Social Vulnerability and Opioid Use Disorder Rate Among Older Medicare Beneficiaries in US Counties: How Has This Relationship Evolved Since Baby Boomers Entered Older Adulthood?

The prevalence of older Medicare beneficiaries (aged 65+) with opioid use disorder (OUD) has increased by almost four times between 2013 and 2018, and the increasing pattern holds regardless of race/ethnicity, sex, or poverty status (Shoff et al., 2021). This rise in older adults with OUD, particularly since 2010, has been identified as a hidden aspect of the opioid crisis in the United States (US) (Huhn et al., 2018) because prior research has concentrated on the working-age population (aged 15–64) and overlooked the older population. There are two explanations for why older adults are uniquely susceptible to OUD. On one hand, the aging process leads to several physical, mental, and behavioral health challenges (Le Roux et al., 2016; Rodin, 1986), such as bone density loss, anxiety, and substance misuse, and these conditions may heighten the exposure to prescription opioids among older adults (Rose et al., 2019). As the onset of OUD is positively associated with exposure to prescription opioids (Butler et al., 2016; Zhang et al., 2018), older adults may experience a high risk of OUD. It should be emphasized that other social risk factors, such as social exclusion and loneliness, may enhance the relationship between OUD and exposure to prescription opioids (Cochran et al., 2017; Huhn et al., 2018; Maree et al., 2016). On the other hand, older adults tend to overlook the negative consequences of opioid use, such as tolerance and physical dependence (Wang & Andrade, 2013), and they are less sensitive to opioid misuse or early symptoms of OUD than the younger population (West & Dart, 2016). As a result, older adults are likely to develop OUD.

The underrepresentation of older adults in the literature may reflect the focus of clinical practice and policies. As suggested by the Substance Abuse Mental Health Services Administration (SAMHSA) (2020), health care providers, professionals, and the general public often share the false belief that older adults do not develop or need treatment for OUD. This misconception contributes to low rates of diagnosis, treatment, and recovery among older adults. To address this issue, the federal government has developed several policies targeting OUD among older adults. For example, Medicare provides screening, brief intervention, and referral to treatment (SBIRT) services, which aim to prevent adverse outcomes of OUD and serve as an early intervention tool. From a clinical practice perspective, prescription drug monitoring programs (PDMPs) assist clinicians in improving opioid prescribing practices by reviewing patients’ history of controlled substance prescriptions and guiding clinical decisions regarding the initiation or continuation of opioid therapy (CDC, 2024b). Furthermore, Medicare’s opioid treatment programs (OTPs) provide another avenue for medication-assisted treatment for OUD. The US Centers for Disease Control and Prevention (CDC) also offers clinical practice guidelines for prescribing opioids for pain. These guidelines cover the entire process, from deciding to initiate opioid therapy to assessing and addressing the harms of opioid use (CDC, 2024a).

Despite these programs and policies, the number of older adults with OUD has increased and has imposed a heavy financial burden on Medicare. In 2019, it was estimated that OUD-attributable Medicare spending among older beneficiaries (age 65+) was approximately $2.9 billion, which was more than 60% higher than the spending among younger beneficiaries (age < 65) (Mark et al., 2023). The financial burden associated with OUD on Medicare may be further exacerbated due to baby boomers entering older adulthood. Baby boomers, a generation of roughly 73 million adults born between 1946 and 1964, had lived through an era when substance use was more socially acceptable (Le Roux et al., 2016) and a period of time when the opioid prescription rate increased rapidly (i.e., 1990–2010) (Kenan et al., 2012; Olsen et al., 2006). The development of OUD may reflect the accumulation of baby boomers’ prior experiences with substance use, the health care system, and social changes that continue into older adulthood. As the boomer generation has changed the demographic landscape of the US for decades, the influx of baby boomers into the older adult population since 2011, often referred to as the “gray tsunami” (US Census Bureau, 2019), may shape the future of OUD among older adults.

Scholars have investigated the determinants of OUD. At the individual level, older adults with poor health or living in poverty are more likely to develop OUD than their counterparts with good health or high socioeconomic status (Park & Lavin, 2010; Wu & Blazer, 2011). At the ecological level, areas with high levels of social isolation or concentrations of marginalized populations (e.g., people who are disabled or living alone) tend to have higher OUD rates among older adults (Yang, Shoff, & Kim, 2022; Yang, Shoff, Kim, et al., 2022). Although these studies have improved our understanding of OUD in the older population, an important factor related to older adults residential environment, namely social vulnerability, has not received sufficient attention (Dufort & Samaan, 2021; Yang, Shoff, Choi, et al., 2022).

