Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2021 Sep 1.
Published in final edited form as: Rural Sociol. 2020 Jun 11;86(1):26–49. doi: 10.1111/ruso.12338

Income inequality and opioid prescribing rates: Exploring rural/urban differences in pathways via residential stability and social isolation

Tse-Chuan Yang 1, Seulki Kim 2, Carla Shoff 3
PMCID: PMC8045985  NIHMSID: NIHMS1611550  PMID: 33867589

Abstract

While opioid prescribing rates have drawn researchers’ attention, little is known about the mechanisms through which income inequality affects opioid prescribing rates and even less focuses on whether there is a rural/urban difference in mediating pathways. Applying mediation analysis techniques to a unique ZIP code level dataset from several sources maintained by the Centers for Medicare and Medicaid Services, we explicitly examine two mechanisms through residential stability and social isolation by rural/urban status and find that (1) income inequality is not directly related to opioid prescribing rates, but it exerts its influence on opioid prescribing via poor residential stability and elevated social isolation; (2) social isolation accounts for two-thirds of the mediating effect of income inequality on opioid prescribing rates among urban ZIP codes, but the proportion halves among rural ZIP codes; (3) residential stability plays a larger role in understanding how income inequality matters in rural than in urban ZIP codes; and (4) beneficiary characteristics only matter in urban ZIP codes. These findings offer nuanced insight into how income inequality affects opioid prescribing rates and suggests that the determinants of opioid prescribing rates vary by rural/urban status. Future research may benefit from identifying place-specific factors for opioid prescribing rates.

Keywords: rural/urban, opioids, income inequality, residential stability, social isolation, mediation analysis

INTRODUCTION

Prescription opioids play a significant role in the growing opioid epidemic in the United States (US) since the 1980s (Kolodny et al. 2015). In 2017, it is estimated that 2.1 million people aged 12 or older suffer from opioid use disorder (OUD) (Substance Abuse and Mental Health Services Administration 2018) and most of them begin with medical use of opioid pain relievers. A recent study (Scholl et al. 2019) reports that more than two-thirds of drug overdose deaths are related to opioids and among the opioid-related deaths, over 35 percent are the result of prescription opioids. That is, out of the 70,000 drug overdose deaths in 2017, more than 17,000 deaths can be attributed to prescription opioids (Scholl et al. 2019).

Opioid prescribing is not evenly distributed across social groups. It has been suggested that the opioid prescribing rate is higher among Medicare beneficiaries than individuals enrolled in private insurance. For example, 26 percent of aged Medicare beneficiaries (65+) use prescription opioids, in contrast to 14 percent of commercial beneficiaries (Jeffery et al. 2018). This difference could not be explained by individual characteristics (e.g., age, sex, and race/ethnicity). Similarly, in Minnesota, a Medicare beneficiary, on average, receives almost 1.5 opioid prescriptions, compared with only 0.37 prescription for an individual with private insurance (Li, Bell, and Chollet 2018). Coupled with the growing trend in the opioid prescribing rate among Medicare beneficiaries (Kuo et al. 2016), individuals enrolled in Medicare are more susceptible to OUD or other negative consequences of opioid misuse than their counterparts with other sources of coverage, including Medicaid (Li et al. 2018).

Moreover, types of residence affect how health care providers prescribe opioids. Patients living in rural areas are more likely to receive an opioid prescription than those residing in urban areas (Keyes et al. 2014; Monnat and Rigg 2016; Rigg and Monnat 2015). While prescribing practices across different types of residence have been improved, the opioid prescribing rate of rural areas (i.e., 9 opioid prescriptions per 100 patient-weeks) almost doubles that of urban areas (i.e., 5 opioid prescriptions per 100 patient-weeks) (García et al. 2019). Similar findings can be applied to other measures of prescription opioids, such as average morphine milligram equivalents (MME) prescribed per person and high-dose opioid prescriptions (Heins et al. 2018; Lund et al. 2019). These findings indicate that it is easier to obtain prescription opioids for rural residents than for urban dwellers, which may lead to rural/urban disparities in adverse outcomes related to prescription opioids (Monnat and Rigg 2016; Rigg and Monnat 2015).

Though the significance of prescription opioids has drawn researchers’ attention (Chen et al. 2019; Keyes et al. 2014; Monnat and Rigg 2016), there are two gaps in the literature that this study aims to fill. First, the relationship between income inequality and the opioid prescribing rate is equivocal, even though it is expected that high levels of income inequality are associated with unequal distributions of resources, which leads to poor access to quality medical care particularly among the impoverished and results in higher rates of opioid prescribing (Webster et al. 2009). Some scholars offer support to this claim (Webster et al. 2009), but others find a null relationship (McDonald, Carlson, and Izrael 2012). Second, little research has investigated the potential mechanisms through which income inequality may affect the opioid prescribing rate and even less has explored whether the mechanisms can be applied to both rural and urban areas.

This study goes beyond prior literature by applying mediation analysis techniques to a unique ZIP-code level dataset aggregated from the 2017 Medicare Part D Prescription Drug Event data, along with the 2017 Centers for Medicare and Medicaid Services Master Beneficiary Summary File Base and the 2010–2015 American Community Survey, to address the two gaps. Using data collected from different time periods, we argue that residential stability and social isolation may serve as mediators linking income inequality to the opioid prescribing rate and examine whether these mechanisms differ between rural and urban ZIP codes.

BACKGROUND AND HYPOTHESES

Income Inequality and Opioid Prescribing Rates

Several reasons explain why income inequality may increase opioid prescriptions. First, drawing from the stress coping theory (Lazarus 1993), being exposed to high levels of income inequality could serve as a chronic stressor in daily life, which has adverse impacts on a range of health and social outcomes, such as individual immune systems, mental health, interpersonal social skills, and other wellbeing (Cohen et al. 2007; Fernald and Gunnar 2009; Pelzer, Schaffrath, and Vernaleken 2014). The compromised social and health outcomes increase the demand for opioids because the use of opioids may help individuals to cope with stress by creating temporary euphoria, sense of power, self-confidence, and energy (Wise and Koob 2014). The temporary, yet intense, feeling of pleasure shelters individuals from the daily stress or negative emotions and fills a void in one’s life. Consequently, individuals may become dependent on opioid treatment to overcome daily stress, which may reinforce the demand for opioids.

Second, high levels of income inequality indicate that the distribution of resources across different social dimensions is unequal (Kawachi 2000). When most resources are concentrated on few individuals or social groups in an area, the competition for limited resources among others becomes tenser, compared to societies where resources are shared by most members. The stiff competition, thus, requires individuals to perform better than others, both mentally and physically. Otherwise, the chance of getting resources will be thin. This scenario may be particularly obvious in labor-intensive industries. The desire to improve performance or focus at work may initiate the use of opioid pain relievers or continue to use prescribed opioid drugs that effectively relieve symptoms (NIDA 2018; Weiss et al. 2014). Doing so helps individuals to survive in areas with high levels of income inequality. Should individuals depend on opioid drugs to maintain their performance, they may create a high demand for opioid prescriptions in areas with high levels of income inequality, indicating a positive relationship between inequality and opioid prescription rates.

Finally, related to the unequal distribution of resources discussed previously, areas with high income inequality are less likely to invest in social services (e.g., library) or public infrastructure (Kawachi 2000) and the marginalized populations will not have access to the public good and quality medical care (Webster et al. 2009). As a consequence, a strong sense of relative deprivation will be developed among marginalized populations and they are more likely to have low self-esteem or efficacy (Osborne, Sibley, and Sengupta 2015), which may undermine their social interaction skills and increases social distance from others (Wilkinson 1997). The lack of self-esteem or confidence may lead to mental illnesses, which increases the odds of using prescribed opioid drugs to alleviate psychiatric symptoms (NIDA 2018) or elevates the risk of drug use (Park and Yang 2017; Subramanian and Kawachi 2004). The lack of access to quality medical care among the poor who live in areas with high levels of income inequality also indicates that opioids may be used in lieu of the unavailability of other appropriate treatments.

