Purpose Results of a study to determine if disparities in drug pricing, pharmacy services, and community pharmacy access exist in a Tennessee county with a predominantly minority population are reported.
Methods A cross-sectional survey of community pharmacies in Shelby County, a jurisdiction with a total population more than 60% composed of racial and ethnic minority groups, was conducted. Data collection included “out-of-pocket” (i.e., cash purchase) prices for generic levothyroxine, methylphenidate, and hydrocodone–acetaminophen; pharmacy hours of operation; availability of selected pharmacy services; and ZIP code–level data on demographics and crime risk. Analysis of variance, chi-square testing, correlational analysis, and data mapping were performed.
Results Survey data were obtained from 90 pharmacies in 25 of the county’s 33 residential ZIP code areas. Areas with fewer pharmacies per 10,000 residents tended to have a higher percentage of minority residents (p = 0.031). Methylphenidate pricing was typically lower in areas with lower employment rates (p = 0.027). Availability of home medication delivery service correlated with income level (p = 0.015), employment rate (p = 0.022), and crime risk (p = 0.014).
Conclusion A survey of community pharmacies in Shelby County, Tennessee, found that areas with a high percentage of minority residents had lower pharmacy density than areas with a high percentage of white residents. Pharmacies located in communities with low average income levels, low employment rates, and high scores for personal crime risk were less likely to offer home medication delivery services.
- community pharmacy
- healthcare access
- medication costs
- pharmacy services
- racial disparities
- socioeconomic disparities
In Shelby County, Tennessee, areas with predominantly minority populations generally have fewer community pharmacies per 10,000 residents than areas with majority white populations, suggesting disparities in pharmacy access.
Data from a survey of 90 Shelby County community pharmacies indicated a trend of lower prices for generic methylphenidate in areas of high unemployment; pricing of generic levothyroxine or hydrocodone–acetaminophen did not appear to be affected by socioeconomic variables.
Surveyed pharmacies in low-income areas and areas of high unemployment and personal crime risk were unlikely to offer home medication delivery.
In the United States, there are over 1 billion visits to physicians per year, and nearly 75% of these visits result in receipt of a medication prescription.1 While the use of medications to decrease morbidity and mortality and improve quality of life continues to increase, healthcare and medication access disparities exist and involve various chronic disease states and areas of healthcare, including asthma, cardiovascular disease, diabetes mellitus, human immunodeficiency virus infection, mental health, oncology, and pain control.2,3 Healthcare disparities may be based on a number of factors, such as race, ethnicity, sex, geographic location, and health literacy.4
Socioeconomic factors such as insurance status also play a key role in health disparities and medication access. For example, individuals without prescription drug coverage are more likely to forgo needed medication therapy and cite cost as a motivating factor.5 In 2012, uninsured patients 18–64 years of age were 4 times more likely than their insured counterparts to not receive medications due to financial reasons2; this may result in medication nonadherence and negative health outcomes.5 Outcomes associated with cost-related nonadherence include poorer health and increased hospitalizations.6
The proximity of an individual’s residence to a community pharmacy and the possible lack of services and drug supplies offered by neighborhood community pharmacies are important factors in medication access. A study of the availability of pharmacies in a Midwestern city (Chicago, Illinois) indicated that “segregated minority neighborhoods” (defined as those in which less than 50% of the population was non-Hispanic white) had fewer community pharmacies.7 During the period 2000–12, these “pharmacy deserts,” located in predominantly black or Hispanic, socioeconomically disadvantaged, and medically underserved areas, experienced less growth in numbers of pharmacies than predominantly white neighborhoods even after adjustments for population density and growth.7 These data are consistent with findings in other studies examining disparities in medication access. For example, not only are black patients, Hispanic patients, and patients of low socioeconomic status less likely to be prescribed opioids in emergency departments than are white patients with equivalent pain, but those living in predominantly nonwhite areas are less likely to have adequate access to opioid analgesics due to insufficient stocking of supplies by pharmacies in these neighborhoods.8–11 In another example, Amstislavski and colleagues12 noted socioeconomic disparities in medication access in a study of New York City pharmacies. Specifically, they found that pharmacies in communities with higher poverty levels were more likely than those in areas with lower poverty levels to be out of stock of common prescription medications used to treat health conditions such as high cholesterol, diabetes, hypertension, and depression.