The concept of social vulnerability refers to a community’s susceptibility to both natural and manmade emergencies and its ability to manage the stress associated with these emergencies (CDC, 2022). The CDC has developed the social vulnerability index (SVI) to measure this concept, which has been found to have a detrimental association with population health. For example, counties with high SVIs report high cancer and cardiovascular disease mortality rates (Ganatra et al., 2022). Similarly, teen birth and COVID-19 infection rates are higher among counties with high SVIs than those with low SVIs (Dasgupta et al., 2020; Yee et al., 2019). To our knowledge, only one prior study has shown a positive relationship between SVI and the OUD rate among older adults (Yang et al., 2023). Although these studies advance our understanding of how social vulnerability is related to population health outcomes, most rely on a cross-sectional research design and do not investigate the temporal trends since 2011, when baby boomers began entering older adulthood.

Drawing from the extant literature, this study offers two explanations for why SVI is a risk factor for OUD among older adults. First, areas with higher SVI tend to host more socioeconomically vulnerable groups, which may undermine local opportunity structure, reduce investment in public infrastructure (e.g., transportation), and limit residents’ access to health care services (Phelan et al., 2010). Older adults may thus live in an environment with substandard public services and health care systems, ultimately leading to opioid misuse or OUD. Second, the social vulnerability of an area could be regarded as a context that hinders social integration and the development of social ties. Without a shared belief or tightly knit social network, forming collective efficacy (Sampson et al., 1997) becomes challenging, and a community’s ability to address emerging public health concerns is impeded. The lack of social support or interaction is a risk factor for opioid misuse (Cance et al., 2021); therefore, living in areas with high SVIs may heighten the risk of OUD and a high prevalence of OUD among older adults.

The goal of this study is twofold: (1) to examine how the OUD rate among older adults evolved between 2011 and 2022 by social vulnerability in US counties, and (2) to estimate the relationship between social vulnerability and the OUD rate among older adults. To achieve the goal, this study assembles a unique county-level dataset using 2011–2022 Medicare claims data and CDC’s SVI and then explores the OUD rate among older Medicare beneficiaries by SVI quartiles. The fixed-effects Poisson regression modeling is applied to the data to quantify the effect of SVI on OUD.

Data and Methods

Data Sources

This study sourced data from two government agencies, the Centers for Medicare & Medicaid Services (CMS) and CDC. The CMS 2011–2022 Medicare Beneficiary Summary file base segment was used to determine the beneficiary demographic, enrollment, and hierarchical condition category (HCC) risk score, and the 2011–2022 Medicare Part A and B administrative claims data were used to identify OUD diagnoses. The SVI is sourced by the CDC. This study utilizes SVIs from 2010, 2014, 2016, 2018, 2020, and 2022. Notably, the CDC has calculated SVIs biennially since 2014, resulting in a 4-year gap between 2010 and 2014. Annual Medicare data are matched to the closest preceding SVIs. For instance, Medicare data from 2011 to 2013 are linked to the 2010 SVIs, whereas data from 2014 to 2015 are matched to the 2014 SVIs.

Study Population

This study includes Medicare Fee-for-Service, or Original Medicare, beneficiaries 65 years or older. To be included, beneficiaries must have been enrolled in Medicare Parts A and B for all 12 months and have not been enrolled in the Medicare Advantage program at any point during the year.

Outcome and Measures

The main outcome of this study is the OUD rate. Beneficiaries are considered to have OUD if they have an International Statistical Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM) diagnosis for OUD (F1110, F1111, F11120, F11121x, F1114, F1115x, F1118x, F1119, F112x F1122x, F1123, F1124, F1125x, F1128x, or F1129) in any position on a paid Part A or B service claim during the year. The OUD rate is expressed as the number of beneficiaries with OUD per 1,000 beneficiaries.

The demographic measures include the percentage of female beneficiaries, the percentage of beneficiaries ages 65 to 74, and the percentage of beneficiaries ages 75 to 84. The Social Security Administration collected the beneficiary’s race and ethnicity and sent it to CMS when the beneficiary enrolled in Medicare. The race/ethnicity measures include the percentage of non-Hispanic Black beneficiaries, the percentage of Hispanic beneficiaries, the percentage of Asian/Pacific Islander beneficiaries, and the percentage of other race beneficiaries. The percentage of dually eligible beneficiaries is the percentage of beneficiaries who are eligible for both Medicare and Medicaid. The HCC risk score is calculated using an algorithm by CMS to gauge beneficiaries’ health risk and understand whether certain beneficiaries are less or more costly to treat than the average population. The HCC risk scores are normalized to 1.0. Beneficiaries with scores over 1.0 are more costly than those with scores less than 1.0. The average HCC risk scores are used in the analysis.