The discussion above suggests that living in areas featured with high levels of income inequality may increase the demand for opioid treatment. Hence, we propose the first hypothesis (H1): High levels of income inequality are associated with high opioid prescribing rates, even after adjusting for other potential confounders.

In addition to the direct impact of income inequality on the opioid prescribing rates, we provide a research framework where residential stability and social isolation serve as the mediators. We discuss the reasons why at least part of the impact of income inequality on the opioid prescribing rates may be mediated by these two factors.

Residential Stability as a Mediator

Conceptually, residential stability refers to community dwellers’ effort and time to connect with other residents (Sampson, Morenoff, and Earls 1999) and it is often assessed with home ownership and long-term residence in the same area. To understand why residential stability may serve as the mediator in this study, it is imperative to first clarify the relationship between income inequality and residential stability. Areas with high levels of income inequality are often characterized with high unemployment, high poverty, and low levels of social control (Hipp 2007; Kawachi et al. 1997; Wilkinson 1997). These undesirable socioeconomic conditions are likely to be translated into residential dissatisfaction, which may serve as a driving force for a move and hence decrease residential stability (South and Deane 1993). In addition, income inequality may undermine residential stability because the opportunity structure of an area is likely to be compromised by high levels of income inequality. Moving somewhere else becomes a viable strategy to maximize the return of one’s investment in human capital or to increase life chances by escaping from poverty and unemployment (Borjas 1987; Frey et al. 1996). As a result, high levels of income inequality should be associated with low levels of residential stability.

Furthermore, residential stability has been identified as a critical component of building social networks, social ties, or social capital (Kawachi et al. 1997; Sampson 1991; Warner and Rountree 1997). These factors facilitate the interactions among residents and could serve as a source of self-esteem and mutual respect, which improves one’s mental health (Kawachi and Berkman 2000) and decreases the odds of having prescription opioids. In addition, high levels of residential stability are likely to build a close-knit personal safety net that may offer instrumental support when individuals are in need (Turney and Harknett 2010). The beneficial effect of living in a stable area may lower the demand for opioids and hence, decrease the opioid prescribing rates. Several studies have reported that living in a stable neighborhood decreases the odds of prescription opioids misuse (Ford and Rigg 2015; Monnat and Rigg 2016) or illicit drug use (Kipke, Weiss, and Wong 2007); however, it remains underexplored if residential stability is negatively associated with opioid prescribing rates.

High levels of residential stability may also affect how health care providers practice. For example, social connections are stronger in stable than unstable neighborhoods and strong connections have been found to facilitate nonmedical opioid use among residents (Jonas et al. 2012). As such, physicians may be more cautious when prescribing in stable neighborhoods than in unstable neighborhoods. Moreover, the stigma associated with nonmedical opioid use may be exacerbated in stable neighborhoods given the strong social ties (Corrigan and Nieweglowski 2018), which may lead health care providers to minimize opioid prescribing.

While the aforementioned relationships among income inequality, residential stability, and opioid prescribing rates seem intuitive, this pathway fills two knowledge gaps. On the one hand, to our knowledge, little research has explored whether residential stability is associated with opioid prescribing rates and even less has investigated the possible mediating process. Specifically, it is suggested that a frequent turnover may create great vulnerability to drug use and abuse in an area because an unstable workforce undermines economic infrastructure and moving serves as a selection process that traps individuals with low human capital (Keyes et al. 2014). However, most previous studies have focused on the supply-side factors (Han et al. 2012; Webster et al. 2009; Wright et al. 2014), such as availability of health care services, and largely overlooked the variables that can create a higher demand for opioids (Dasgupta, Beletsky, and Ciccarone 2018). On the other hand, little attention has been paid to how income inequality may affect health providers’ behaviors. Even among the few studies that explored the relationship between income inequality and opioid prescribing (McDonald et al. 2012; Zhou, Yu, and Losby 2018), residential stability is absent in the analysis.

The extant literature builds the first pathway via residential stability and we propose the second hypothesis (H2): High levels of income inequality are associated with low levels of residential stability, and the decrease in residential stability further increases the opioid prescribing rates. As the relationship between income inequality and residential stability and the association between residential stability and opioid prescribing rates are both negative, the overall indirect effect should be positive, which not only follows the same direction of the direct impact of income inequality on the opioid prescribing rates, but also indicates a mediating mechanism.

Social Isolation as a Mediator

The second pathway linking income inequality and opioid prescribing rates is through social isolation. High levels of income inequality often create a social environment that offers limited life chances from both economic and social standpoints. Residents living in areas with high levels of income inequality are more likely to experience social isolation or exclusion than their counterparts living in equal neighborhoods (Scharf, Phillipson, and Smith 2004) as income inequality is strongly associated with relative deprivation (Nolan and Ive 2009). When income inequality increases in an area, residents, particularly marginalized populations, are less likely to interact with one another and even regular communications may not be sustained (Wilson 1987). The explanation is that the macro level income inequality may provoke an individual’s perceived unfairness and distrust (Kawachi et al. 1997), and both factors prevent individuals from having regular interactions with others. The lack of social interactions, which is known as social isolation, is reflected by the reduced participation in local organizations and contacts with friends (Rankin and Quane 2000). As a result, income inequality should be positively associated with social isolation.

How is social isolation related to opioid prescribing rates? As a recent study points out (Dasgupta et al. 2018), prior research on the opioid epidemic focuses on the supply dimension of this growing public health concern (i.e., health care system) and overlooks the demand dimension of the problem. For example, using in-depth interviews to understand the root causes of the opioid epidemic in an economically distressed community, a study finds that individuals addicted to opioids cite social isolation and hopelessness as reasons for drug use (McLean 2016). Similarly, an ecological study concludes that social isolation contributes to opioid use in the US (Zoorob and Salemi 2017). That is, for those who are isolated from other social groups and/or regular contacts, opioids may offer them a refuge from psychological trauma, hopelessness, and loneliness. The euphoria and the sense of regaining control of one’s life created by opioids initiate the positive reinforcement process, i.e., seeking comfort, (Wise and Koob 2014) and may lead to a high demand for opioids.

In addition, social isolation is, at least partly, a function of aging in several ways. First, aging commonly increases ones’ psychosocial vulnerabilities, such as depression and feeling of helplessness (Cochran et al. 2017; Maree et al. 2016). These negative affections undermine physiological resilience (Hawkley and Cacioppo 2007) and increase social isolation among older adults. Second, from the life-course perspective, older adults are more likely to experience bereavement and loss of companionship (Cornwell and Waite 2009). These factors not only elevate social isolation, but also increase the likelihood of opioid use (Huhn et al. 2018). Finally, the aging process hinders individual physical health and several disease states (e.g., hearing loss and arthritis) may limit older adults’ daily activities, which in turn creates social isolation (Kobayashi, Cloutier-Fisher, and Roth 2009).

The aforementioned features associated with social isolation may also affect the way in which prescribers practice. The concern about disseminating opioids to others for nonmedical use may be lessened in neighborhoods with high levels of social isolation, which may be positively associated with opioid prescriptions. Furthermore, neighborhood social isolation may make individuals hypersensitive to stressors, which aggravates chronic pain (Hruschak and Cochran 2017). Opioid treatments may be one of the options for pain management and consequently, prescribing rates may increase.

Our second pathway via social isolation can be drawn from the discussion above. We hypothesize that (H3): high levels of income inequality are associated with high levels of social isolation, and the increasing social isolation is also related to the growing opioid prescribing rates. As the linkage between income inequality and social isolation and the association of social isolation with opioid prescribing rates are both positive, we expect that the indirect effect via social isolation should also be positive.

