[Skip to Content]
Sign In
Individual Sign In
Create an Account
Institutional Sign In
OpenAthens Shibboleth
[Skip to Content Landing]
Table 1.  
Comparison of Demographic Characteristics and Medical Conditions Between All Medicare Enrollees Aged 21 to 64 Years Qualified for Disability in 2016 and Those Who Died of Opioid Overdose
Comparison of Demographic Characteristics and Medical Conditions Between All Medicare Enrollees Aged 21 to 64 Years Qualified for Disability in 2016 and Those Who Died of Opioid Overdose
Table 2.  
Demographic Characteristics and Medical Conditions Associated With Opioid Overdose Deaths in Medicare Enrollees Aged 21 to 64 Years Qualified for Disability in 2016a
Demographic Characteristics and Medical Conditions Associated With Opioid Overdose Deaths in Medicare Enrollees Aged 21 to 64 Years Qualified for Disability in 2016a
Table 3.  
Proportion of Enrollees and the Rate of Opioid Overdose Deaths per 100 000 in 2016 Among Medicare Beneficiaries Aged 21 to 64 Years Who Qualified for Disability With 3 Major Condition Categories: Psychiatric Diseases, Substance Abuse, and Chronic Pain
Proportion of Enrollees and the Rate of Opioid Overdose Deaths per 100 000 in 2016 Among Medicare Beneficiaries Aged 21 to 64 Years Who Qualified for Disability With 3 Major Condition Categories: Psychiatric Diseases, Substance Abuse, and Chronic Pain
1.
Song  Z.  Mortality quadrupled among opioid-driven hospitalizations, notably within lower-income and disabled white populations.  Health Aff (Millwood). 2017;36(12):2054-2061. doi:10.1377/hlthaff.2017.0689PubMedGoogle ScholarCrossref
2.
Peters  JL, Durand  WM, Monteiro  KA, Dumenco  L, George  P.  Opioid overdose hospitalizations among Medicare-disability beneficiaries.  J Am Board Fam Med. 2018;31(6):881-896. doi:10.3122/jabfm.2018.06.180152PubMedGoogle ScholarCrossref
3.
Meara  E, Horwitz  JR, Powell  W,  et al.  State legal restrictions and prescription-opioid use among disabled adults.  N Engl J Med. 2016;375(1):44-53. doi:10.1056/NEJMsa1514387PubMedGoogle ScholarCrossref
4.
Equator Network. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) Statement: guidelines for reporting observational studies. http://www.equator-network.org/reporting-guidelines/strobe/. Updated November 15, 2018. Accessed August 31, 2019.
5.
Mues  KE, Liede  A, Liu  J,  et al.  Use of the Medicare database in epidemiologic and health services research: a valuable source of real-world evidence on the older and disabled populations in the US.  Clin Epidemiol. 2017;9(9):267-277. doi:10.2147/CLEP.S105613PubMedGoogle ScholarCrossref
6.
Filice  CE, Joynt  KE.  Examining race and ethnicity information in Medicare administrative data.  Med Care. 2017;55(12):e170-e176. doi:10.1097/MLR.0000000000000608PubMedGoogle ScholarCrossref
7.
US Department of Agriculture. Rural-urban continuum codes. https://www.ers.usda.gov/data-products/rural-urban-continuum-codes/. Updated October 12, 2016. Accessed July 14, 2017.
8.
Centers for Medicare & Medicaid Services. Chronic Condition Data Warehouse: chronic condition categories. https://www2.ccwdata.org/web/guest/condition-categories. Accessed August 31, 2019.
9.
May  WL, Johnson  WD.  A SAS macro for constructing simultaneous confidence intervals for multinomial proportions.  Comput Methods Programs Biomed. 1997;53(3):153-162. doi:10.1016/S0169-2607(97)01809-9PubMedGoogle ScholarCrossref
10.
Scholl  L, Seth  P, Kariisa  M, Wilson  N, Baldwin  G.  Drug and opioid-involved overdose deaths: United States, 2013–2017.  MMWR Morb Mortal Wkly Rep. 2018;67(5152):1419-1427. doi:10.15585/mmwr.mm675152e1PubMedGoogle ScholarCrossref
