.86 (13.59 - 61.29) | ||
| Fourth quartile | 1.78 (1.50 - 2.10) | 76.01 (35.97 - 160.61) | ||
| Age group, years | ||||
| < 60 | - | - | - | Ref |
| 60-64 | - | - | - | d |
| 65-75 | - | - | - | |
| ≥ 75 | - | - | - | |
| Significant interactions | ||||
| Gender-Race a | ||||
| Female vs. Male; Race = Black | 0.86 (0.73 - 1.01) | 0.93 (0.81 - 1.07) | - | - |
| Female vs. Male; Race = White | 0.84 (0.74 - 0.96) | 1.33 (1.20 - 1.48) | - | - |
| Black vs. White; Gender = Female | 1.12 (0.97 - 1.29) | 1.12 (0.99 - 1.26) | - | - |
| Black vs. White; Gender = Male | 1.06 (0.91 - 1.24) | 1.60 (1.41 - 1.82) | - | - |
| Total number of medications-Race b | ||||
| Second vs. first quartile; Race = Black | 5.06 (4.28 - 5.97) | 3.22 (2.74 - 3.77) | - | - |
| Third vs. first quartile; Race = Black | 26.19 (19.61 - 34.97)* | 11.32 (9.18 - 13.97)* | - | - |
| Fourth vs. first quartile; Race = Black | 65.53 (40.79 - 105.29)* | 31.45 (22.72 - 43.53)* | - | - |
| Second vs. first quartile; Race = White | 4.62 (3.99 - 5.34) | 2.50 (2.18 - 2.87) | - | - |
| Third vs. first quartile; Race = White | 12.48 (10.36 - 15.04)* | 5.96 (5.12 - 6.93)* | - | - |
| Fourth vs. first quartile; Race = White | 38.13 (28.97 - 50.18)* | 16.98 (14.03 - 20.52)* | - | - |
| Income-Race c | ||||
| Less than $20,000 vs. ≥ $75,000; race = Black | - | - | 1.40 (0.94 - 2.08) | - |
| $20,000- $34,999 vs. ≥ $75,000; race = Black | - | - | 1.00 (0.67 - 1.47) | - |
| $35,000- $74,999 vs. ≥ $75,000; race = Black | - | - | 1.06 (0.71 - 1.58) | - |
| Less than $20,000 vs. ≥ $75,000; race = White | - | - | 0.76 (0.51 - 1.01) | - |
| $20,000- $34,999 vs. ≥ $75,000; race = White | - | - | 0.87 (0.72 - 1.06) | - |
| $35,000- $74,999 vs. ≥ $75,000; race = White | - | - | 0.83 (0.69 – 1.00) | - |
| Black vs. White; income = Less than $20,000 | - | - | 1.32 (1.08 - 1.61) | - |
| Black vs. White; income = $20,000- $34,999 | - | - | 0.81 (0.61 - 1.02) | - |
| Black vs. White; income = $35,000- $74,999 | - | - | 0.91 (0.75 - 1.10) | - |
| Black vs. White; income = ≥ $75,000 | - | - | 0.71 (0.48 - 1.06) | - |
| Age group-Race d | ||||
| 60-64 vs. < 60; race = Black | - | - | - | 1.16 (0.79 - 1.71) |
| 65-75 vs. < 60; race = Black | - | - | - | 0.89 (0.63 - 1.26) |
| ≥ 75 vs. < 60; race = Black | - | - | - | 0.74 (0.46 - 1.20) |
| 60-64 vs. < 60; race = White | - | - | - | 1.12 (0.86 - 1.45) |
| 65-75 vs. < 60; race = White | - | - | - | 1.05 (0.84 - 1.32) |
| ≥ 75 vs. < 60; race = White | - | - | - | 1.48 (1.16 - 1.90) |
| Black vs. White; age = < 60 | - | - | - | 0.60 (0.43 - 0.83) |
| Black vs. White; age = 60-64 | - | - | - | 0.63 (0.45 - 0.88) |
| Black vs. White; age = 65-75 | - | - | - | 0.51 (0.39 - 0.66) |
| Black vs. White; age ≥ 75 | - | - | - | 0.30 (0.19 - 0.46) |
Multivariable logistic regression findings (odds ratio [OR] and 95% CIs) for the associations of PIM use (including OTC and prescription-only drugs) and DDIs with disparity parameters (gender, age, race, income, education, and location of residence) and total medication use.