How Does Rural/Urban Status Matter?

Several scholars have emphasized the rural/urban differences in opioid use or misuse (Keyes et al. 2014; Monnat and Rigg 2016; Rigg and Monnat 2015) and we argue that the direct and indirect effect of income inequality on opioid prescribing rates should differ and we explain as follows. Prior research suggests that rural residents are more likely to be exposed to prescription opioids in contrast to their urban counterparts (García et al. 2019; Keyes et al. 2014). The relatively low socioeconomic profile among rural residents (e.g., low educational attainment) may decrease the awareness of the potential harm of prescription opioids. Furthermore, rural residents tend to use prescription opioids at an earlier age than urban dwellers due to the difference in industrial structure (Keyes et al. 2014; Monnat and Rigg 2016) and providers’ personal relationship with their patients affects opioid prescribing in rural areas (Click et al. 2018). Both these factors indicate that given the same level of income inequality, opioid prescribing rates should be higher (i.e., having a stronger impact) in rural than in urban areas. As a result, we hypothesize that (H4a) the association between income inequality and opioid prescribing rates is stronger in rural areas than in urban areas.

Regarding the indirect effects, little is known about whether the importance of each mediator differs between rural and urban areas. However, the literature of rural/urban differences in social ties and social capital sheds some light on this issue. Explicitly, it is found that personal social ties contain greater intensity and complexity and consist of more kinship and neighborhood solidarities in rural than in urban setting (Beggs, Haines, and Hurlbert 1996). In addition, compared with urban families, rural ones are more likely to offer social support to others (Hofferth and Iceland 1998). These advantages in rural areas have been documented to improve population health or reduce health risks (Yang, Jensen, and Haran 2011). Situating residential stability into this context, the protective effect of residential stability on opioid prescribing rates should be more profound (i.e., strengthened) in rural than in urban areas because of strong social capital and close-knit networks in rural areas. Similarly, because social capital and social ties could mitigate the detrimental impact of social isolation on opioid prescribing rates, social isolation should have a weaker impact in rural (i.e., mitigated more) than in an urban setting. That is, social isolation should be more important in urban than in rural areas. Extending the rationale to this study, we expect that (H4b) residential stability mediates the impact of income inequality on opioid prescribing rates more in rural areas than in urban areas and that (H4c) social isolation plays a larger role in urban areas than rural areas in mediating the impact of income inequality on opioid prescribing rates.

DATA AND METHODS

This study uses multiple data sources from the Centers for Medicare and Medicaid Services (CMS) and the 2010–2015 American Community Survey (ACS). All variables have been aggregated to the ZIP-code level. The data has been limited to the aged Medicare beneficiaries (65 years of age and older). To assure the privacy of Medicare beneficiaries, ZIP codes with fewer than 11 beneficiaries aged 65 or older utilizing the Part D program and ZIP codes with fewer than 11 Part D prescription claims have been excluded, which is the conventional practice at CMS. The Rural-Urban Commuting Area Codes (RUCA) developed by the Department of Agriculture Economic Research Service are used to determine a ZIP code’s rural and urban status (Economic Research Service 2010). Though the RUCA codes are originally designed for census tracts, the Rural Health Center ZIP Code RUCA Approximation crosswalk allows us to extend the rural/urban definitions to ZIP codes (University of Washington Rural Health Center 2010). It should be noted that 87 urban and 32 rural ZIP codes have missing data on key variables (see below) and are excluded from the analysis.1 The final sample contains 18,007 ZIP codes (12,176 urban and 5,828 rural), which covers 95% of all ZIP codes with any CMS data. It should be emphasized that we use data at different time periods to establish the temporal order among our key independent variable, mediators, and dependent variable.

Measures

The dependent variable, the opioid prescribing rate, comes from the 2017 Medicare Part D Prescription Drug Event (PDE) data and this analysis only includes prescriptions prescribed to Medicare Part D beneficiaries aged 65 or older. The opioid prescribing rate is the number of opioid drug claims, including original prescriptions and refills, divided by overall claims and expressed as a percentage. As the distribution of this variable is positively skewed, we transform the original variable with square rooting (the power transformation λ ≈ 0.50).

The data for the key independent variable and mediators come from the 2010–2015 ACS. Income inequality in a ZIP code is assessed with the Gini index and serves as the key independent variable. It ranges from 0 (completely equal) to 1 (completely unequal). While there are other income inequality indicators (e.g., Robin Hood index), the Gini index is a widely used measure in health research (Kawachi and Kennedy 1997).

As for mediators, we have two variables: residential stability and social isolation. Residential stability in a ZIP code is gauged with the percentage of housing units occupied by owners.2 Despite the lack of a universal measurement of social isolation, based on a recent theory-driven approach (United Health Foundation 2018), we construct a social isolation index by first standardizing the following variables focusing on the elderly (65+): the percent of older adults with a disability, the percent of older adults who live alone, and the percent of older adults living in poverty. We then use the average of the three standardized scores to assess the level of social isolation among the older population in a ZIP code.3

As for the other independent variables, we consider 5 beneficiary characteristics and 1 prescriber characteristic and 3 community environment variables. The beneficiary characteristic measures come from the 2017 CMS Master Beneficiary Summary File Base and for this analysis, only characteristics of aged beneficiaries who are enrolled in and utilize the Medicare Part D program are included. The beneficiary characteristics in a ZIP code include: average age, percent non-white beneficiaries, percent female beneficiaries, percent of Medicare Part D beneficiaries dually eligible for Medicaid, and the average Medicare Hierarchical Condition Categories (HCC) risk score. The HCC score is a composite score created by CMS. The prescriber characteristic is the percentage of pain management doctors, sports medicine doctors, surgeons, and oncologists. These practitioners are more likely to prescribe opioids due to the nature of their patients. To construct this measure, the PDE, the National Plan and Provider Enumeration System (NPPES), and Medicare Part B Fee-for-Service (FFS) claims data were linked using the National Provider Identifier (NPI) and then aggregated to the ZIP code level.

Community (i.e., ZIP code) environment variables include the percent of older people without a high school diploma, the percent of employed population in natural resources, construction, and maintenance, and the percent of older married people. Data from the 2010–2015 ACS were used for these community environment measures.

Analytic strategy

Our analytic strategy encompasses three stages. First, we conduct descriptive analysis with the full samples in order to gain a basic understanding of our data. The results by rural/urban status are then obtained and t-tests are used to examine whether the group differences exist. The second stage is to conduct two linear regression models. One includes all covariates except for the two mediators, and the other model takes into account residential stability and the social isolation index. The two models allow us to observe how the relationship between income inequality and the opioid prescribing rate changes after the mediators are considered. The nested linear regression models are applied to all samples, urban, and rural ZIP codes, respectively.

In the final stage, we use the PROCESS macro developed by Hayes (2017) to conduct mediation analysis. This macro is designed for continuous or binary observed variables and has been commonly used in the social sciences to estimate both direct and indirect effects. PROCESS allows multiple mediators and provides bootstrapping as an estimation approach for statistical inference. As bootstrapping yields robust standard errors for both direct and indirect effects, we use this approach to obtain the estimates and present the results accordingly. Further details about PROCESS can be found elsewhere (Hayes 2017; Hayes and Rockwood 2020).

RESULTS

The descriptive statistics of the variables used in this study are presented in Table 1. Several findings are notable. First, overall, there are 4.723 opioid prescriptions per 100 claims in all ZIP codes (i.e., 1.981 square rooted opioid prescription rate) with a relatively large standard deviation (5.114). Though the original opioid prescribing rate is higher in rural ZIP codes (4.740) than urban ZIP codes (4.715), we do not find any significant difference between rural and urban ZIP codes for this variable. This pattern is also observed for the transformed opioid prescribing rates. Second, the average Gini index is fairly stable across groups and the standard deviation is relatively small compared to the mean value. As a result, there is no rural/urban difference in income inequality in our data.