With respect to social vulnerability, the CDC developed a composite SVI score consisting of four separate themes: (1) “Socioeconomic Status theme,” (2) “Household Composition and Disability theme,” (3) “Minority Status and Language theme,” and (4) “Housing Type and Transportation theme.” The CDC uses the ranking method (Flanagan et al., 2011) to create the SVI score. Specifically, there are several variables (e.g. poverty) under each theme. Each variable is assigned a percentile ranking value from 0 to 1, with higher values indicating greater vulnerability. The percentile ranking values of all variables are summed within each theme to generate theme-specific percentile rankings. The final SVI scores are the percentile rankings generated with the sum of the four theme-specific percentile rankings values.

Statistical Analysis

The fixed-effects Poisson regression modeling is the main analytic technique in this study, which can be expressed as follows (Allison, 2009):

where $λit$ is the expected number of OUD beneficiaries for county i in year t, $μ t$ is an intercept varying with time, and $β$ represents the vector of coefficient estimates of the relationship between time-varying independent variable $xit$⁠. The total population at risk (i.e., all eligible Medicare beneficiaries in a county) is included in the model as an offset. Each county serves as its own control group under the fixed-effects modeling framework so that the unobserved and time-invariant effects are dropped out of the estimation (Allison, 2009).

The analytic strategy has two phases. First, counties are divided into quartiles using SVIs each year. The average county-level prevalence of OUD among older Medicare beneficiaries is then calculated for each quartile by year. The descriptive findings are summarized into tables and line plots to understand the temporal trend in the OUD rate by SVI quartiles. In the second phase, the fixed-effects Poisson regression modeling is applied to the county-year longitudinal data to estimate the relationship between SVI and the prevalence of OUD.

Results

Table 1 presents the summary statistics of variables used in this study by SVI quartiles over the study period. Three key findings are drawn from this table. First, the OUD rate among older beneficiaries gradually increases with SVI quartiles in that the counties in the first quartile (i.e., the least vulnerable) have the lowest OUD rate (3.450 per 1,000 older beneficiaries), and those in the fourth quartile (i.e., the most vulnerable) have the highest rate (7.404 per 1,000). The differences in the OUD rate across SVI quartiles are statistically significant, providing preliminary support that SVI is positively associated with the OUD rate (results available on request). Second, the age and sex structures across SVI quartiles are largely comparable across SVI quartiles. For example, the percentage of the population aged 65–74 ranges between 52% to 53%, and approximately 55% of beneficiaries are female. However, racial/ethnic composition seems to vary greatly over the SVI quartiles. For example, on average, only 1% of beneficiaries are non-Hispanic Black in the least vulnerable counties, but this number increases to over 7% in the most vulnerable counties. This pattern is also observed for other racial/ethnic minority groups. Third, regarding dual-eligibility and HCC, there is a clear rising gradient from the first to the fourth SVI quartile. For example, almost 8%of total beneficiaries are dually eligible among the least vulnerable counties, compared with almost 20% among the most vulnerable counties.

The OUD rates among older beneficiaries over SVI quartiles are examined by year, and the results are shown in Table 2. The results not only confirm that the OUD rate among older beneficiaries is positively associated with SVI quartiles every year but also demonstrate a rising pattern, especially from 2011 and 2019, for all SVI quartiles. The OUD rate among older beneficiaries dropped in 2020 when the COVID-19 pandemic hit. The last two rows of Table 2 display the differences in OUD rates among older adults and the ratio of OUD rates in the fourth quartile compared to the first quartile for each year. The gap in OUD rate among older beneficiaries between the first and fourth quartile ranges between 0.653 and 5.816 cases per 1,000 older beneficiaries. Moreover, the OUD rate among older beneficiaries in the most vulnerable counties is consistently double that in the least vulnerable counties.

Figure 1 further illustrates the temporal trend in the OUD rate among older beneficiaries by SVI quartiles between 2011 and 2022. The widening gap in OUD rates has become increasingly evident since 2011. Although OUD rates have risen across all four quartiles, the most vulnerable counties have experienced the steepest increase, approximately sixfold (9.853/1.643 = 5.997), between 2011 and 2016. In contrast, the least vulnerable counties have seen a relatively smoother, yet still significant, increase of about fourfold (4.037/0.990 = 4.078) during the same period. The upward trend continued until 2019, followed by a decline in 2020, likely due to the COVID-19 pandemic. However, since 2021, OUD rates among older beneficiaries have begun to rise again. If this trend persists, these rates are likely to surpass prepandemic levels in the near future.