11.
Jeffery  MM, Hooten  WM, Henk  HJ,  et al.  Trends in opioid use in commercially insured and Medicare Advantage populations in 2007-16: retrospective cohort study.  BMJ. 2018;362:k2833. doi:10.1136/bmj.k2833PubMedGoogle ScholarCrossref
12.
Zhou  C, Yu  NN, Losby  JL.  The association between local economic conditions and opioid prescriptions among disabled Medicare beneficiaries.  Med Care. 2018;56(1):62-68. doi:10.1097/MLR.0000000000000841PubMedGoogle ScholarCrossref
13.
Davis  MA, Lin  LA, Liu  H, Sites  BD.  Prescription opioid use among adults with mental health disorders in the United States.  J Am Board Fam Med. 2017;30(4):407-417. doi:10.3122/jabfm.2017.04.170112PubMedGoogle ScholarCrossref
14.
Deyo  RA, Von Korff  M, Duhrkoop  D.  Opioids for low back pain.  BMJ. 2015;350:g6380. doi:10.1136/bmj.g6380PubMedGoogle ScholarCrossref
15.
Social Security Administration. Annual statistical report on the Social Security Disability Insurance Program, 2016: beneficiaries in current-payment status. https://www.ssa.gov/policy/docs/statcomps/di_asr/2016/index.html. Published October 2017. Accessed March 25, 2019.
16.
Moyo  P, Zhao  X, Thorpe  CT,  et al.  Dual receipt of prescription opioids from the Department of Veterans Affairs and Medicare Part D and prescription opioid overdose death among veterans: a nested case-control study.  Ann Intern Med. 2019;170(7):433-442. doi:10.7326/M18-2574PubMedGoogle ScholarCrossref
17.
Guy  GP  Jr, Zhang  K, Bohm  MK,  et al.  Vital signs: changes in opioid prescribing in the United States, 2006-2015.  MMWR Morb Mortal Wkly Rep. 2017;66(26):697-704. doi:10.15585/mmwr.mm6626a4PubMedGoogle ScholarCrossref
18.
Goodwin  JS, Kuo  YF, Brown  D, Juurlink  D, Raji  M.  Association of chronic opioid use with presidential voting patterns in US counties in 2016.  JAMA Netw Open. 2018;1(2):e180450. doi:10.1001/jamanetworkopen.2018.0450PubMedGoogle Scholar
19.
Han  B, Compton  WM, Blanco  C, Crane  E, Lee  J, Jones  CM.  Prescription opioid use, misuse, and use disorders in US adults: 2015 National Survey on Drug Use and Health.  Ann Intern Med. 2017;167(5):293-301. doi:10.7326/M17-0865PubMedGoogle ScholarCrossref
20.
Liaw  V, Kuo  YF, Raji  MA, Baillargeon  J.  Opioid prescribing in disabled adults in the United Sates following the 2014 federal hydrocodone rescheduling regulation.  Public Health Rep. In press.Google Scholar
21.
Brady  JE, Giglio  R, Keyes  KM, DiMaggio  C, Li  G.  Risk markers for fatal and non-fatal prescription drug overdose: a meta-analysis.  Inj Epidemiol. 2017;4(1):24. doi:10.1186/s40621-017-0118-7PubMedGoogle ScholarCrossref
22.
Ruhm  CJ.  Corrected US opioid-involved drug poisoning deaths and mortality rates, 1999-2015.  Addiction. 2018;113(7):1339-1344. doi:10.1111/add.14144PubMedGoogle ScholarCrossref
23.
Harrell  FE.  Regression Modeling Strategies: With Applications to Linear Models, Logistic Regression, and Survival Analysis. New York, NY: Springer; 2001. doi:10.1007/978-1-4757-3462-1
24.
National Institute on Disability, Independent Living, and Rehabilitation Research. Summary of responses from a request for information: people with disabilities and opioid use disorder. https://acl.gov/sites/default/files/news%202018-05/20180502NIDILRROpioidRFIFindings.pdf. Updated September 24, 2018. Accessed May 15, 2019.
25.
Lauer  EA, Henly  M, Brucker  DL.  Prescription opioid behaviors among adults with and without disabilities: United States, 2015-2016.  Disabil Health J. 2018;12(3):519-522. doi:10.1016/j.dhjo.2018.12.001PubMedGoogle ScholarCrossref
Limit 200 characters
Limit 25 characters
Conflicts of Interest Disclosure