Significant gender-race interaction
Significant total number of medications-race interaction
Significant income-race interaction
Significant age group-race interaction
PIM: Potentially inappropriate medication; DDI: Drug-drug interaction; OTC: Over the counter; na: not applicable.
ORs such big are usually driven by small cell size and are unstable and may overestimate the study interpretation.53
Males compared with females (female vs male: OR= 0.61, 95% CI 0.56 – 0.67), individuals with lower education (less than high school vs college graduate and above: OR= 1.16, 95% CI 1.01 – 1.33), and people residing in rural areas (small rural vs urban: OR= 1.29, 95% CI 1.07 – 1.54); large rural vs urban: OR= 1.18, 95% CI 1.03 – 1.35) had significantly higher odds of OTC PIM use. A significant income-race interaction in this group elucidated that blacks compared with whites had higher odds of OTC PIM use in the lower income subgroups (for annual income less than $20,000, black vs white: OR= 1.32, 95% CI 1.08 – 1.61). We also found similar disparities across the use of interacting drugs together. Males compared with females (female vs male: OR= 0.55, 95% CI 0.48 – 0.63) and individuals residing in rural areas (small rural vs urban: OR 1.37, 95% CI 1.07 – 1.76) had significantly higher odds of having DDIs. Higher medication use also was a significant predictor of DDIs. A significant age-race interaction suggested that whites compared with blacks have higher odds of DDIs regardless of their age.
Multivariable Poisson regression model revealed that females compared with males, whites compared with blacks, individuals with lower income, education and residing in rural areas had higher medication use across all drugs and also for prescription-only drugs (Table 4). Similarly, for OTC drugs, females compared with males, whites compared with blacks, and individuals with lower education were more likely to use more medications. Evaluation of factors affecting the total number of PIM use revealed that all the disparity parameters and total number of medications were significant predictors of total PIM count across all drugs, prescription-only, and OTC drugs except for gender (Figure 1). Gender was not a significant predictor for OTC PIM count. We found that females compared with males, blacks compared with whites, and individuals with lower income, education and residing in rural areas had significantly higher PIM prevalence compared with the individuals with higher income and education, and urban dwellers. We also found that more medication use was significantly associated with higher PIM prevalence.
Table 4.
Effect of different disparity parameters on total medication count
| Characteristic | PIM use across all drugs, PR (95% CI) |
PIM use across prescription-only drugs, PR (95% CI) |
PIM use across OTC drugs, PR (95% CI) |
|---|---|---|---|
| Gender | |||
| Male | Ref | Ref | Ref |
| Female | 1.11 (1.09 - 1.12) | 1.09 (1.08 - 1.11) | 1.12 (1.10 - 1.14) |
| Race | |||
| White | Ref | Ref | Ref |
| Black | 0.86 (0.85 - 0.88) | 0.86 (0.85 - 0.87) | 0.89 (0.87 - 0.91) |
| Income | |||
| ≥ $75,000 | Ref | Ref | Ref |
| Less than $20,000 | 1.03 (1.01 - 1.07) | 1.04 (1.01 - 1.08) | 0.99 (0.96 - 1.02) |
| $20,000- $34,999 | 1.01 (0.98 - 1.04) | 1.02 (0.99 - 1.05) | 0.99 (0.96 - 1.02) |
| $35,000- $74,999 | 1.00 (0.97 - 1.03) | 1.00 (0.98 - 1.03) | 1.00 (0.97 - 1.03) |
| Education | |||
| College graduate and above | Ref | Ref | Ref |
| Less than high school | 1.04 (1.01 - 1.06) | 1.03 (1.01 - 1.05) | 1.04 (1.01 - 1.07) |
| High school | 1.00 (0.98 - 1.02) | 0.99 (0.97 - 1.01) | 0.99 (0.97 - 1.01) |
| Some college | 1.03 (1.01 - 1.05) | 1.02 (1.01 - 1.04) | 1.03 (1.01 - 1.05) |
| Location of residence | |||
| Urban | Ref | Ref | Ref |
| Isolated | 1.01 (0.97 - 1.06) | 1.02 (0.97 - 1.06) | 0.98 (0.93 - 0.99) |
| Small rural | 0.99 (0.96 - 1.02) | 0.99 (0.96 - 1.02) | 0.96 (0.93 - 1.00) |
| Large rural | 1.03 (1.01 - 1.05) | 1.03 (1.01 - 1.05) | 1.01 (0.99 - 1.04) |
Multivariable Poisson regression findings (prevalence ratio [PR] and 95% CIs) to study the associations of disparity parameters (gender, race, income, education, and location of residence) with total medication use (including OTC and prescription-only drugs).