While ostensibly an artificial and surrogate boundary, the border of the ZIP code area in which an individual resides may influence access to community pharmacies and, in turn, medications and services, thereby creating potential disparities. One study in Wayne County, Michigan, which has a population that is 54.4% white and 40.1% African-American, with the remaining 5.5% consisting of various nonblack minority groups, evaluated out-of-pocket costs for medications and the availability of various pharmacy services in relation to resident characteristics within 63 ZIP code areas.13 The researchers concluded that individuals belonging to racial and ethnic minorities and those of lower socioeconomic status may be at a disadvantage in terms of access to medications and pharmacy services. For instance, in ZIP code areas characterized by lower average annual household incomes, costs for generic medications such as levothyroxine were significantly higher (mean ± S.D. costs for levothyroxine were 11.0 ± 2.10 in the lowest income quartile, 10.79 ± 2.10 in the second-lowest quartile, 9.25 ± 1.87 in the second-highest quartile, and 8.89 ± 2.53 in the highest quartile; p = 0.02), fewer pharmacies offered discount generic drug programs and immunizations, and pharmacy hours of operation were fewer on average.13 On the other hand, residents of ZIP code areas with a larger white population had greater access to pharmacy services, including discount generic drug programs and immunizations.
Few studies in the literature on pharmacy access factors have focused on geographic locations with resident populations predominantly composed of racial or ethnic minority groups and the possible impact that increased demographic representation might have on pharmacy access. The study described here, conducted in Shelby County, Tennessee (which includes the city of Memphis), sought to address this issue. Although economically and racially diverse, with minority groups constituting more than 60% of the population, Shelby County faces ongoing issues regarding racial disparities in some aspects of healthcare and healthcare outcomes. For example, there is a higher rate of diabetes-related leg amputations in black residents versus nonblack residents, and cancer-related mortality is higher among minority residents versus white residents.14 However, potential disparities in community pharmacy access, medication pricing, and access to pharmacy services have not been examined. Determination of disparities is critical to development of a better understanding of the barriers to equitable medication and pharmacy access; such understanding is needed in order to design and implement interventions to reduce identified disparities in Shelby County. Therefore, the purpose of the survey-based study was to determine if there are racial and socioeconomic disparities in community pharmacy access, drug pricing, and services in Shelby County.
Background. Shelby County was selected as the study site due to its mid-South location and investigator familiarity with the county. Moreover, because previous studies largely focused on regions with a majority white population, Shelby County was selected due to its racial and economic diversity. It has a population more than 60% composed of minority groups, with a median annual household income of $46,250 and an unemployment rate of 6.7% (as of May 2015).15,16 The study was approved by the institutional review board of the University of Tennessee Health Science Center.
Survey design. A cross-sectional survey of community pharmacies located in Shelby County (inclusive of the city of Memphis) was conducted. The aim of the study was to evaluate measures of community pharmacy access and drug pricing in relation to resident characteristics. Of the 78 ZIP code areas in Shelby County, 33 include a residential population on which U.S. census data are available; these areas were the focus of the study. Information gathered in the 2010 national census and in the U.S. Census Bureau’s American Community Survey17 were used to collect the following ZIP code–level data (these were the independent variables for study): population total, median age, percentage of white residents, percentage of female residents, median household income, percentage of residents with incomes below the federal poverty line, percentage of residents who graduated from high school, percentage of residents who graduated from college (i.e., with a bachelor’s degree), percentage of residents who were employed, and percentage of residents who were uninsured. Data available from the federal Health Resources and Services Administration (HRSA)18,19 were used to identify primary care shortage areas, mental health services shortage areas, and medically underserved areas within Shelby County. The HRSA data were used to create the following study variables, which were analyzed to characterize pharmacy access by ZIP code area, percentage of pharmacies in mental health services shortage areas, percentage of pharmacies in primary care shortage areas, and percentage of pharmacies in medically underserved areas. Areas designated as having a primary care shortage are those with a physician-to-population ratio of more than 1:3,500; mental health service shortage areas are those for which there are at least 30,000 residents for every psychiatrist.18 Designation of medically underserved areas is based on a calculation involving the ratio of primary care physicians to residents, the infant mortality rate, the percentage of residents below the poverty line, and the percentage of residents who are 65 years of age or older.19 ZIP code–level crime risk scores20 were also considered as independent variables. For each area, 3 scores were evaluated: a total crime risk score (national average = 100, with higher scores indicating greater risk) representing the combined risks of rape, murder, assault, robbery, burglary, larceny, and vehicle theft; a personal crime risk score representing the combined risks of rape, murder, assault, and robbery; and a property crime risk score representing the combined risks of burglary, larceny, and vehicle theft.