Table 1.

Descriptive statistics of all variables, all samples by rural/urban status, and group-comparison results.

All ZIP Codes
N=18,007		 Urban ZIP Codes
N=12,179		 Rural ZIP Codes
N=5,828		 Two-sample Comparison
Mean SD Mean SD Mean SD Mean difference
Dependent Variable
 Square Rooted Opioid Prescribing Rate 1.981 0.893 1.984 0.883 1.976 0.915 0.008
 (Original Opioid Prescribing Rate) 4.723 5.114 4.715 5.111 4.740 5.119 −0.025
Key Independent Variable
 Gini Index 0.431 0.054 0.431 0.056 0.432 0.051 −0.001
Mediators
 Residential Stability 68.940 16.327 67.188 18.337 72.601 10.058 −5.414 ***
 Social Isolation 0.009 0.708 −0.067 0.728 0.168 0.635 −0.235 ***
Beneficiary Characteristics
 Average Age of Beneficiaries 75.143 1.055 75.109 1.053 75.214 1.056 −0.105 ***
 Percent Non-white Beneficiaries 18.981 22.617 22.652 24.081 11.309 16.784 11.340 ***
 Percent Female Beneficiaries 58.573 3.492 58.719 3.450 58.268 3.557 0.451 ***
 Percent Dual Beneficiaries 18.633 13.319 18.354 14.066 19.217 11.583 −0.863 ***
 Average HCC Score 1.151 0.169 1.169 0.173 1.114 0.155 0.056 ***
Prescriber Characteristic
 Percent Doctors 3.876 7.395 4.347 7.205 2.892 7.687 1.455 ***
Community Environment Variables
 Percent of Older Population without a High School Diploma 19.657 12.499 18.832 12.586 21.380 12.135 −2.548 ***
 Percent of Employed Population in Natural Resources, Construction, and Maintenance 10.512 5.853 9.141 5.197 13.377 6.107 −4.236 ***
 Percent of Older Married Population 55.613 11.831 54.760 12.380 57.393 10.370 −2.633 ***
Open in a new tab
*

p<0.05;

**

p<0.01;

***

p<0.001.

Third, both residential stability and social isolation differ between rural and urban ZIP codes. Specifically, residential stability, on average, is roughly 5 percentage points higher in rural (72.60%) than urban ZIP codes (67.19%). Older adults in rural ZIP codes are more socially isolated (0.168) than their counterparts living in urban ZIP codes (−0.067). The statistically significant differences in mediators provide auxiliary evidence that the mechanisms may differ between rural and urban areas. Finally, we find that all other covariates vary by rural/urban status. For example, percent of older adults without a high school diploma is higher in rural (21.38%) than urban ZIP codes (18.83%), and the concentration of specialty doctors is higher in urban areas (4.35%) than rural areas (2.89%).

The nested linear regression models for all samples, urban, and rural ZIP codes are shown in Table 2. We summarize the key findings as follows. First, when analyzing all samples without mediators (Model 1), income inequality is positively associated with the opioid prescribing rate. For every 0.1 unit increase in the Gini index, the square rooted opioid prescribing rate increases by 0.03 (0.304*0.1=0.03). However, including residential stability and social isolation in the analysis makes the positive impact of income inequality on the opioid prescribing rate statistically non-significant. The change in the coefficient of the Gini index between Models 1 and 2 indicates that residential stability and social isolation may account for the association between income inequality and the opioid prescribing rate. Moreover, in Model 2, residential stability is negatively associated with the opioid prescribing rate and social isolation has a positive association with the opioid prescribing rate, both relationships following our expectation. Second, the results of urban ZIP codes suggest that income inequality may not be a determinant of the opioid prescribing rate as the Gini index is not significant in both Models 3 and 4. Nonetheless, it should be noted that both residential stability and social isolation still demonstrate expected associations with the opioid prescribing rate. These findings suggest that income inequality may not directly affect the opioid prescribing rate, but it may indirectly influence opioid prescribing via residential stability and social isolation.

Table 2.

Linear regression analysis of square rooted opioid prescribing rate, by rural/urban status.‡

All ZIP Codes
N= 18,007		 Urban ZIP Codes
N=12,179		 Rural ZIP Codes
N=5,828
Model 1 Model 2 Model 3 Mode 4 Model 5 Model 6
 Gini Index 0.304 (0.134)* −0.099 (0.142) 0.183 (0.162) −0.281 (0.173) 0.646 (0.250)* 0.274 (0.260)
Mediators
 Residential Stability −0.003 (0.001)*** −0.002 (0.001)*** −0.007 (0.001)***
 Social Isolation 0.107 (0.016)*** 0.130 (0.020)*** 0.066 (0.028)*
Beneficiary/Prescriber Characteristics
 Average Age of Beneficiaries −0.106 (0.007)*** −0.107 (0.007)*** −0.128 (0.009)*** −0.125 (0.009)*** −0.061 (0.013)*** −0.058 (0.013)***
 Percent Non-white Beneficiaries −0.001 (0.000)* −0.000 (0.001) −0.001 (0.001) 0.000 (0.001) −0.000 (0.001) −0.001 (0.001)
 Percent Female Beneficiaries 0.017 (0.002)*** 0.015 (0.002)*** 0.019 (0.003)*** 0.017 (0.003)*** 0.007 (0.004) 0.002 (0.004)
 Percent Dual Beneficiaries −0.003 (0.001)** −0.006 (0.001)*** −0.005 (0.001)*** −0.008 (0.001)*** 0.001 (0.002) −0.001 (0.002)
 Average HCC Score 0.153 (0.063)* 0.132 (0.063)* 0.236 (0.081)** 0.175 (0.082)* 0.080 (0.110) 0.064 (0.110)
 Percent Doctors 0.019 (0.001)*** 0.019 (0.001)*** 0.023 (0.001)*** 0.023 (0.001)*** 0.014 (0.002)*** 0.013 (0.002)***
Community Environment
 Percent of Older Population without a High School Diploma −0.005 (0.001)*** −0.006 (0.001)*** −0.005 (0.001)*** −0.006 (0.001)*** −0.006 (0.001)*** −0.007 (0.001)***
 Percent of Employed Population in Natural Resources, Construction, and Maintenance 0.009 (0.001)*** 0.008 (0.001)*** 0.008 (0.002)*** 0.008 (0.002)*** 0.005 (0.002)** 0.005 (0.002)*
 Percent of Older Married Population −0.002 (0.001)** 0.002 (0.001)* −0.003 (0.001)** 0.002 (0.001) −0.001 (0.001) 0.001 (0.002)
Constant 8.785 (0.493)*** 9.184 (0.496)*** 10.271 (0.602)*** 10.428 (0.603)*** 5.892 (0.913)*** 6.551(0.920)***x
Open in a new tab
‡

Bootstrapping standard errors are in the parentheses.

*

p<0.05;

**

p<0.01;

***

p<0.001.

Third, regarding the findings drawn from rural ZIP codes, income inequality is a strong determinant of the opioid prescribing rate in Model 5 and a 0.1-unit increase in the Gini index is associated with a 0.065-unit (0.646*0.1=0.065) increase in the square rooted opioid prescribing rate. The inclusion of residential stability and social isolation drops the statistical significance of income inequality (Model 6), which indicates that these variables may mediate the direct impact of income inequality on the opioid prescribing rate among rural ZIP codes. It should be noted that unlike the findings in Models 1–4, many beneficiary characteristics are not associated with the opioid prescribing rate for rural ZIP codes, such as average HCC score and percent of dual beneficiaries. That being said, the determinants of the opioid prescribing rate differ between rural and urban ZIP codes in that beneficiary characteristics are important in urban ZIP codes and contextual variables (e.g., income inequality) are critical in rural areas.