Identify all potential conflicts of interest that might be relevant to your comment.

Conflicts of interest comprise financial interests, activities, and relationships within the past 3 years including but not limited to employment, affiliation, grants or funding, consultancies, honoraria or payment, speaker's bureaus, stock ownership or options, expert testimony, royalties, donation of medical equipment, or patents planned, pending, or issued.

Err on the side of full disclosure.

If you have no conflicts of interest, check "No potential conflicts of interest" in the box below. The information will be posted with your response.

Not all submitted comments are published. Please see our commenting policy for details.

Limit 140 characters
Limit 3600 characters or approximately 600 words
    Original Investigation
    Substance Use and Addiction
    November 15, 2019

    Association of Disability With Mortality From Opioid Overdose Among US Medicare Adults

    Author Affiliations
    • 1Sealy Center on Aging, Department of Internal Medicine, The University of Texas Medical Branch at Galveston, Galveston
    • 2Department of Preventive Medicine and Population Health, The University of Texas Medical Branch at Galveston, Galveston
    • 3Institute for Translational Science, The University of Texas Medical Branch at Galveston, Galveston
    JAMA Netw Open. 2019;2(11):e1915638. doi:10.1001/jamanetworkopen.2019.15638
    Key Points español 中文 (chinese)

    Question  What is the rate of opioid overdose deaths among Medicare enrollees younger than 65 years who qualified for Medicare because of disability?

    Findings  In this cohort study, the 1 766 790 adult Medicare enrollees who qualified for disability—representing 14.9% of the Medicare population—accounted for 80.8% of all opioid overdose deaths among all Medicare enrollees. Among the 11.1% of the enrollees with disability who had 3 co-occurring conditions of substance abuse, psychiatric diseases, and chronic pain syndrome, the opioid overdose death rate was 23.4 times higher than that for enrollees without any of the conditions.

    Meaning  Understanding the heterogeneity of medical and psychiatric conditions associated with opioid use and misuse is key to developing specific, data-driven interventions targeted for each subgroup of high-risk populations.

    Abstract

    Importance  Patients qualifying for Medicare disability have the highest rates of opioid use compared with older Medicare beneficiaries and commercial insurance beneficiaries. Research on opioid overdose deaths in this population can help identify appropriate interventions.

    Objective  To assess the rate of opioid overdose death and to identify its associated risk factors.

    Design, Setting, and Participants  This cohort study included a 20% national sample of Medicare enrollees aged 21 to 64 years whose initial Medicare entitlement was based on disability and who resided in 50 US states and Washington, DC, in 2012 to 2016. Data analyses were performed from March 15, 2019, through September 23, 2019.

    Exposures  Fifty-five chronic or potentially disabling conditions were selected from the Centers for Medicare & Medicaid Services Chronic Disease Data Warehouse.

    Main Outcomes and Measures  Opioid overdose death rate estimated from Medicare National Death Index linkage data.

    Results  Among 1 766 790 Medicare enrollees younger than 65 years who qualified for Medicare because of disability, the mean (SD) age was 52.2 (10.2) years, and 866 914 (49.1%) were women. These enrollees represent 14.9% (95% CI, 14.9%-15.0%) of the entire Medicare population and accounted for 80.8% (95% CI, 78.9%-82.7%) of opioid overdose deaths among all Medicare enrollees. Opioid overdose mortality in this population increased from 57.4 per 100 000 (95% CI, 53.9-61.0 per 100 000) in 2012 to 77.6 per 100 000 (95% CI, 73.5-81.8 per 100 000) in 2016. Results from the stepwise logistic regression model revealed that 3 categories of conditions are associated with opioid overdose death: substance abuse, psychiatric diseases, and chronic pain. Among the 11.1% (95% CI, 11.0%-11.2%) of adults with disability who had all 3 conditions, the rate of opioid overdose death was 363.7 per 100 000 (95% CI, 326.7-402.6 per 100 000), which is 23.4 times higher than the rate for individuals with none of the conditions (15.5 per 100 000; 95% CI, 11.6-20.1 per 100 000).

    Conclusions and Relevance  This study identifies differences in opioid overdose mortality among subgroups of Medicare enrollees younger than 65 years who qualify for Medicare because of disability. Understanding the heterogeneity of medical and psychiatric conditions associated with opioid use and misuse is key to developing specific, data-driven interventions targeted to each subgroup of high-risk populations.

    Introduction

    Medicare beneficiaries who qualify for Medicare because of disability constitute a growing population of patients hospitalized for opioid or heroin overdose1,2 and account for 25% of deaths from prescription opioid overdose annually.3 The Centers for Disease Control and Prevention (CDC) regularly generates reports of opioid overdose deaths by demographic variables and states, but studies on policy-actionable risk factors (eg, clusters of medical and psychiatric conditions or types of disabling conditions) for overdose mortality are lacking in this population.

    Methods

    The University of Texas Medical Branch institutional review board approved this study and waived any informed consent requirement because the research used deidentified data. This study follows the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.4

    Our initial cohort was a 20% national sample of all Medicare enrollees aged 21 to 100 years residing in 50 US states and Washington, DC, in 2012 to 2016. Centers for Medicare & Medicaid Services (CMS) selects a random sample of 20% Medicare beneficiaries according to the last 2 digits of their health insurance claim number. Studies based on these data should provide generalizable estimates for the Medicare population.5 We used Medicare summary files linked to the National Death Index. From the National Death Index, we identified drug overdose deaths according to International Statistical Classification of Diseases and Related Health Problems, Tenth Revision, codes for cause of death: X40-44, X60-64, X85, or Y10-Y14. Opioid overdose deaths were identified by codes T40.0 (opium), T40.1 (heroin), T40.2 (natural or semisynthetic opioids), T40.3 (methadone), T40.4 (synthetic opioids other than methadone), or T40.6 (other and unspecified narcotics) from the underlying conditions.