During February and March 2015, a telephone survey targeting pharmacies located in the identified residential ZIP codes in Shelby County was conducted. Up to 3 attempts to contact each pharmacy were made. To be eligible for study participation, a pharmacy had to be verified as a retail community pharmacy (defined as an independent, supermarket, chain, or mass merchandiser pharmacy21). Of the 157 pharmacies initially identified, 53 were excluded from survey participation because they were not community pharmacies or could not be verified as community pharmacies; the remaining 104 were verified as retail community pharmacies eligible for study inclusion based on oral confirmation by the pharmacy manager or a designee. Publicly available data on the 104 verified community pharmacies were used to determine pharmacy density, defined as the number of community pharmacies per 10,000 residents per ZIP code.7
A standardized survey questionnaire was administered to the pharmacist manager, the pharmacist-in-charge, or a designee at each participating pharmacy; questionnaire administration took about 5 minutes. The questionnaire included open-ended questions to ascertain drug pricing and hours of operation and yes/no questions to determine if a pharmacy offered the services of interest. The following data were collected and entered into the study database (the accuracy of data entry was verified by 2 investigators): “out-of-pocket” (i.e., cash purchase) prices of 30-tablet supplies of generic levothyroxine sodium 50 µg, generic methylphenidate hydrochloride extended-release (ER) 20 mg, and generic hydrocodone bitartrate 5 mg–acetaminophen 325 mg); pharmacy hours of operation from Monday through Friday and on Saturdays and Sundays (hours of operation were considered pharmacy access variables for study purposes); and availability of services offered by many pharmacies (a generic drug program, home medication delivery, influenza immunizations, other immunizations, and medication therapy management [MTM]). Levothyroxine, methylphenidate, and hydrocodone–acetaminophen were selected for examination of disparities in pricing because they are frequently prescribed, are included in the list of the top 200 prescription drugs dispensed in the United States, and are used in the treatment and management of common disease states.22 As a drug often prescribed to children and adolescents, methylphenidate ER was of particular interest because possible disparities in its pricing have not been previously assessed in the literature. Since methylphenidate is commonly used in the pediatric population to treat attention-deficit/hyperactivity disorder, its use pertains to an understudied area of potential health disparities: pediatric mental health.
Data analysis. Data analysis was conducted using IBM SPSS Statistics 22.0 (IBM Corporation, Armonk, NY). Demographic data, crime scores, and pharmacy density data were aggregated and expressed as mean ± S.D. values. Pearson’s correlations (r values) were calculated for the ZIP code–delineated variables. Due to non-normal distribution, data on the percentages of pharmacies in mental health services shortage areas, primary care shortage areas, and medically underserved areas were summarized as medians and modes with interquartile ranges (IQRs).
The analysis approach was based on the plan used by Erickson and Workman,13 in which pharmacy data (medication out-of-pocket prices, pharmacy services, and pharmacy hours of operation) were aggregated at the ZIP code level. Analysis of variance (ANOVA) was performed to assess associations between ZIP code area characteristics (with data segmented into quartiles) and pharmacy variables. Likewise in our study, data from survey participants were aggregated at the ZIP code level: (1) out-of-pocket prices of the 3 drugs of interest (hydrocodone–acetaminophen, levothyroxine, and methylphenidate) and hours of operation were summarized as means, and (2) data on the availability of each pharmacy service of interest were summarized as percentages based on the number of pharmacies offering the service within each ZIP code area. The aggregated data were then used to calculate mean ± S.D. values for the dependent variables (out-of-pocket drug prices, pharmacy hours of operation, and percentages of pharmacies offering selected services).