The findings in the nested linear regression models offer some support for the mediating mechanisms between income inequality and the opioid prescribing rate; however, it is necessary to conduct formal mediation analyses to thoroughly examine the proposed pathways. Following the analytic strategy, we present the PROCESS results in Table 3 and highlight the key findings below. There are three panels in Table 3 and panels (a), (b), and (c) show the PROCESS results for all samples, urban ZIP codes, and rural ZIP codes, respectively. For all samples, the PROCESS results indicate that income inequality is not directly related to the opioid prescribing rate (direct effect= −0.099, bootstrapping S.E.=0.170) and the effect mediated by residential stability and social isolation is 0.404 (bootstrapping S.E.=0.055). Between the two mediators, residential stability accounts for 38.79% of the mediating effect and social isolation explains 61.23%. As expected, both mediating effects are positive, which follows theoretical expectations. Specifically, income inequality worsens residential stability (negative association) and residential stability is also negatively associated with the opioid prescribing rate. As a result, the overall mediating impact should be positive. Similarly, income inequality increases social isolation and social isolation, in turn, increases the opioid prescribing rate. The mediating effect, hence, should be positive.

Table 3.

The PROCESS path analysis of square rooted opioid prescribing rate, by rural/urban status✝

Estimates Boot. S.E. Boot. LLCI Boot. ULCI Mediation percentage
(a) Model for All Zip Codes
 Total effect 0.304 (0.134) NA NA NA
 Direct effect −0.099 (0.170) −0.435 0.236 NA
 Mediating effect 0.404 (0.055) 0.294 0.513 NA
Through
 Residential Stability 0.157 (0.037) 0.084 0.229 38.79
 Social Isolation 0.247 (0.045) 0.161 0.336 61.23
(b) Model for Urban ZIP Codes
 Total effect 0.184 (0.162) NA NA NA
 Direct effect −0.281 (0.207) −0.694 0.126 NA
 Mediating effect 0.464 (0.075) 0.320 0.615 NA
Through
 Residential Stability 0.159 (0.055) 0.053 0.267 34.20
 Social Isolation 0.306 (0.058) 0.198 0.426 65.80
(c) Model for Rural ZIP Codes
 Total effect 0.646 (0.250) NA NA NA
 Direct effect 0.274 (0.306) −0.336 0.886 NA
 Mediating effect 0.372 (0.086) 0.204 0.540 NA
Through
 Residential Stability 0.256 (0.060) 0.145 0.382 68.92
 Social Isolation 0.116 (0.058) 0.001 0.233 31.06
Open in a new tab
✝

If 0 is not included in the bootstrapping confidence interval, the estimated direct or indirect is significant. All models include all other covariates discussed in this study. NA = Not applicable. PROCESS does not yield bootstrapping confidence levels for total effects. Boot S.E.: Bootstrapping standard errors; Boot LLCI: Bootstrapping lower limit of 95% confidence interval; Boot ULCI: Bootstrapping upper limit of 95% confidence interval

Second, the PROCESS results of urban ZIP codes confirm that income inequality indirectly affects the opioid prescribing rate by poor residential stability and elevated social isolation. The former accounts for more than one-third of the total mediating effect, whereas the latter explains almost 66 percent of the indirect effect of the Gini index on the opioid prescribing rate. As for rural ZIP codes, the PROCESS results suggest that residential stability plays a larger role in explaining the mediating effect than social isolation does because the former accounts for almost 70 percent of the indirect effect, which almost doubles those found in all samples or urban ZIP codes.

Third, these findings offer strong evidence to support the residential stability and social isolation mechanisms through which income inequality exerts its influence on the opioid prescribing rate. Residential stability is more important among rural ZIP codes than urban ZIP codes; nonetheless, social isolation channels a larger proportion of the indirect effect on the opioid prescribing rate than residential stability under the urban setting.

Beyond the key relationships among income inequality, residential stability, social isolation, and the opioid prescribing rates, our results offer consistent evidence for several community environment variables. For example, in Table 2, percent of employed population in natural resources, construction, and maintenance is positively associated with the opioid prescribing rates in all models. This relationship suggests areas featured with labor-intensive industries are likely to have a high demand for opioid treatment. Similarly, a high average age of beneficiaries is negatively associated with opioid prescribing rates, indicating the age-dependent prescription patterns and highlighting the importance of age structure in opioid prescribing rates.

CONCLUSIONS AND DISCUSSION

Prescription opioids significantly contribute to the growing opioid epidemic in the US; however, little is known about how income inequality shapes opioid prescribing rates and even less about whether the mechanisms linking income inequality and opioid prescribing rates differ between rural and urban areas. This study fills these gaps by examining the mediating roles of residential stability and social isolation and comparing the results by rural/urban status. The findings above are used to revisit our hypotheses. We first hypothesized that high levels of income inequality are associated with high opioid prescribing rates, net of other potential confounders. This hypothesis is partially supported and our findings suggest that the relationship between income inequality and opioid prescribing rates is complex. Specifically, the linear regression results of all samples confirm the positive relationship between income inequality and opioid prescribing rates but this relationship becomes non-significant with the inclusion of residential stability and social isolation. In addition, the significant and positive relationship in the all samples analysis is driven by rural ZIP codes, rather than urban ZIP codes.

Our second hypothesis is concerned with the residential stability pathway. We hypothesized that income inequality is negatively associated with residential stability, and residential stability is also negatively related to opioid prescribing rates. The linear regression and PROCESS results offer evidence to bolster this mechanism. The linear regression modeling (Table 2) yields a negative relationship between residential stability and opioid prescribing rates. Given the positive mediating effect through residential stability (Table 3), the association between income inequality and residential stability should also be negative. In other words, we obtain evidence to support the following mechanism: high levels of income inequality are associated with low levels of residential stability, which in turn increases opioid prescribing rates. This mechanism holds for both rural and urban ZIP codes.

Similar to the residential stability pathway, our results also support the mediating role of social isolation. The positive association between social isolation and opioid prescribing rates is statistically significant in Table 2. As the mediating effect via social isolation is positive, income inequality has to be positively related to social isolation. That is, this pathway can be understood as the following: increasing income inequality is associated with increasing social isolation, and high prevalence of social isolation further raises opioid prescribing rates. Our third hypothesis (H3) is supported.

Our fourth hypothesis consists of three sub-hypotheses. We first argued that the association between income inequality and opioid prescribing rates is stronger in rural areas than in urban areas (H4a). This sub-hypothesis receives support from the linear regression and PROCESS results as the total effect of income inequality on the opioid prescribing rate of rural ZIP codes (0.65) is stronger than that of urban ZIP codes (0.18). Moreover, we expected that residential stability mediates the effect of income inequality on opioid prescribing rates more in rural areas than in urban areas (H4b). This sub-hypothesis is also supported by the findings as residential stability mediates almost 70 percent of the total indirect effect among rural ZIP codes (Table 3), compared to 34 percent among urban ZIP codes. Finally, hypothesis (4c) stated that social isolation plays a larger role in urban areas than in rural areas in mediating the impact of income inequality on opioid prescribing rates. Similar to H4b, this sub-hypothesis is bolstered in that social isolation accounts for almost two-thirds of the total indirect effect among urban ZIP codes, in contrast to slightly more than 30 percent among rural ZIP codes.