    Patient characteristics included age, sex, race/ethnicity, years under Medicare disability entitlement, health maintenance organization enrollment, and Medicare Part D enrollment for at least 1 month in a year. Race/ethnicity is based on Social Security Administration data, with a surname algorithm enhancing the identification of Hispanic and Asian origin.6 Medicaid enrollment was used to represent low income. Education and household income were obtained from the 2011 to 2015 American Community Survey by zip code linkage and were divided into quartiles. Residential area was classified using rural-urban continuum codes.7

    The CMS Chronic Conditions Data Warehouse listed 27 chronic conditions and 36 potentially disabling conditions as of March 2019. Chronic Conditions Data Warehouse is a research database designed to improve the quality of care and reduce costs and utilization. The specific definition and reference for each condition are available at the Chronic Conditions Data Warehouse website.8 We combined these 2 sets of conditions, removed the duplicate conditions (eg, depression and depressive disorder), and consolidated the cancer diagnoses (breast, lung, colorectal, endometrial, prostate cancer, or leukemia and lymphomas). This resulted in 55 chronic or potentially disabling conditions for our study (Table 1).

    Statistical Analysis

    We calculated the number of deaths from opioid overdose among all Medicare recipients aged 21 to 100 years and the percentage of opioid overdose deaths by age and by reason for initial entitlement (age, disability, or end-stage renal disease) in 2016. Confidence intervals of these percentages were calculated according to multinomial distribution.9 Our main analyses focused on enrollees aged 21 to 64 years who initially qualified for Medicare because of disability. We first calculated the rate of opioid overdose deaths from 2012 to 216. Then we restricted the analyses to the 2016 data to compare their characteristics with those who died from opioid overdose, in bivariate analyses using goodness-of-fit test, and reported P values with Bonferroni correction for the multiple comparisons. For multivariable analyses, only those with Medicare Part A and B continuous enrollment and without health maintenance organization coverage in 2015 and 2016 were included and analyzed, by a stepwise logistic regression model, to identify conditions and characteristics associated with opioid overdose deaths. Thirteen conditions associated with fewer than 11 deaths from opioid overdose were not included because of the CMS cell size suppression policy to protect the confidentiality of enrollees. No obvious multicollinearity was found, with all correlations among characteristics and study conditions being less than 0.38. On the basis of the results of stepwise logistic regression, we grouped the conditions that were positively associated with opioid overdose death into 3 major common categories of conditions: substance abuse, psychiatric disease, and chronic pain. We analyzed the overlap across these 3 categories to show the proportion of enrollees and opioid overdose mortality for their combinations. The 95% CI of opioid overdose mortality was calculated assuming a Poisson distribution. In addition, we repeated analyses using the 2012 to 2015 data for 2 purposes: to validate our final model in the 2016 data and to expand the conditions associated with fewer than 11 deaths in the 2016 analyses. To compare our results with the CDC report, we calculated age-adjusted rates using the direct method and the 2000 US standard population aged 20 to 64 years. All tests of statistical significance were 2-sided with significance set at P < .05. Analyses were performed with SAS Enterprise statistical software version 7.12 (SAS Inc). Data analyses were performed from March 15, 2019, through September 23, 2019.

    Results

    In 2016, 1 766 790 Medicare enrollees in the 20% sample were aged 21 to 64 years and received their initial Medicare entitlement because of disability. The mean (SD) age of this cohort was 52.2 (10.2) years; 866 914 (49.1%) were women, and 899 876 (50.9%) were men. This 14.9% (95% CI, 14.9%-15.0%) of all enrollees accounted for 80.8% (95% CI, 78.9%-82.7%) of opioid overdose deaths among all Medicare enrollees. The remaining analyses focus on opioid overdose deaths in this Medicare population younger than 65 years who qualified for disability. The opioid overdose mortality rate in this population increased from 57.4 per 100 000 (95% CI, 53.9-61.0 per 100 000) in 2012 to 77.6 per 100 000 (95% CI, 73.5-81.8 per 100 000) in 2016.