ANOVA was conducted to determine associations between ZIP code–delineated population characteristics (with the exception of mental health shortage area, primary care shortage area, and medically underserved area) and pharmacy access variables (out-of-pocket drug prices, pharmacy hours of operation, and pharmacy services offered). ANOVA was also used to determine associations between pharmacy density and pharmacy access and pricing variables. Post hoc comparisons were performed after ANOVA using the Bonferroni correction. For ANOVA, data on local population characteristics and pharmacy density were converted to quartile groupings. Chi-square analysis was performed to examine associations of pharmacy density with resident characteristics and crime risk. Due to non-normal distribution of the independent variables, Spearman correlations were conducted to assess the relationship of the pharmacy variables to the percentages of pharmacies in mental health shortage areas, primary care shortage areas, and medically underserved areas. The a priori significance level was 0.05.
Select demographic and pharmacy characteristics were mapped using ArcGIS Online, version 10.2.2 (Esri, Redlands, CA); all ZIP code areas completely within the boundaries of Shelby County for which the data of interest could be obtained were included. The base map layer was derived from the ArcGIS Web Map collection (Esri), and a Mercator Auxiliary Sphere projection was used. Data collected in the survey and demographic information from public sources were then combined with geographic data to populate the mapped ZIP code areas.
Of 104 eligible community pharmacies, 90 (86.5%) participated in the survey. The 14 pharmacies that declined to participate did so because company policy precluded participation. The participating pharmacies were located in 25 of the county’s 33 ZIP code areas. Quartile-based groupings of ZIP code–level data on Shelby County resident characteristics, crime risk, and pharmacy density are presented in Table 1; data on those characteristics and pharmacy-reported data, aggregated as mean ± S.D. values at the ZIP code level, are presented in Table 2. The median percentage of pharmacies in mental health shortage areas per ZIP code was 100% (mode, 100%; IQR, 100%), the median percentage of pharmacies in primary care shortage areas was 80% (mode, 100%; IQR, 100%), and the median percentage of pharmacies in medically underserved areas was 0% (mode, 0%; IQR, 66.5%).
Pharmacy-reported data on out-of-pocket prices for the study drugs, stratified by ZIP code area variable and grouped by quartile, are displayed in Table 3. Select variables were significantly associated with pricing of methylphenidate and levothyroxine (but not hydrocodone–acetaminophen). Specifically, the post hoc analysis found a significant difference in methylphenidate pricing based on employment rate (F = 3.821, df = 19, p = 0.027); prices were lower in ZIP code areas with the lowest employment rates (quartile 1 mean, $74.19), as compared with areas with higher employment rates (quartile 3 mean, $165.71; p = 0.03). Pricing of levothyroxine differed significantly according to pharmacy density (F = 3.826, df = 20, p = 0.026); pricing was higher in ZIP code areas with the fewest pharmacies per 10,000 residents (quartile 1 mean, $16.37) relative to areas with higher pharmacy density (quartile 2 mean, $10.27; p = 0.023). Pricing of methylphenidate and levothyroxine did not differ significantly in relation to any other evaluated ZIP code area variable.
Similarly, few significant associations were noted with regard to the availability of pharmacy services (Table 4 presents a quartile distribution of ZIP code–stratified data on the percentages of pharmacies offering the 5 services of interest). Among those services, only the availability of home medication delivery was found to be significantly associated with one or more of the ZIP code area variables. A statistically significant difference was found based on income (F = 4.383, df = 21, p = 0.015; Figure 1). The post hoc analysis indicated significant differences in the percentages of pharmacies offering home delivery between ZIP code areas with the lowest income levels (quartile 1 mean, 0%) and higher-income areas (quartile 2 mean, 45.9% [p = 0.048]; quartile 3 mean, 44.9% [p = 0.044]). Differences in home medication delivery on the basis of employment rate were also found (F = 3.97, df = 21, p = 0.022; Figure 1); ZIP code areas with the lowest rates of employment had a lower percentage of pharmacies offering the service (quartile 1 mean, 0%) than pharmacies in areas with higher employment rates (quartile 2 mean, 55.0%; p = 0.15). The availability of home medication delivery was also associated with personal crime risk score (F = 4.442, df = 21, p = 0.014; Figure 1), with delivery service offered by a higher percentage of pharmacies in lower-risk ZIP code areas (quartile 2 mean, 48.1%) versus the highest-risk areas (quartile 4 mean, 0%; p = 0.034).