How do our findings advance the extant literature? First, unlike prior cross-sectional studies (Chen et al. 2019; Ford and Rigg 2015; Monnat and Rigg 2016), this study built a dataset where income inequality, residential stability, and social isolation are assessed prior to the opioid prescribing rates. Our data structure allows us to yield strong evidence to answer the question of whether high levels of income inequality may lead to high opioid prescribing rates. Second, this study identifies the pathways between income inequality and opioid prescribing rates. Previous studies aim to explore the determinants of opioid prescribing rates or other opioid-related outcomes (García et al. 2019; Heins et al. 2018; Lund et al. 2019) and overlook the pathways linking the macroeconomic conditions or structural factors to the opioid epidemic (Dasgupta et al. 2018). Our findings clarify the intertwined relationships among key factors. Third, while scholars have paid attention to the rural/urban differences or geographical heterogeneity in opioid-related outcomes (Keyes et al. 2014; Monnat 2019; Rigg and Monnat 2015; Rigg, Monnat, and Chavez 2018), few researchers investigate whether the mechanisms are the same between rural and urban areas. Our findings advance the literature by suggesting that not only the determinants, but also the pathways between structural factors and opioid-related outcomes, may differ by rural/urban status.

We conducted several sensitivity analyses to assure the robustness of our findings and conclusions. For example, we implemented the mediation analysis with lavaan in R (Rosseel 2012) and obtained the same results. That is, our conclusions are not altered by the choice of analytic methods or estimation approach. Also, the spatial correlation in our dependent variable is fairly low (Moran’s I =0.04), suggesting that the potential bias caused by spatial dependency should not be a concern. All the sensitivity analysis results are available upon request.

Some policy implications can be drawn from the findings, especially given the relatively small unit of analysis (i.e., ZIP code) in this study. For one, it is important to keep the turnover rates low in an area and housing subsidy policies, such as first-time home buyer incentives or tax credits, may help individuals to either become homeowners or stay in the same community for a long time. Furthermore, it is important to engage marginalized older adults (e.g., living in poverty or living alone) in regular social activities or contacts. Reducing the percentages of socially isolated populations should reduce the demand for opioids. Finally, it remains critical to create an equal environment or to facilitate the access to resources in order to alleviate the adverse impact of income inequality on opioid prescribing rates.

This study is subject to several limitations. First, like other ecological studies, the findings and conclusions may be altered if the CMS data are aggregated into different geographical levels, such as county or health service areas. This is known as the modifiable areal unit problem (Openshaw 1984). Second, our findings cannot be generalized to other populations because the individual data are from the older beneficiaries enrolled in the Medicare Part D program. Third, we employ the Gini index to assess income inequality and using a different income inequality indicator (e.g., decile ratio) may lead to different conclusions. Fourth, due to data limitations, we are unable to consider some variables related to opioid prescribing rates. For example, no data are available to measure social capital and social ties at the ZIP code level (to our knowledge, the finest spatial resolution is county for such variables) and ACS does not provide information on individuals who live in the same house for at least five years, which is also a common indicator of residential stability (Boardman 2004). Finally, there may be other possible pathways linking income inequality and opioid prescribing rates. Future research should endeavor to identify other mechanisms beyond residential stability and social isolation.

Despite the limitations, this study contributes to the extant literature in three ways. First, the relationship between income inequality and the opioid prescribing rate may be largely explained by residential stability and social isolation. Our findings provide nuanced insight into how structural factors, particularly macroeconomic conditions, may affect the opioid prescribing rate. Second, while both residential stability and social isolation are important pathways, the PROCESS findings indicate that the former is more important for rural ZIP codes and the latter is more critical for urban ZIP codes. To our knowledge, no prior research offers evidence for the different pathways between rural and urban areas. Third, beneficiary characteristics are not significant determinants of the opioid prescribing rate among rural ZIP codes, suggesting that future effort to explore rural/urban differences in the opioid prescribing rate should consider other contextual factors, especially for rural areas.

Footnotes

1

As the number of ZIP codes excluded in the analysis is low, they do not affect our findings and conclusions.

2

The housing tenure variable is not further divided by age so that we are unable to calculate this variable for the older population (aged 65+).

3

The Cronbach’s alpha is 0.6 (fairly strong given the small number of items) and the inter-item correlation among the three variables is 0.31, which suggests good reliability (Briggs and Cheek 1986).

Contributor Information

Tse-Chuan Yang, University at Albany, 1400 Washington Ave., Arts & Sciences 351, Albany, NY 12222

Seulki Kim, University at Albany, 1400 Washington Ave., Arts & Sciences 356, Albany, NY 12222.

Carla Shoff, Centers for Medicare & Medicaid Services, 7500 Security Boulevard, Baltimore, MD 21244