    Table 1 shows the characteristics and conditions for Medicare enrollees aged 21 to 64 years with a disability entitlement and also for the subset who died because of opioid overdose. In general, those who died from opioid overdose were more likely than the study population to be younger (mean [SD] age, 49.7 [9.9] years vs 52.2 [10.2] years), male (773 [56.4%] vs 899 876 [50.9%]), white (1114 [81.3%] vs 1 170 832 [66.3%]), and eligible for Medicaid (796 [58.1%] vs 866 544 [49.0%]). A larger proportion of these individuals also had psychiatric diseases (depression, 436 [56.8%] vs 264 809 [30.2%]; anxiety, 487 [63.4%] vs 279 297 [30.9%]; and bipolar disorder, 273 [35.6%] vs 122 830 [14.0%]), substance use disorder (tobacco use disorder, 437 [56.9%] vs 207 378 [23.7%]; alcohol use disorder, 186 [24.2%] vs 58 582 [6.7%]; and drug use disorder, 432 [56.3%] vs 105 625 [12.1%]), and chronic pain (489 [63.7%] vs 271 913 [31.1%]). Table 2 presents the results of a stepwise logistic regression model estimating opioid overdose deaths in 2016 among Medicare enrollees younger than 65 years who qualified for disability. In the final models, we found that being male and white, having higher income, coverage with Medicare Part D, enrollment under disability entitlement less than 15 years, and living in metropolitan areas were associated with higher rates of opioid overdose death. Chronic conditions associated with a higher rate of opioid overdose death were associated with substance abuse, including opioid use disorder (adjusted odds ratio [aOR], 3.23; 95% CI, 2.56-4.06), drug use disorder (aOR, 1.86; 95% CI, 1.47-2.37), and tobacco use disorder (aOR, 1.77; 95% CI, 1.49-2.09); psychiatric diseases, including anxiety (aOR, 1.65; 95% CI, 1.37-1.97), bipolar disorder (aOR, 1.51; 95% CI, 1.28-1.79), and depression (aOR, 1.29; 95% CI, 1.09-1.53); and chronic pain (aOR, 1.86; 95% CI 1.57-2.21). Chronic kidney disease (aOR, 1.59; 95% CI, 1.32-1.91), pressure and chronic ulcers (aOR, 1.48; 95% CI, 1.11-1.96), and hepatitis (aOR, 1.68; 95% CI, 1.37-2.05) were also associated with a greater likelihood of opioid overdose death. The C-statistic for the model was 0.863.

    Table 3 shows the opioid overdose mortality rates for the 3 major categories of conditions: substance abuse, psychiatric diseases, and chronic pain. The logistic model including only these conditions with no demographic variables had a C-statistic of 0.82. Among the 37.5% (95% CI, 37.4%-37.7%) of beneficiaries younger than 65 years with Medicare disability entitlement who did not have any of the 3 categories of conditions, the opioid overdose mortality rate in 2016 was 15.5 per 100 000 (95% CI, 11.6-20.1 per 100 000). In contrast, the 11.1% (95% CI, 11.0%-11.2%) of beneficiaries who had all 3 categories of conditions had an opioid overdose mortality rate of 363.7 per 100 000 (95% CI, 326.7-402.6 per 100 000), which is 23.4 times higher than the rate for enrollees without any of the conditions and accounts for 45.8% of all opioid overdose deaths. The 8.1% (95% CI, 8.0%-8.2%) of beneficiaries with substance abuse plus psychiatric diseases without a pain diagnosis had an opioid overdose mortality rate of 187.5 per 100 000 (95% CI, 157.0-220.7 per 100 000), and the 9.3% (95% CI, 9.2%-9.4%) of beneficiaries with pain plus psychiatric diseases without a substance abuse diagnosis had an opioid overdose mortality of 84.6 per 100 000 (95% CI, 65.9-105.7 per 100 000) (Table 3). Overall, 78.0% of all opioid overdose deaths occurred in the 32.0% of enrollees qualified for disability who had at least 2 major condition categories.

    To explore the robustness of study results from 2016, we repeated analyses with person-year data from 2012 to 2015. Seven additional conditions (intellectual disability, multiple sclerosis, spina bifida, blindness, traumatic brain injury, acute myocardial infarction, and glaucoma) excluded in the 2016 analysis because of CMS cell size suppression were added in the multivariable analyses because they were associated with 11 or more opioid overdose deaths in the 2012 to 2015 data. The final stepwise logistic regression model from this sensitivity analyses had C-statistics of 0.857 compared with 0.863 in the main analyses. Aside from pressure and chronic ulcers, positive associations with opioid overdose deaths for all other conditions from our main analyses persisted. Three additional conditions—ischemic heart disease, arthritis, and alcohol use disorder—were found to be positively associated with outcomes in the final model of sensitivity analyses.