Pharmacy access was examined in relation to hours of operation per week and pharmacy density (Table 4 presents pharmacy-reported data on mean hours of operation stratified by ZIP code variable quartiles). Significant differences based on ZIP code area were found in Monday–Friday hours of operation (F = 6.684, df = 21, p = 0.002), Saturday hours of operation (F = 5.52, df = 21, p = 0.006), and Sunday hours of operation (F = 7.792, df = 21, p = 0.001). Regarding Monday–Friday hours of operation, pharmacies in the least-populated areas typically reported fewer hours per week (quartile 1 mean, 50 hours) than those in the most-populated ZIP codes (quartile 3 mean, 68.1 hours [p = 0.008]; quartile 4 mean, 70.4 hours [p = 0.004]). Likewise for Saturday hours of operation, pharmacies in the least-populated ZIP code areas tended to report fewer hours of operation (quartile 1 mean, 4.9 hours) than pharmacies in the most-populated areas (quartile 3 mean, 10.1 hours [p = 0.016]; quartile 4 mean, 10.6 hours [p = 0.010]). In the post hoc analysis of Sunday hours of operation, pharmacies in the least-populated ZIP code areas averaged fewer hours of operation (quartile 1 mean, 1.9 hours) than pharmacies in more populated areas (quartile 2 mean, 7.5 hours [p = 0.015]; quartile 3 mean, 8.0 hours [p = 0.005]; quartile 4 mean, 9.1 hours [p = 0.002]).
In the chi-square analysis of pharmacy density, only 1 variable (percentage of white residents) was statistically significant (χ2 = 18.353, p = 0.031), indicating that ZIP code areas with more pharmacies per 10,000 residents had a higher percentage of white residents (and thus a lower percentage of minority residents). To better illustrate this finding, Figure 1 maps data on pharmacy density and minority population percentage per ZIP code area.
No significant associations were found when data on the percentages of pharmacies in mental health service shortage areas, primary care shortage areas, and medically underserved areas were analyzed in relation to the dependent variables (medication out-of-pocket prices, pharmacy services offered, and pharmacy access factors [hours of operation and pharmacy density]). Significant correlations (p < 0.01) of independent variables (those pertaining to race, income, employment, poverty, insurance, education, and crime risk) were found, with predominantly white ZIP code areas more likely than predominantly minority areas to have higher proportions of insured residents, better socioeconomic indicators, higher levels of education, and lower crime risk scores.
Previous studies have indicated that neighborhood of residence may affect pharmacy access, particularly among minority and low-income populations.7,12,13 However, for the most part, these studies focused on regions in which whites constituted the largest racial group. Our study sought to examine pharmacy access issues in an area where minority groups constitute a majority of the population.
The study findings regarding pharmacy density and race suggest that the risk of poor pharmacy access is increased in predominantly minority ZIP code areas; this is problematic given that major chronic disease states such as hypertension and diabetes are more prevalent among minority populations, particularly African Americans, as compared with whites.23 As medication therapy is the typical treatment modality for most chronic conditions, limited community pharmacy access and, in turn, limited medication access jeopardize treatment efforts and may result in poorer health outcomes in minority groups. Strategies proposed by Qato et al.7 to increase the number of community pharmacies (and thus pharmacy access) include expansion of public policies such as economic incentives to prompt pharmacy openings in these communities. Another option may be to create partnerships with health clinics in predominantly minority neighborhoods wherein pharmacies operate on clinic premises. Mobile pharmacies (analogous to mobile blood donation centers) may also be a strategy by which greater access can be provided.
In contrast to racial disparities in pharmacy access (measured in our study through evaluation of pharmacy density), economic forces such as consumer demand and costs of operation seem the more likely factors driving pharmacy access, as measured by pharmacy hours of operation. We found that pharmacies in the least-populated ZIP code areas were, on average, open fewer hours per week than pharmacies in more heavily populated areas. Perhaps due to their smaller consumer base and subsequently reduced consumer demand, pharmacies in less populated ZIP code areas may limit their hours per week to decrease operational costs.