No significant interaction was found
Figure 1.
Effect of different disparity parameters on total PIM count
Multivariable Poisson regression findings (prevalence ratio [PR] and 95% CIs) to study the associations of disparity parameters (gender, race, income, education, and location of residence) with total number of PIM use (including OTC and prescription-only drugs).
* No significant interaction was found
Discussion
We found that race, sex, and socioeconomic status was strongly associated with the likelihood of a prescribing pattern of PIMs and DDIs. Black males and white females were more likely to have PIM use. Men and residents of rural areas were more likely to be exposed to DDIs.
Demographic and socioeconomic disparities in different aspects of healthcare have been reported in prior studies. For example, white women were reported to have an increase in death rates and a decline in life expectancies than white men23 which are often linked with the rise in the use of prescription opioids.24 According to the Beers Criteria, opioids are considered to be potentially inappropriate in older adults and are recommended to avoid with a history of fall or fracture.19 It has been found that women are more likely than men to be diagnosed with depression and take antidepressants.25-27 Moreover, whites have more access to psychiatric services than blacks and are more likely to take antidepressants.26,27 Many antidepressants are also considered as potentially inappropriate according to the Beers Criteria.19 In our study, white females compared to white males had significantly higher odds of receiving PIM prescription which is consistent with some of the prior studies focusing on gender-race inequality.
Prior studies demonstrated that racial disparities are most persistent, most difficult to address, and shapes other socioeconomic disparities.28,29 It has been found that African-American patients have a less participatory relationship with their physicians than whites.30 Oliver et al. found that white physicians spent more time with white patients than African-American patients for planning a treatment, evaluating health literacy, providing health education, and answering questions.31 In our study, although whites received more medications than blacks, blacks were more likely than whites to receive more PIMs. It is possible that blacks are being treated differently by the providers. However, further study is needed to know the providers’ prescribing behavior.
In general, low income individuals across all races have comparatively poor health status than their higher income counterparts.32 Additionally, people with lower income have inadequate healthcare coverage.25 Lower income was associated with a significantly higher odds of receiving PIM prescription in our study which supports the findings of prior studies. The reasons mentioned above could also be the driving factors for higher PIM prescription in these lower income subgroups.
We also explored disparities across education and location of residence. Our study found a significant association between lower education and higher PIM use. Studies have shown that higher education is linked to better job, better income, and better health literacy and behavior.33 Lower health literacy could be an influential factor for using more OTC PIMs. Moreover, individuals with lower income and education are more likely to live in poor neighborhoods which may lack resources for good health.33 Disparities among individuals living in smaller geographic location has been described in many studies.34,35 Individuals living in the rural areas lack timely access to healthcare providers and have limited availability of subspecialty physicians.36 All these factors can interplay and contribute to inappropriate use of medications in the population with lower education and residing in rural areas.
Similar factors also could contribute to the higher odds of DDIs in rural residents compared to the urban dwellers in our study. Total number of medications used by the study participants remained a significant predictor of DDIs in our study. We observed that higher the number of medications, higher the odds of DDIs. We also found that whites compared with blacks had significantly higher odds of having DDIs, and older age was a significant predictor of DDI for whites only. Prior studies have found that whites are more likely than blacks to receive more medications.26,37-39 Moreover, older adults are often present with multiple chronic conditions that may require the use of multiple drugs concomitantly which in turn can increase the risk DDIs.40 It has been found that patients taking two medications face a 13% risk of DDIs. This percentage rises up to 38% for patients taking four medications and rises up to 82% if seven or more medications are given concomitantly.41 Our findings also support the conclusions derived from the prior studies.