REFERENCES

  1. Beggs John J, Valerie A Haines and Jeanne S Hurlbert. 1996. “Revisiting the Rural‐Urban Contrast: Personal Networks in Nonmetropolitan and Metropolitan Settings 1.” Rural Sociology 61(2):306–25. [Google Scholar]
  2. Boardman Jason D. 2004. “Stress and Physical Health: The Role of Neighborhoods as Mediating and Moderating Mechanisms.” Social science & medicine 58(12):2473–83. [DOI] [PubMed] [Google Scholar]
  3. Borjas George J. 1987. “Self-Selection and the Earnings of Immigrants.” National Bureau of Economic Research Cambridge, Mass., USA. [Google Scholar]
  4. Briggs Stephen R and Jonathan M Cheek. 1986. “The Role of Factor Analysis in the Development and Evaluation of Personality Scales.” Journal of personality 54(1):106–48. [Google Scholar]
  5. Chen Qiushi, Marc R Larochelle, Davis T Weaver, Anna P Lietz, Peter P Mueller, Sarah Mercaldo, Sarah E Wakeman, Kenneth A Freedberg, Tiana J Raphel and Amy B Knudsen. 2019. “Prevention of Prescription Opioid Misuse and Projected Overdose Deaths in the United States.” JAMA network open 2(2):e187621–e21. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Click Ivy A, Jeri Ann Basden, Joy M Bohannon, Heather Anderson and Fred Tudiver. 2018. “Opioid Prescribing in Rural Family Practices: A Qualitative Study.” Substance use & misuse 53(4):533–40. [DOI] [PubMed] [Google Scholar]
  7. Cochran Gerald, Rosen Daniel, Rebecca M McCarthy and Rafael J Engel. 2017. “Risk Factors for Symptoms of Prescription Opioid Misuse: Do Older Adults Differ from Younger Adult Patients?”. Journal of gerontological social work 60(6–7):443–57. [DOI] [PubMed] [Google Scholar]
  8. Cohen Frances, Margaret E Kemeny, Leonard S Zegans, Paul Johnson, Kathleen A Kearney and Daniel P Stites. 2007. “Immune Function Declines with Unemployment and Recovers after Stressor Termination.” Psychosomatic medicine 69(3):225–34. [DOI] [PubMed] [Google Scholar]
  9. Cornwell Erin York and Linda J Waite. 2009. “Social Disconnectedness, Perceived Isolation, and Health among Older Adults.” Journal of health and social behavior 50(1):31–48. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Corrigan Patrick W and Katherine Nieweglowski. 2018. “Stigma and the Public Health Agenda for the Opioid Crisis in America.” International Journal of Drug Policy 59:44–49. [DOI] [PubMed] [Google Scholar]
  11. Dasgupta Nabarun, Beletsky Leo and Ciccarone Daniel. 2018. “Opioid Crisis: No Easy Fix to Its Social and Economic Determinants.” American journal of public health 108(2):182–86. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Economic Research Service. 2010, “Rural-Urban Commuting Area Codes”. Retrieved August 5, 2019 (https://www.ers.usda.gov/data-products/rural-urban-commuting-area-codes.aspx).
  13. Fernald Lia CH and Megan R Gunnar. 2009. “Poverty-Alleviation Program Participation and Salivary Cortisol in Very Low-Income Children.” Social science & medicine 68(12):2180–89. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Ford Jason A and Khary K Rigg. 2015. “Racial/Ethnic Differences in Factors That Place Adolescents at Risk for Prescription Opioid Misuse.” Prevention Science 16(5):633–41. [DOI] [PubMed] [Google Scholar]
  15. Frey William H, Kao-Lee Liaw, Yu Xie and Marcia J Carlson. 1996. “Interstate Migration of the Us Poverty Population: Immigration “Pushes” and Welfare Magnet “Pulls”.” Population and Environment 17(6):491–533. [Google Scholar]
  16. García Macarena, Charles M Heilig, Scott H Lee, Mark Faul, Gery Guy, Michael F Iademarco, Katherine Hempstead, Dorrie Raymond and Josh Gray. 2019. “Opioid Prescribing Rates in Nonmetropolitan and Metropolitan Counties among Primary Care Providers Using an Electronic Health Record System—United States, 2014–2017.” Morbidity and Mortality Weekly Report 68(2):25. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Han Huijun, Philip H Kass, Barth L Wilsey and Chin-Shang Li. 2012. “Individual and County-Level Factors Associated with Use of Multiple Prescribers and Multiple Pharmacies to Obtain Opioid Prescriptions in California.” PloS one 7(9). [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Hawkley Louise C and John T Cacioppo. 2007. “Aging and Loneliness: Downhill Quickly?”. Current Directions in Psychological Science 16(4):187–91. [Google Scholar]
  19. Hayes Andrew F. 2012. “Process: A Versatile Computational Tool for Observed Variable Mediation, Moderation, and Conditional Process Modeling.” University of Kansas, KS. [Google Scholar]
  20. Hayes Andrew F. 2017. Introduction to Mediation, Moderation, and Conditional Process Analysis: A Regression-Based Approach: Guilford publications. [Google Scholar]
  21. Hayes Andrew F and Nicholas J Rockwood. 2020. “Conditional Process Analysis: Concepts, Computation, and Advances in the Modeling of the Contingencies of Mechanisms.” American Behavioral Scientist 64(1):19–54. [Google Scholar]
  22. Heins Sara E, Mark J Sorbero, Christopher M Jones, Andrew W Dick and Bradley D Stein. 2018. “High-Risk Prescribing to Medicaid Enrollees Receiving Opioid Analgesics: Individual-and County-Level Factors.” Substance use & misuse 53(10):1591–601. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Hipp John R. 2007. “Income Inequality, Race, and Place: Does the Distribution of Race and Class within Neighborhoods Affect Crime Rates?”. Criminology 45(3):665–97. [Google Scholar]
  24. Hofferth Sandra L and John Iceland. 1998. “Social Capital in Rural and Urban Communities 1.” Rural Sociology 63(4):574–98. [Google Scholar]
  25. Hruschak Valerie and Cochran Gerald. 2017. “Psychosocial and Environmental Factors in the Prognosis of Individuals with Chronic Pain and Comorbid Mental Health.” Social work in health care 56(7):573–87. [DOI] [PubMed] [Google Scholar]
  26. Huhn Andrew S, Eric C Strain, Tompkins D Andrew and Kelly E Dunn. 2018. “A Hidden Aspect of the Us Opioid Crisis: Rise in First-Time Treatment Admissions for Older Adults with Opioid Use Disorder.” Drug and alcohol dependence 193:142–47. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Jeffery Molly Moore, Hooten W Michael, Henk Henry J, Bellolio M Fernanda, Hess Erik P, Meara Ellen, Ross Joseph S and Shah Nilay D. 2018. “Trends in Opioid Use in Commercially Insured and Medicare Advantage Populations in 2007–16: Retrospective Cohort Study.” bmj 362:k2833. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Jonas Adam B, April M Young, Carrie B Oser, Carl G Leukefeld and Jennifer R Havens. 2012. “Oxycontin® as Currency: Oxycontin® Use and Increased Social Capital among Rural Appalachian Drug Users.” Social science & medicine 74(10):1602–09. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Kawachi Ichiro and Bruce P Kennedy. 1997. “The Relationship of Income Inequality to Mortality: Does the Choice of Indicator Matter?”. Social science & medicine 45(7):1121–27. [DOI] [PubMed] [Google Scholar]
  30. Kawachi Ichiro, Bruce P Kennedy, Kimberly Lochner and Deborah Prothrow-Stith. 1997. “Social Capital, Income Inequality, and Mortality.” American journal of public health 87(9):1491–98. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Kawachi Ichiro. 2000. “Income Inequality and Health.” Social epidemiology 76:94. [Google Scholar]
  32. Kawachi Ichiro and Berkman Lisa. 2000. “Social Cohesion, Social Capital, and Health.” Social epidemiology 174(7). [Google Scholar]
  33. Keyes Katherine M, Magdalena Cerdá, Joanne E Brady, Jennifer R Havens and Sandro Galea. 2014. “Understanding the Rural–Urban Differences in Nonmedical Prescription Opioid Use and Abuse in the United States.” American journal of public health 104(2):e52–e59. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Kipke Michele D, George Weiss and Carolyn F Wong. 2007. “Residential Status as a Risk Factor for Drug Use and Hiv Risk among Young Men Who Have Sex with Men.” AIDS and Behavior 11(2):56–69. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Kobayashi Karen M, Denise Cloutier-Fisher and Marilyn Roth. 2009. “Making Meaningful Connections: A Profile of Social Isolation and Health among Older Adults in Small Town and Small City, British Columbia.” Journal of Aging and Health 21(2):374–97. [DOI] [PubMed] [Google Scholar]
  36. Kolodny Andrew, David T Courtwright, Catherine S Hwang, Peter Kreiner, John L Eadie, Thomas W Clark and Alexander G Caleb. 2015. “The Prescription Opioid and Heroin Crisis: A Public Health Approach to an Epidemic of Addiction.” Annual review of public health 36:559–74. [DOI] [PubMed] [Google Scholar]