Figure 1.

Trend in the opioid use disorder rate among older beneficiaries by social vulnerability index quartile between 2011 and 2022. Text description: Between 2011 and 2022, the least vulnerable counties (solid line) have the lowest prevalence of opioid use disorder (OUD) among older beneficiaries and the most vulnerable counties (long-dashed line) have the highest prevalence of OUD among older beneficiaries. The gaps across social vulnerability index quartiles had widened between 2011 and 2019 and in 2020, the first year of the COVID-19 pandemic, the gaps narrowed slightly. After 2021, the prevalence of OUD among beneficiaries have increased to the pre-pandemic levels. Notes: COVID-19 = coronavirus disease 2019; SVI = social vulnerability index.

Open in new tabDownload slide

The fixed-effects Poisson modeling results are summarized in Table 3. Model 1 focuses solely on the SVI. The results suggest that the OUD rate increases by approximately 8% for every 0.25 unit increase in SVI ([1.383^0.25-1]×100% = 8.4%). Including age and sex structures in Model 2 slightly accounts for the association between SVI and OUD rate. Specifically, a 0.25 unit increase in SVI is associated with a 7.5% increase in the OUD rate among older beneficiaries ([1.336^0.25-1]×100% = 7.5%), assuming the percentages of female beneficiaries and older beneficiaries aged 65–74 and 75–84 do not vary across counties. Model 3 further incorporates racial/ethnic composition, the percentage of dually eligible beneficiaries, and the average HCC. These variables partially explain why SVI has a detrimental relationship with the OUD rate among older beneficiaries. Model 3 indicates that a 0.25 unit increase in SVI raises the OUD rate by approximately 6% ([1.263^0.25-1]×100% = 6.0%).

Beyond the relationship between SVI and the OUD rate among older beneficiaries, several findings are drawn from Model 3. One is that larger proportions of younger beneficiaries tend to increase the OUD rate in US counties. For example, a 1 percentage point increase in the population aged 65–74 raises the OUD rate among older beneficiaries by 3.5%. Moreover, the concentrations of non-Hispanic Black and Hispanic beneficiaries are negatively related to the OUD rate. When the Hispanic beneficiary population grows by 1 percentage point, the OUD rate among older beneficiaries decreases by 3% ([0.970-1]×100% = -3.0%). Finally, counties with high concentrations of dually eligible beneficiaries or average HCC scores have high OUD rates among older beneficiaries. For example, when the average HCC score increases by 0.1 unit, the OUD rate is expected to increase 37.7% ([24.454^0.1-1]×100% = 37.7%). As the HCC score serves as a proxy for a beneficiary’s health conditions, counties with more beneficiaries who need health care are subject to higher OUD rates.

Conclusions and Discussion

The results above indicate that this study achieved the twofold research goal. First, the OUD rate among older beneficiaries has increased since 2011, when baby boomers began to enter older adulthood. The upward trajectory is observed for all SVI quartiles; nonetheless, the most vulnerable counties have experienced a more rapid increase in OUD rates than other counties between 2011 and 2022. The gaps in OUD rates among SVI quartiles have widened, especially before the COVID-19 pandemic. Although the OUD rates decreased in 2020, the first year of the pandemic, the OUD rate among older beneficiaries has gradually risen to the prepandemic level. Second, using the fixed-effects Poisson regression, this study shows that SVI is positively associated with the OUD rate, even after controlling for the sociodemographic and health conditions of beneficiaries in a county. According to Model 3, the OUD rate among older beneficiaries in the top 10% of most vulnerable counties is at least 20% ([1.263^0.8-1]×100% = 20.5%) higher than in the bottom 10% of the least vulnerable counties.

"[T]he OUD rate among older beneficiaries has increased since 2011, when baby boomers began to enter older adulthood. The upward trajectory is observed for all SVI quartiles; nonetheless, the most vulnerable counties have experienced a more rapid increase in OUD rates than other counties between 2011 and 2022. The gaps in OUD rates among SVI quartiles have widened, especially before the COVID-19 pandemic."

"[T]he OUD rate among older beneficiaries in the top 10% of most vulnerable counties is at least 20%... higher than in the bottom 10% of the least vulnerable counties."