The root causes of health inequality are very diverse, complex, and often difficult to understand.25 One reason for health disparity is considered to be the implicit bias from the providers’ perspective which is defined by John Dovidio as “unconscious discrimination”.42 These biases may not be arbitrary and could be shaped by the racial stratification and societal norms.25 Due to the implicit biases, physicians are found to treat patients differently based on their race, ethnicity, or gender rather than the actual underlying conditions.43,44 Functional magnetic resonance imaging (MRI) of brains also revealed that white providers have implicit biases against African-Americans.45 Experts often suggest for patient-provider concordance to overcome such biases.44,46 Patient-provider concordance can happen in terms of gender, social class, age, ethnicity, race, language, sexual orientation, beliefs about health and illness, values, and health care decisions.47 However, debate exists as to whether this concordance would help overcome health disparities.47 Studies showed that patients prefer providers who treat them more respectfully rather than the providers of their own race or ethnicity.48,49 Pharmacists also can play an important role in reducing health disparities in terms of PIM use. For example, when they receive a new prescription for a drug which is potentially inappropriate for older patients, they can consider discussing it with the prescriber for a safer but equally efficacious alternative regardless of the race, gender, or other pertaining disparity parameters. While advising the patients about OTC drugs, pharmacists can recommend the drugs that are not considered as PIMs. Appropriate training programs for the pharmacists on the harmful effect of PIMs and the related disparities can help reduce this problem.
Our study has some limitations. First, we used the 2015 Beers Criteria to identify PIMs. There are other explicit criteria to identify PIMs which have shown differences in detecting PIMs in prior studies.50,51 However, the Beers criteria was developed in the US and is the most widely used tool for PIM identification.52 Since we are looking for disparities in PIM use in the US population, we believe that the Beers Criteria is an appropriate tool for PIM identification for our study. Still, it is possible to have some misclassification bias. Second, our study lacked information regarding the indication for which the medications were prescribed. Beers recommendations are not absolute and confounding by indication is still possible. Additionally, although the trained EMSI personnel performed rigorous pill bottle assessment, medication doses were not recorded. As a result, we could not establish a dose-response relationship. Furthermore, we did not have information regarding the provider’s characteristics. In future studies, it would be helpful to know the pattern of the prescribers and their prescribing behavior. And finally, some of the data may be as old as 15 years and Beers Criteria has evolved over time, thereby potentially changing therapeutic interpretation. Moreover, not only the Beers Criteria been updated but also the prescribing patterns have likely changed over time, so this may not be reflective of actual current practice.
Conclusion
To the best of our knowledge, this is the first attempt to evaluate disparities in the appropriateness of medication use. Our study found significant demographic and socioeconomic disparities in PIM use. Although medication prescription is a process measure, an inappropriate prescription can lead to poor outcomes especially in older adults. Future studies should seek to better understand factors contributing to the disparities in order to guide development of interventions.
Supplementary Material
Acknowledgments
Funding Source
This research project is supported by a cooperative agreement U01 NS041588 from the National Institute of Neurological Disorders and Stroke, National Institutes of Health, Department of Health and Human Service. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institute of Neurological Disorders and Stroke or the National Institutes of Health.
Footnotes
Competing interests
No authors declare any competing interest.
Declaration of interest
In the past 3 years, Richard A. Hansen has provided expert testimony for Daiichi Sankyo and Takeda. Md Motiur Rahman is currently an employee of Johnson and Johnson. No other authors declare any potential competing interest.
References
- 1.US healthcare spending compared to others. Peterson-Kaiser Health System Tracker. http://www.healthsystemtracker.org/chart-collection/health-spending-u-s-compare-countries/?_sf_s=health+spending#item-start. Accessed July 7, 2017.