Economic factors at both the individual level (e.g., employment status) and the pharmacy level (e.g., number of competing pharmacies per ZIP code) were also noted to affect out-of-pocket drug prices. Generic methylphenidate was generally priced lower in areas of greater unemployment. This finding contrasts with results of prior studies indicating socioeconomic disparities in the pricing of a range of commonly used generic drugs.12,13 Our finding provided evidence of favorable pricing of methylphenidate for individuals of low socioeconomic status in Shelby County, suggesting that the medication was more affordable for these residents. Pharmacy-reported pricing of methylphenidate did not appear to be influenced by any other evaluated ZIP code area variable or by pharmacy density, nor were any pricing differences found with regard to hydrocodone–acetaminophen.
The third medication of interest in the study was generic levothyroxine. As previously mentioned, a study conducted by Erickson and Workman13 in Wayne County, Michigan, found that ZIP codes with lower household income were associated with significantly higher out-of-pocket prices for generic levothyroxine. In contrast, our study found that prices for generic levothyroxine did not differ significantly on the basis of income or any other racial or socioeconomic variable. Higher prices for levothyroxine were noted in ZIP codes with fewer pharmacies per 10,000 residents, suggesting that when pharmacies have less competition (and residents have fewer purchasing options) the prices for select medications may be affected.
The pharmacy services findings of our study also differed from those of Erickson and Workman,13 who found that pharmacies in lower-income ZIP code areas was associated with decreased availability of generic drug programs as well as influenza and other immunization programs; they also found no significant association between income and home medication delivery. In our analysis of the availability of pharmacy services in relation to multiple ZIP code area variables, no significant socioeconomic or racial differences were found with regard to generic drug programs, MTM services, and influenza and other immunizations. However, pharmacies in ZIP code areas characterized by lower median household incomes and lower employment rates were less likely to offer home medication delivery. Thus, socioeconomic disparities influence access to home medication delivery. Such access may be critical to patients who experience transportation barriers. Additionally, availability of home medication delivery is more limited in ZIP code areas with higher scores for personal crime risk. We speculate that pharmacies in these areas may not offer the service due to concerns over employee safety.
Although home medication delivery may not be as commonplace as other pharmacy services, it was offered by more than one third of pharmacies participating in our study. This finding provides evidence of consumer demand for the service in Shelby County and suggests the importance of reducing barriers to home medication delivery. Therefore, one suggestion to improve access to home medication delivery is to explore the possibility of pharmacies in low-income and low-employment areas partnering and contracting with community health workers. Community health workers are trained to participate in health education and outreach and are embedded in local neighborhoods. They have relationships and familiarity with the communities they serve that may allow them to navigate these areas with greater ease than pharmacy employees. Contracts with community health workers may also involve a lower cost to pharmacies relative to hiring employees specifically to perform medication deliveries. Although such partnerships face challenges (e.g., adhering to Health Insurance Portability and Accountability Act rules regarding patient health information), they may also help to increase community access to pharmacy home medication delivery services.
There were limitations to the study. Individuals may travel across ZIP code boundaries to procure goods and services on a regular basis.13 However, prior studies have successfully used ZIP code–related characteristics or similar geographic variables to examine disparities in drug pricing, pharmacy services availability, and community pharmacy access.7,12,13 Additionally, ZIP code–level data are readily available from the U.S. Census Bureau. Therefore, ZIP code of residence was selected as an appropriate indicator of geographic access to community pharmacies. Another limitation was that the study focused only on one U.S. jurisdiction, Shelby County, in which the largest racial or ethnic group is a minority group (African Americans). Therefore, the results may have limited generalizability, reinforcing the need for individual counties and cities to examine and address the specific pharmacy access disparities that may exist within a particular geographic region.
A survey of community pharmacies in Shelby County, Tennessee, found that areas with a high percentage of minority residents had lower pharmacy density than areas with a high percentage of white residents. Pharmacies located in communities with low average income levels, low employment rates, and high scores for personal crime risk were less likely to offer home medication delivery services.
The research project was supported by funding from the National Association of Chain Drug Stores Foundation. The authors have declared no potential conflicts of interest.
- Copyright © 2017 by the American Society of Health-System Pharmacists, Inc. All rights reserved.