    We compared our results with the CDC report on opioid overdose deaths,10 which reported an age-adjusted rate of opioid overdose deaths in the US population in 2016 of 13.3 per 100 000. Assuming that the rate at age 15 to 24 years was the same as that for age 20 to 24 years, the age-adjusted rate was 20.8 per 100 000 for those aged 20 to 64 years in the United States. In contrast, the age-adjusted rate was 99.9 per 100 000 in our study of Medicare enrollees aged 21 to 64 years who qualified for disability, which is at least 4.8 times higher than that of the general US population in 2016.10 On the basis of our findings of 1371 opioid overdose deaths in this population from the 20% Medicare sample, we estimated that approximately 16% of the 42 249 opioid overdose deaths in 2016 reported by the CDC occurred in this population.

    Discussion

    The finding of increased opioid overdose deaths among adults who qualified for Medicare because of disability is consistent with the CDC report of increased mortality across multiple demographic groups.10 The mortality rate in our study is 4.8 times higher than that of the general US population in 2016.10 The 37.5% of the study population with no diagnosis of psychiatric diseases, chronic pain, or substance abuse had an age-adjusted rate of opioid overdose death of 15.5 per 100 000, which is lower than that of the US population.10 The rate of overdose death greatly increased for patients with any of these 3 conditions, with the highest rate among those with all 3 condition categories.

    Previous studies3,11 have reported high rates of prescription opioid use among Medicare adults who qualified for disability. In addition to contextual characteristics, such as county income, income disparity, and unemployment rates,12 other risk factors associated with higher opioid use in this population were mental illness and chronic pain.13,14 The proportion of adult Medicare enrollees with disability entitlement who had a chronic pain diagnosis in our study population (31.1%) was consistent with that of a previous report.15 Our finding of an association of Medicare Part D enrollment with opioid overdose death is also consistent with the results from a Veterans Affairs study,16 presumably attributable to increased access to prescription opioids.

    In the multivariable model, dual-eligible enrollment with Medicaid was not associated with opioid overdose death. We found an unexpected positive association between opioid overdose death and residence in the higher-income zip codes. There is a negative association between income and opioid prescriptions at the county level.17,18 At the individual level, a study19 based on data from the National Survey on Drug Use and Health also showed that those with lower income were more likely to misuse opioids and had higher rates of opioid use disorder than the general US population. Another study20 of the Medicare population showed that enrollees residing in higher-income zip codes had lower rates of long-term opioid prescriptions. It is unclear why higher rates of opioid overdose death were associated with higher income in our study, despite the previously reported lower rate of prescription opioid use in this population. Our findings of higher rates of opioid overdose death in metropolitan areas add additional evidence to the conflict surrounding this topic.21 Future studies assessing the association of overdose mortality with income and urban or rural residence stratified by opioid type may give more insight to our unexpected findings.

    Subgroups of adults enrolled in Medicare because of disability with different risk profiles for opioid overdose death might benefit from different interventions. For example, the subset of beneficiaries with the lowest rate of opioid overdose deaths likely represents a population for whom prescription opioids can be safely offered by their physicians as part of comprehensive pain management if the clinical condition warrants opioid analgesics. For those at the highest risk of opioid overdose deaths, examples of targeted interventions include case management developed for and tailored to a high-risk population and Medicare-mandated requirement for comanagement by pain, addiction, and mental health specialists in treating pain in these populations. These interventions would have the biggest impact on the subgroups with at least 2 major condition categories that comprised 32.0% of all enrollees who qualified for disability but accounted for 78.0% of all opioid overdose deaths in this population. Patient-informed and evidence-supported comanagement strategies and targeted interventions (including enhanced access to substance use disorder treatment) have the potential to reduce opioid overdose deaths in this high-risk population while ensuring effective pain management.

    Limitations

    This study has limitations. First, the quality and accuracy of death certificate data associated with overdose varies across states.10,22 In addition, we could not distinguish from our data between deaths deemed accidental, suicide, or homicide or between the deaths that occurred in the inpatient or outpatient setting. Second, we restricted the analyses of associations with medical conditions to enrollees with 2 years of continuous enrollment with fee-for-service coverage. Our results may not be generalizable to health maintenance organization populations. Third, the validity of medical conditions derived from claims data varies. Fourth, we provided descriptive results to show the differences in rates of opioid overdose death across subgroups. We did not conduct analyses to incorporate competing causes of death. Fifth, we did not analyze the association of drug interactions or contaminated street drugs with opioid overdose mortality. Future studies examining the use of pain medications, psychiatric medications, and the 2 in combination is important for understanding why the opioid overdose death rate is much higher in this population. Sixth, we used stepwise regression to find the most important patient characteristics and health conditions associated with opioid overdose mortality in a parsimonious model, with the understanding of the concerns for the bias in parameter and SE estimation.23 However, our findings in isolating the 3 major condition categories were robust in our analyses of 2012 to 2015 data.