- 2.Squires DA. The U.S. health system in perspective: a comparison of twelve industrialized nations. Issues in International Health Policy. 2011. http://www.commonwealthfund.org/~/media/Files/Publications/Issue%20Brief/2011/Jul/1532_Squires_US_hlt_sys_comparison_12_nations_intl_brief_v2.pdf. Accessed July 7, 2017. [PubMed]
- 3.Gary Claxton CC, Selena Gonzales, Rabah Kamal, Larry Levitt. Measuring the quality of healthcare in the U.S. https://www.healthsystemtracker.org/brief/measuring-the-quality-of-healthcare-in-the-u-s/#item-start. Accessed April 22, 2018 2015.
- 4.Howard G Ancel Keys Lecture: Adventures (and misadventures) in understanding (and reducing) disparities in stroke mortality. Stroke. 2013;44(11):3254–3259. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.AHRQ. 2015. National Healthcare Quality and Disparities Report. http://www.ahrq.gov/sites/default/files/wysiwyg/research/findings/nhqrdr/nhqdr15/2015nhqdr.pdf. Accessed on September 1, 2016.
- 6.RURAL AND URBAN HEALTH. Georgetown University; Health Policy Institute. https://hpi.georgetown.edu/agingsociety/pubhtml/rural/rural.html. Accessed August 5, 2018. [Google Scholar]
- 7.Agency for Healthcare Research and Quality, 2014. Health care coverage analyses of the National Healthcare Quality and Disparities Reports: 2000–2008 trends. https://www.medicaid.gov/medicaid/quality-of-care/downloads/health-coverage-analyses-of-nhqr-2000-2008-trends.pdf. Accessed July 8, 2017. [Google Scholar]
- 8.Hsia AW, Edwards DF, Morgenstern LB, et al. Racial disparities in tissue plasminogen activator treatment rate for stroke: a population-based study. Stroke. 2011;42(8):2217–2221. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Go AS, Mozaffarian D, Roger VL, et al. Heart disease and stroke statistics−-2013 update: a report from the American Heart Association. Circulation. 2013;127(1):e6–e245. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Ayala C, Greenlund KJ, Croft JB, et al. Racial/ethnic disparities in mortality by stroke subtype in the United States, 1995–1998. Am J Epidemiol. 2001;154(11):1057–1063. [DOI] [PubMed] [Google Scholar]
- 11.American Geriatrics Society updated Beers Criteria for potentially inappropriate medication use in older adults. J Am Geriatr Soc. 2012;60(4):616–631. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Chen LL, Tangiisuran B, Shafie AA, Hassali MA. Evaluation of potentially inappropriate medications among older residents of Malaysian nursing homes. Int J Clin Pharm. 2012;34(4):596–603. [DOI] [PubMed] [Google Scholar]
- 13.Gallagher P, Barry P, O’Mahony D. Inappropriate prescribing in the elderly. J Clin Pharm Ther. 2007;32(2):113–121. [DOI] [PubMed] [Google Scholar]
- 14.Institute of Medicine Committee on Quality of Health Care in A. In: Kohn LT, Corrigan JM, Donaldson MS, eds. To Err is Human: Building a Safer Health System. Washington (DC): National Academies Press (US)Copyright 2000 by the National Academy of Sciences. All rights reserved.; 2000.