Two explanations exist for the drop in the OUD rate among older beneficiaries in 2020. In the early stage of the pandemic, older beneficiaries susceptible to OUD were also at a higher risk of COVID-19 infection or death (Miller, 2020; Shahid et al., 2020), which may have prevented older beneficiaries from developing OUD. Moreover, the fear of infection and looming shortages of medical resources could discourage older beneficiaries from seeking health care or interrupt their regular doctor visits (Schuster et al., 2021). Consequently, older beneficiaries are less likely to be diagnosed with OUD by primary health care providers, ultimately lowering the OUD rate in 2020. Since March 2020, CMS has relaxed regulations to give health care providers maximum flexibility in offering telehealth services to Medicare beneficiaries. This change has improved the treatment and related outcomes for beneficiaries with OUD (Jones et al., 2022). However, there are well-documented barriers to the use of telehealth among older beneficiaries, such as a lack of technical literacy, willingness, and cost (Kruse et al., 2020), which may undermine the receipt of health care or timely assistance.

The positive relationships between younger age structures (e.g., percentage of beneficiaries aged 65–74) and the OUD rate among older beneficiaries in US counties can be understood as follows. The youngest old population may still be adjusting to new challenges in older adulthood, such as retirement and the loss of social roles (Waite & Das, 2010). These challenges can increase the risk of substance misuse and OUD. In addition, the youngest older population tends to be more physically active than the middle or oldest older populations. Physical activities can lead to injuries that are often managed with chronic opioid prescriptions, ultimately raising the risk of OUD.

The findings of this study should be interpreted with the following caveats. First, the county is the unit of analysis. Using a different administrative unit, such as ZIP codes, census tracts, or health service areas, may alter our conclusions, a limitation common to all ecological studies (Fotheringham & Wong, 1991). Second, the resurgence of OUD among older beneficiaries since 2021 may be either facilitated or mitigated when younger baby boomers enter older adulthood. Although our findings show an upward trajectory for all SVI quartiles, future research should revisit this temporal trend with new data. Third, fixed-effects Poisson modeling takes into account the unobserved time-invariant effects and yields robust estimates (Allison, 2009); however, this approach is specific to the observations in our analysis and cannot be generalized to other settings.

Despite the limitations, several policy implications emerge from the findings. First, whereas PDMPs aim to improve clinicians’ opioid prescribing practices, research indicates that opioid dispensing rates are higher in counties with greater social vulnerability (Bounthavong & Yip, 2024). In light of our findings, we recommend more educational programs for patients and providers in these vulnerable counties, focusing on proper medication adherence, nonopioid treatments, and recognizing symptoms of OUD. These educational initiatives align with the first two recommendations of the CDC’s clinical practice guideline for prescribing opioids for pain. They could help reduce OUD disparities among older adults across the SVI quartiles. Moreover, increasing the availability of medications for opioid use disorder (MOUD) for older adults with OUD could mitigate disparities along the SVI spectrum. The Consolidated Appropriations Act of 2023 removed the federal requirement for a special waiver to prescribe buprenorphine, but states are not mandated to adopt these changes. Therefore, states should ensure MOUD accessibility for those in need. Finally, our finding that higher percentages of young-older (age 65–74) and middle-older (age 75–84) populations are associated with higher OUD rates indicates that the SBIRT services may be more cost-effective if providers encourage their use among older adults under 85.

In sum, this study is among the first to investigate how the OUD rate among older beneficiaries has evolved since baby boomers first entered older adulthood in 2011 by SVI quartiles in US counties and to quantify the relationship between SVI and the OUD rate in the geriatric population from a longitudinal perspective. With the increasing number of older adults experiencing OUD, especially among those living in our most vulnerable counties, efforts need to be made to ensure that they are getting the care that they need. Research has shown that the utilization of MOUDs among the entire Medicare FFS population with OUD is very low (less than 13%) (Jones et al., 2022). Given the unique perspective of the baby boomers, the utilization of MOUD among 65 and older beneficiaries could be even lower. As previously mentioned, CMS has relaxed regulations to provide health care providers with maximum flexibility in offering telehealth services to Medicare beneficiaries, improving the treatment and related outcomes for beneficiaries with OUD (Jones et al., 2022). However, older adults may not be as comfortable using telehealth (Li et al., 2025). More efforts need to be made to ensure older adults with OUD are receiving MOUD, especially those living in the most vulnerable counties. It is also critical to educate medical providers about the rise in OUD among this vulnerable population and what they can do to treat these patients. Continuing education courses would be a great way to educate providers about OUD among older adults, how to best treat older adults with OUD, and how to prevent OUD from occurring.

Funding

None.

Conflict of Interest

None.

References