    Conclusions

    Despite the high rates of opioid use in the Medicare population entitled for disability, research24,25 shows lower utilization of opioid treatment programs. This population is heterogeneous, with physical and cognitive disorders present at birth along with conditions acquired later in life. The conditions most significantly associated with opioid overdose deaths in our analyses most commonly occur or are recognized in adolescence or later. There is currently a major federal effort to increase access to opioid use disorder treatment programs and to promote knowledge and use of opioid antagonists in cases of suspected overdose. Such programs work best with accurate targeting of populations at high risk. Our findings suggest that straightforward analyses of Medicare data can identify subgroups of Medicare enrollees with the highest rates of death from opioid overdose. Further studies are needed to understand the heterogeneity of medical and psychiatric conditions associated with opioid use, misuse, addiction, and overdose in this population. Such understanding is key to developing specific, data-driven interventions (eg, programs to manage co-occurring substance use disorder and requirement for comanagement by pain, mental health, and addiction specialists) targeted for each subgroup of high-risk populations.

    Back to top
    Article Information

    Accepted for Publication: September 26, 2019.

    Published: November 15, 2019. doi:10.1001/jamanetworkopen.2019.15638

    Open Access: This is an open access article distributed under the terms of the CC-BY License. © 2019 Kuo Y-F et al. JAMA Network Open.

    Corresponding Author: Yong-Fang Kuo, PhD, Department of Preventive Medicine and Population Health, The University of Texas Medical Branch at Galveston, 301 University Blvd, Galveston, TX 77555 (yokuo@utmb.edu).

    Author Contributions: Dr Kuo had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

    Concept and design: All authors.

    Acquisition, analysis, or interpretation of data: All authors.

    Drafting of the manuscript: All authors.

    Critical revision of the manuscript for important intellectual content: Kuo, Raji.

    Statistical analysis: Kuo, Goodwin.

    Obtained funding: Kuo, Raji.

    Administrative, technical, or material support: Kuo, Raji.

    Conflict of Interest Disclosures: Dr Kuo reported receiving grants from the Agency for Healthcare Research and Quality outside the submitted work. No other disclosures were reported.

    Funding/Support: This work was supported by grants R01-DA039192 and UL1TR001439 from the National Institute on Drug Abuse, National Institutes of Health.

    Role of the Funder/Sponsor: The National Institutes of Health had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