- 15.Vieira de Lima TJ, Garbin CA, Garbin AJ, Sumida DH, Saliba O. Potentially inappropriate medications used by the elderly: prevalence and risk factors in Brazilian care homes. BMC Geriatr. 2013;13:52. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Felton M, Hanlon JT, Perera S, Thorpe JM, Marcum ZA. Racial differences in anticholinergic use among community-dwelling elders. Consult Pharm. 2015;30(4):240–245. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Healthy people 2020. Disparities Overview by Income. https://www.healthypeople.gov/2020/data/disparities/summary/Chart/4155/6.1. Accessed on April 25, 2017.,. [Google Scholar]
- 18.Howard VJ, Cushman M, Pulley L, et al. The reasons for geographic and racial differences in stroke study: objectives and design. Neuroepidemiology. 2005;25(3):135–143. [DOI] [PubMed] [Google Scholar]
- 19.American Geriatrics Society 2015 Updated Beers Criteria for Potentially Inappropriate Medication Use in Older Adults. J Am Geriatr Soc. 2015;63(11):2227–2246. [DOI] [PubMed] [Google Scholar]
- 20.Ament PW, Bertolino JG, Liszewski JL. Clinically significant drug interactions. Am Fam Physician. 2000;61(6):1745–1754. [PubMed] [Google Scholar]
- 21.Hansen RA, Khodneva Y, Glasser SP, Qian J, Redmond N, Safford MM. Antidepressant Medication Use and Its Association With Cardiovascular Disease and All-Cause Mortality in the Reasons for Geographic and Racial Differences in Stroke (REGARDS) Study. Ann Pharmacother. 2016;50(4):253–261. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Barzi F, Woodward M. Imputations of missing values in practice: results from imputations of serum cholesterol in 28 cohort studies. Am J Epidemiol. 2004;160(1):34–45. [DOI] [PubMed] [Google Scholar]
- 23.Arias E Changes in Life Expectancy by Race and Hispanic Origin in the United States, 2013–2014. NCHS Data Brief. 2016(244):1–8. [PubMed] [Google Scholar]
- 24.Nan Marie Astone SM, Laudan Y. Aron. Death Rates for US Women Ages 15 to 54. https://www.urban.org/research/publication/death-rates-us-women-ages-15-54. Accessed August 27, 2018 2015.
- 25.National Academies of Sciences E, Medicine. Communities in Action: Pathways to Health Equity. Washington, DC: The National Academies Press; 2017. [PubMed] [Google Scholar]
- 26.Mundell EJ. U.S. Antidepressant Use Jumps 65 Percent in 15 Years. https://consumer.healthday.com/mental-health-information-25/antidepressants-news-723/u-s-antidepressant-use-jumps-65-percent-in-15-years-725586.html. Accessed August 27, 2018 2017.
- 27.Laura A Pratt DJB, Qiuping Gu. Antidepressant Use Among Persons Aged 12 and Over: United States, 2011–2014. https://www.cdc.gov/nchs/products/databriefs/db283.htm. Accessed August 27, 2018 2017. [PubMed]
- 28.Williams DR, Mohammed SA. Discrimination and racial disparities in health: evidence and needed research. J Behav Med. 2009;32(1):20–47. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Farmer MM, Ferraro KF. Are racial disparities in health conditional on socioeconomic status? Soc Sci Med. 2005;60(1):191–204. [DOI] [PubMed] [Google Scholar]
- 30.Cooper-Patrick L, Gallo JJ, Gonzales JJ, et al. Race, gender, and partnership in the patient-physician relationship. JAMA. 1999;282(6):583–589. [DOI] [PubMed] [Google Scholar]
- 31.Oliver MN, Goodwin MA, Gotler RS, Gregory PM, Stange KC. Time use in clinical encounters: are African-American patients treated differently? J Natl Med Assoc. 2001;93(10):380–385. [PMC free article] [PubMed] [Google Scholar]
- 32.Braveman PA, Cubbin C, Egerter S, Williams DR, Pamuk E. Socioeconomic disparities in health in the United States: what the patterns tell us. Am J Public Health. 2010;100 Suppl 1:S186–196. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Health VCfSa. Why Education Matters to Health: Exploring the Causes. https://societyhealth.vcu.edu/work/the-projects/why-education-matters-to-health-exploring-the-causes.html#healthBenefits. Accessed August 27, 2018 2015.
- 34.Kulkarni SC, Levin-Rector A, Ezzati M, Murray CJ. Falling behind: life expectancy in US counties from 2000 to 2007 in an international context. Popul Health Metr. 2011;9(1):16. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Emily Zimmerman SHW. Understanding the Relationship Between Education and Health. IOM Roundtable on Population Health Improvement. https://nam.edu/wp-content/uploads/2015/06/BPH-UnderstandingTheRelationship1.pdf. Accessed August 27, 2018 2014.