  37. Kuo Yong-Fang, Mukaila A Raji Nai-Wei Chen, Hasan Hunaid and James S Goodwin. 2016. “Trends in Opioid Prescriptions among Part D Medicare Recipients from 2007 to 2012.” The American journal of medicine 129(2):221. e21–21. e30. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Lazarus Richard S. 1993. “Coping Theory and Research: Past, Present, and Future.” Fifty years of the research and theory of RS Lazarus: An analysis of historical and perennial issues:366–88. [Google Scholar]
  39. Li Jiaqi, Bell Thomas and Chollet Deborah. 2018. “Patterns of Opioid Prescribing in Minnesota: 2012 and 2015.” Vol.: Mathematica Policy Research. [Google Scholar]
  40. Lund Brian C, Michael E Ohl, Katherine Hadlandsmyth and Hilary J Mosher. 2019. “Regional and Rural–Urban Variation in Opioid Prescribing in the Veterans Health Administration.” Military medicine. [DOI] [PubMed] [Google Scholar]
  41. Maree Rachel D, Zachary A Marcum, Ester Saghafi, Debra K Weiner and Jordan F Karp. 2016. “A Systematic Review of Opioid and Benzodiazepine Misuse in Older Adults.” The American Journal of Geriatric Psychiatry 24(11):949–63. [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. McDonald Douglas C, Kenneth Carlson and David Izrael. 2012. “Geographic Variation in Opioid Prescribing in the Us.” The journal of Pain 13(10):988–96. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. McLean Katherine. 2016. ““There’s Nothing Here”: Deindustrialization as Risk Environment for Overdose.” International Journal of Drug Policy 29:19–26. [DOI] [PubMed] [Google Scholar]
  44. Monnat Shannon M and Khary K Rigg. 2016. “Examining Rural/Urban Differences in Prescription Opioid Misuse among Us Adolescents.” The Journal of Rural Health 32(2):204–18. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Monnat Shannon M. 2019. “The Contributions of Socioeconomic and Opioid Supply Factors to Us Drug Mortality Rates: Urban-Rural and within-Rural Differences.” Journal of Rural Studies 68:319–35. [Google Scholar]
  46. NIDA. 2018. Drugs, Brains, and Behavior: The Science of Addiction. No. 18-DA-5605Congress. [Google Scholar]
  47. Nolan Brian and Ive Marx. 2009. “Economic Inequality, Poverty, and Social Exclusion.” in The Oxford Handbook of Economic Inequality. [Google Scholar]
  48. Openshaw Stan. 1984. “The Modifiable Areal Unit Problem.” Concepts and techniques in modern geography. [Google Scholar]
  49. Osborne Danny, Chris G Sibley and Nikhil Kumar Sengupta. 2015. “Income and Neighbourhood‐Level Inequality Predict Self‐Esteem and Ethnic Identity Centrality through Individual‐and Group‐Based Relative Deprivation: A Multilevel Path Analysis.” European Journal of Social Psychology 45(3):368–77. [Google Scholar]
  50. Park Kiwoong and Yang Tse-Chuan. 2017. “The Long-Term Effects of Self-Esteem on Depression: The Roles of Alcohol and Substance Use During Young Adulthood.” The Sociological Quarterly 58(3):429–46. [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Pelzer B, Schaffrath S and Vernaleken Ingo. 2014. “Coping with Unemployment: The Impact of Unemployment on Mental Health, Personality, and Social Interaction Skills.” Work 48(2):289–95. [DOI] [PubMed] [Google Scholar]
  52. Rankin Bruce H and James M Quane. 2000. “Neighborhood Poverty and the Social Isolation of Inner-City African American Families.” Social forces 79(1):139–64. [Google Scholar]
  53. Rigg Khary K and Shannon M Monnat. 2015. “Urban Vs. Rural Differences in Prescription Opioid Misuse among Adults in the United States: Informing Region Specific Drug Policies and Interventions.” International Journal of Drug Policy 26(5):484–91. [DOI] [PMC free article] [PubMed] [Google Scholar]
  54. Rigg Khary K, Shannon M Monnat and Melody N Chavez. 2018. “Opioid-Related Mortality in Rural America: Geographic Heterogeneity and Intervention Strategies.” International Journal of Drug Policy 57:119–29. [DOI] [PubMed] [Google Scholar]
  55. Rosseel Yves. 2012. “Lavaan: An R Package for Structural Equation Modeling and More. Version 0.5–12 (Beta).” Journal of statistical software 48(2):1–36. [Google Scholar]
  56. Sampson Robert J. 1991. “Linking the Micro-and Macrolevel Dimensions of Community Social Organization.” Social forces 70(1):43–64. [Google Scholar]
  57. Sampson Robert J, Jeffrey D Morenoff and Felton Earls. 1999. “Beyond Social Capital: Spatial Dynamics of Collective Efficacy for Children.” American sociological review:633–60. [Google Scholar]
  58. Scharf Thomas, Phillipson Chris and Allison E Smith. 2004. “Poverty and Social Exclusion: Growing Older in Deprived Urban Neighbourhoods.” Growing older: Quality of life in old age:81–106. [Google Scholar]
  59. Scholl Lawrence, Seth Puja, Kariisa Mbabazi, Wilson Nana and Baldwin Grant. 2019. “Drug and Opioid-Involved Overdose Deaths—United States, 2013–2017.” Morbidity and Mortality Weekly Report 67(5152):1419. [DOI] [PMC free article] [PubMed] [Google Scholar]
  60. South Scott J and Glenn D Deane. 1993. “Race and Residential Mobility: Individual Determinants and Structural Constraints.” Social forces 72(1):147–67. [Google Scholar]
  61. Subramanian Subu and Ichiro Kawachi. 2004. “Income Inequality and Health: What Have We Learned So Far?”. Epidemiologic reviews 26(1):78–91. [DOI] [PubMed] [Google Scholar]
  62. Substance Abuse and Mental Health Services Administration. 2018. “Key Substance Use and Mental Health Indicators in the United States: Results from the 2017 National Survey on Drug Use and Health (Hhs Publication No. Sma 17–5044, Nsduh Series H-52). Rockville, Md: Center for Behavioral Health Statistics and Quality.” Substance Abuse and Mental Health Services Administration. Retrieved from https://www.samhsa.gov/data. [Google Scholar]
  63. Turney Kristin and Harknett Kristen. 2010. “Neighborhood Disadvantage, Residential Stability, and Perceptions of Instrumental Support among New Mothers.” Journal of Family Issues 31(4):499–524. [DOI] [PMC free article] [PubMed] [Google Scholar]
  64. United Health Foundation. 2018, “America’s Health Rankings Senior Report 2018”, Saint Paul, Minnesota: United Health Foundation. Retrieved August 5, 2019 (https://assets.americashealthrankings.org/app/uploads/ahrsenior18-finalv1.pdf). [Google Scholar]
  65. University of Washington Rural Health Center. 2010, “Zip Code Ruca Approximation.”. Retrieved August 5, 2019 (http://depts.washington.edu/uwruca/ruca-approx.php). [Google Scholar]
  66. Warner Barbara D and Pamela Wilcox Rountree. 1997. “Local Social Ties in a Community and Crime Model: Questioning the Systemic Nature of Informal Social Control.” Social Problems 44(4):520–36. [Google Scholar]
  67. Webster Barbara S, Manuel Cifuentes, Santosh Verma and Glenn Pransky. 2009. “Geographic Variation in Opioid Prescribing for Acute, Work‐Related, Low Back Pain and Associated Factors: A Multilevel Analysis.” American journal of industrial medicine 52(2):162–71. [DOI] [PubMed] [Google Scholar]
  68. Weiss Roger D, Jennifer Sharpe, Margaret L Griffin, McHugh R Kathryn, Deborah Haller, Petra Jacobs, John Gardin II, Dan Fischer and Kristen D Rosen. 2014. “Reasons for Opioid Use among Patients with Dependence on Prescription Opioids: The Role of Chronic Pain.” Journal of substance abuse treatment 47(2):140–45. [DOI] [PMC free article] [PubMed] [Google Scholar]
  69. Wilkinson Richard G. 1997. “Comment: Income, Inequality, and Social Cohesion.” American journal of public health 87(9):1504–06. [DOI] [PMC free article] [PubMed] [Google Scholar]
  70. Wilson William Julius. 1987. “The Truly Disadvantaged: The Inner City.” The Underclass, and Public Policy 8. [Google Scholar]
  71. Wise Roy A, and George F Koob. 2014. “The Development and Maintenance of Drug Addiction.” Neuropsychopharmacology 39(2):254. [DOI] [PMC free article] [PubMed] [Google Scholar]
  72. Wright Eric R, Harold E Kooreman, Marion S Greene, Chambers R Andrew, Aniruddha Banerjee and Jeffrey Wilson. 2014. “The Iatrogenic Epidemic of Prescription Drug Abuse: County-Level Determinants of Opioid Availability and Abuse.” Drug and alcohol dependence 138:209–15. [DOI] [PubMed] [Google Scholar]
  73. Yang Tse‐Chuan, Jensen Leif and Haran Murali. 2011. “Social Capital and Human Mortality: Explaining the Rural Paradox with County‐Level Mortality Data.” Rural Sociology 76(3):347–74. [DOI] [PMC free article] [PubMed] [Google Scholar]
  74. Zhou Chao, Ning Neil Yu and Jan L Losby. 2018. “The Association between Local Economic Conditions and Opioid Prescriptions among Disabled Medicare Beneficiaries.” Medical care 56(1):62. [DOI] [PMC free article] [PubMed] [Google Scholar]
  75. Zoorob Michael J and Jason L Salemi. 2017. “Bowling Alone, Dying Together: The Role of Social Capital in Mitigating the Drug Overdose Epidemic in the United States.” Drug and alcohol dependence 173:1–9. [DOI] [PubMed] [Google Scholar]

RESOURCES