    References
    1.
    Song  Z.  Mortality quadrupled among opioid-driven hospitalizations, notably within lower-income and disabled white populations.  Health Aff (Millwood). 2017;36(12):2054-2061. doi:10.1377/hlthaff.2017.0689PubMedGoogle ScholarCrossref
    2.
    Peters  JL, Durand  WM, Monteiro  KA, Dumenco  L, George  P.  Opioid overdose hospitalizations among Medicare-disability beneficiaries.  J Am Board Fam Med. 2018;31(6):881-896. doi:10.3122/jabfm.2018.06.180152PubMedGoogle ScholarCrossref
    3.
    Meara  E, Horwitz  JR, Powell  W,  et al.  State legal restrictions and prescription-opioid use among disabled adults.  N Engl J Med. 2016;375(1):44-53. doi:10.1056/NEJMsa1514387PubMedGoogle ScholarCrossref
    4.
    Equator Network. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) Statement: guidelines for reporting observational studies. http://www.equator-network.org/reporting-guidelines/strobe/. Updated November 15, 2018. Accessed August 31, 2019.
    5.
    Mues  KE, Liede  A, Liu  J,  et al.  Use of the Medicare database in epidemiologic and health services research: a valuable source of real-world evidence on the older and disabled populations in the US.  Clin Epidemiol. 2017;9(9):267-277. doi:10.2147/CLEP.S105613PubMedGoogle ScholarCrossref
    6.
    Filice  CE, Joynt  KE.  Examining race and ethnicity information in Medicare administrative data.  Med Care. 2017;55(12):e170-e176. doi:10.1097/MLR.0000000000000608PubMedGoogle ScholarCrossref
    7.
    US Department of Agriculture. Rural-urban continuum codes. https://www.ers.usda.gov/data-products/rural-urban-continuum-codes/. Updated October 12, 2016. Accessed July 14, 2017.
    8.
    Centers for Medicare & Medicaid Services. Chronic Condition Data Warehouse: chronic condition categories. https://www2.ccwdata.org/web/guest/condition-categories. Accessed August 31, 2019.
    9.
    May  WL, Johnson  WD.  A SAS macro for constructing simultaneous confidence intervals for multinomial proportions.  Comput Methods Programs Biomed. 1997;53(3):153-162. doi:10.1016/S0169-2607(97)01809-9PubMedGoogle ScholarCrossref
    10.
    Scholl  L, Seth  P, Kariisa  M, Wilson  N, Baldwin  G.  Drug and opioid-involved overdose deaths: United States, 2013–2017.  MMWR Morb Mortal Wkly Rep. 2018;67(5152):1419-1427. doi:10.15585/mmwr.mm675152e1PubMedGoogle ScholarCrossref
    11.
    Jeffery  MM, Hooten  WM, Henk  HJ,  et al.  Trends in opioid use in commercially insured and Medicare Advantage populations in 2007-16: retrospective cohort study.  BMJ. 2018;362:k2833. doi:10.1136/bmj.k2833PubMedGoogle ScholarCrossref
    12.
    Zhou  C, Yu  NN, Losby  JL.  The association between local economic conditions and opioid prescriptions among disabled Medicare beneficiaries.  Med Care. 2018;56(1):62-68. doi:10.1097/MLR.0000000000000841PubMedGoogle ScholarCrossref
    13.
    Davis  MA, Lin  LA, Liu  H, Sites  BD.  Prescription opioid use among adults with mental health disorders in the United States.  J Am Board Fam Med. 2017;30(4):407-417. doi:10.3122/jabfm.2017.04.170112PubMedGoogle ScholarCrossref
    14.
    Deyo  RA, Von Korff  M, Duhrkoop  D.  Opioids for low back pain.  BMJ. 2015;350:g6380. doi:10.1136/bmj.g6380PubMedGoogle ScholarCrossref
    15.
    Social Security Administration. Annual statistical report on the Social Security Disability Insurance Program, 2016: beneficiaries in current-payment status. https://www.ssa.gov/policy/docs/statcomps/di_asr/2016/index.html. Published October 2017. Accessed March 25, 2019.
    16.
    Moyo  P, Zhao  X, Thorpe  CT,  et al.  Dual receipt of prescription opioids from the Department of Veterans Affairs and Medicare Part D and prescription opioid overdose death among veterans: a nested case-control study.  Ann Intern Med. 2019;170(7):433-442. doi:10.7326/M18-2574PubMedGoogle ScholarCrossref
    17.
    Guy  GP  Jr, Zhang  K, Bohm  MK,  et al.  Vital signs: changes in opioid prescribing in the United States, 2006-2015.  MMWR Morb Mortal Wkly Rep. 2017;66(26):697-704. doi:10.15585/mmwr.mm6626a4PubMedGoogle ScholarCrossref
    18.
    Goodwin  JS, Kuo  YF, Brown  D, Juurlink  D, Raji  M.  Association of chronic opioid use with presidential voting patterns in US counties in 2016.  JAMA Netw Open. 2018;1(2):e180450. doi:10.1001/jamanetworkopen.2018.0450PubMedGoogle Scholar
    19.
    Han  B, Compton  WM, Blanco  C, Crane  E, Lee  J, Jones  CM.  Prescription opioid use, misuse, and use disorders in US adults: 2015 National Survey on Drug Use and Health.  Ann Intern Med. 2017;167(5):293-301. doi:10.7326/M17-0865PubMedGoogle ScholarCrossref
    20.
    Liaw  V, Kuo  YF, Raji  MA, Baillargeon  J.  Opioid prescribing in disabled adults in the United Sates following the 2014 federal hydrocodone rescheduling regulation.  Public Health Rep. In press.Google Scholar
    21.
    Brady  JE, Giglio  R, Keyes  KM, DiMaggio  C, Li  G.  Risk markers for fatal and non-fatal prescription drug overdose: a meta-analysis.  Inj Epidemiol. 2017;4(1):24. doi:10.1186/s40621-017-0118-7PubMedGoogle ScholarCrossref
    22.
    Ruhm  CJ.  Corrected US opioid-involved drug poisoning deaths and mortality rates, 1999-2015.  Addiction. 2018;113(7):1339-1344. doi:10.1111/add.14144PubMedGoogle ScholarCrossref
    23.
    Harrell  FE.  Regression Modeling Strategies: With Applications to Linear Models, Logistic Regression, and Survival Analysis. New York, NY: Springer; 2001. doi:10.1007/978-1-4757-3462-1
    24.
    National Institute on Disability, Independent Living, and Rehabilitation Research. Summary of responses from a request for information: people with disabilities and opioid use disorder. https://acl.gov/sites/default/files/news%202018-05/20180502NIDILRROpioidRFIFindings.pdf. Updated September 24, 2018. Accessed May 15, 2019.
    25.
    Lauer  EA, Henly  M, Brucker  DL.  Prescription opioid behaviors among adults with and without disabilities: United States, 2015-2016.  Disabil Health J. 2018;12(3):519-522. doi:10.1016/j.dhjo.2018.12.001PubMedGoogle ScholarCrossref
    ×