- 36.Wong ST, Regan S. Patient perspectives on primary health care in rural communities: effects of geography on access, continuity and efficiency. Rural Remote Health. 2009;9(1):1142. [PubMed] [Google Scholar]
- 37.Khandker RK, Simoni-Wastila LJ. Differences in prescription drug utilization and expenditures between Blacks and Whites in the Georgia Medicaid population. Inquiry. 1998;35(1):78–87. [PubMed] [Google Scholar]
- 38.Wang J, Zuckerman IH, Miller NA, Shaya FT, Noel JM, Mullins CD. Utilizing New Prescription Drugs: Disparities among Non-Hispanic Whites, Non-Hispanic Blacks, and Hispanic Whites. Health Serv Res. 2007;42(4):1499–1519. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Singhal A, Tien YY, Hsia RY. Racial-Ethnic Disparities in Opioid Prescriptions at Emergency Department Visits for Conditions Commonly Associated with Prescription Drug Abuse. PLoS One. 2016;11(8):e0159224. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Hajjar ER, Cafiero AC, Hanlon JT. Polypharmacy in elderly patients. Am J Geriatr Pharmacother. 2007;5(4):345–351. [DOI] [PubMed] [Google Scholar]
- 41.Goldberg RM, Mabee J, Chan L, Wong S. Drug-drug and drug-disease interactions in the ED: analysis of a high-risk population. Am J Emerg Med. 1996;14(5):447–450. [DOI] [PubMed] [Google Scholar]
- 42.Dovidio JF, Gaertner SL, Kawakami K, Hodson G. Why can’t we just get along? Interpersonal biases and interracial distrust. Cultur Divers Ethnic Minor Psychol. 2002;8(2):88–102. [DOI] [PubMed] [Google Scholar]
- 43.Institute of Medicine Committee on U, Eliminating R, Ethnic Disparities in Health C In: Smedley BD, Stith AY, Nelson AR, eds. Unequal Treatment: Confronting Racial and Ethnic Disparities in Health Care. Washington (DC): National Academies Press (US)Copyright 2002 by the National Academy of Sciences. All rights reserved.; 2003.
- 44.Zestcott CA, Blair IV, Stone J. Examining the Presence, Consequences, and Reduction of Implicit Bias in Health Care: A Narrative Review. Group Process Intergroup Relat. 2016;19(4):528–542. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45.Hall WJ, Chapman MV, Lee KM, et al. Implicit Racial/Ethnic Bias Among Health Care Professionals and Its Influence on Health Care Outcomes: A Systematic Review. Am J Public Health. 2015;105(12):e60–76. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46.Betancourt JR, Corbett J, Bondaryk MR. Addressing disparities and achieving equity: cultural competence, ethics, and health-care transformation. Chest. 2014;145(1):143–148. [DOI] [PubMed] [Google Scholar]
- 47.Cooper LA and Powe RN. 2004. Disparities in Patient Experiences, Health Care Processes, and Outcomes: The Role of Patient-Provider Racial, Ethnic, and Language Concordance. http://www.commonwealthfund.org/~/media/files/publications/fund-report/2004/jul/disparities-in-patient-experiences--health-care-processes--and-outcomes--the-role-of-patient-provide/cooper_disparities_in_patient_experiences_753-pdf.pdf. Accessed February 28, 2018.
- 48.Dale HE, Polivka BJ, Chaudry RV, Simmonds GC. What young African American women want in a health care provider. Qual Health Res. 2010;20(11):1484–1490. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 49.Volandes AE, Paasche-Orlow M, Gillick MR, et al. Health literacy not race predicts end-of-life care preferences. J Palliat Med. 2008;11(5):754–762. [DOI] [PubMed] [Google Scholar]
- 50.Mann E, Haastert B, Bohmdorfer B, et al. Prevalence and associations of potentially inappropriate prescriptions in Austrian nursing home residents: secondary analysis of a cross-sectional study. Wien Klin Wochenschr. 2013;125(7–8):180–188. [DOI] [PubMed] [Google Scholar]
- 51.Oliveira MG, Amorim WW, de Jesus SR, Heine JM, Coqueiro HL, Passos LC. A comparison of the Beers and STOPP criteria for identifying the use of potentially inappropriate medications among elderly patients in primary care. J Eval Clin Pract. 2015;21(2):320–325. [DOI] [PubMed] [Google Scholar]
- 52.O’Mahony D, Gallagher P, Ryan C, et al. STOPP & START criteria: A new approach to detecting potentially inappropriate prescribing in old age. European Geriatric Medicine.1(1):45–51. [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.