- Review previous evidence on employer costs associated with opioid use disorder (OUD).
- Summarize the findings of the new analysis of the impact of OUD on employee healthcare and productivity costs.
- Discuss the findings on the cost impact of pharmacotherapy for employees with OUD.
Employee health can affect employer health care and productivity costs.1–3 Employers, particularly self-insured employers, increasingly are concerned specifically about the costs associated with opioid misuse among their employees. Individuals prescribed opioids are at risk for dependence,4–6 and many employees receive such prescriptions. Doctors prescribe painkillers to nearly 40% of individuals who seek help for lower back pain, one of the most common workplace ailments.7 Although opioid prescribing to adolescents is rare, dependents of employees are at risk for misusing opioids from leftover prescriptions.8
Prior research has measured the impact of employee opioid use disorder (OUD) on employer costs.9–11 One study found that employees who are dependent on opioids but have not been diagnosed with OUD have lower at-work productivity, which costs employers approximately $16 million a year.9 Another study using 2006 to 2012 data reported that individuals with OUD had seven more medically related absenteeism days annually relative to matched controls.12 A third study found that US adults who misuse prescription pain relievers have higher work absenteeism than do employees who do not.13 Studies focusing on health care costs have found that individuals who misuse opioids have more than $10,000 more in annual expenditures.12,14–16
Employers do not have a recent or full picture of costs related to OUD. Employees who have a spouse or dependent with an OUD may have additional lost productivity days and days absent because of family member health concerns. Employees may have to help their family member navigate health care benefits during business hours, including identifying appropriate and available providers for substance use disorder (SUD). Employees also may assume a caregiving role, particularly during relapse or potential relapse. A cross-sectional study of caregivers of individuals with advanced cancer found a 23% drop in average productivity.17 Another study that looked at caregivers of patients with poststroke spasticity found that lost-productivity cost per employed caregiver was $835 per month, with 72% attributable to presenteeism.18
As payers, employers have the opportunity to improve access to OUD treatment by covering pharmacotherapy such as methadone, buprenorphine/naloxone, and naltrexone as part of their health care benefits. The use of these Food and Drug Administration-approved pharmacotherapies, in combination with counseling and behavioral therapies, provide a whole-patient approach to the treatment of OUD. Not all people with OUD seek treatment, but those who do seek treatment do not necessarily receive this evidence-based treatment for multiple reasons—including stigma, cost, and lack of available providers. Health-plan restrictions such as limited provider networks, prior authorization, counseling requirements, quantity limits, step therapy requirements, duration limitations, and network requirements also are obstacles to access for many.19 Treatment admissions data show that only 18% of people admitted for prescription OUD have a treatment plan that includes pharmacotherapy.20 Employers that already cover these pharmacotherapies can invest in efforts to increase awareness and uptake of these important treatment options. In making these decisions, employers need to consider the business case for these investments, such as whether pharmacotherapies increase or decrease total health care expenditures.
In this study, we describe the prevalence of OUD among employees and their spouses and dependents. We measure the association of employee OUD and presenteeism, health-related absences, and health care costs. We also examine whether family member health affects these outcomes and whether receipt of pharmacotherapy to treat OUD reduces days lost to absences, preseenteism, or health care costs associated with OUD. Finally, we compare the costs of treating OUD with the costs of treating mental illness, another condition that is associated with high health care costs for employers.21
Study Design and Data Source
We conducted a retrospective, cross-sectional analysis of a large commercial claims database linked with survey data to measure the influence of employee and family member mental illness and substance use on presenteeism, absences, and health care costs. Demographic data, medical claims, and health risk assessment data were extracted from the 2016 and 2017 IBM® MarketScan® Commercial and Health Risk Assessment Databases. The Marketscan Commercial data include deidentified patient-level enrollment, inpatient, outpatient, outpatient prescription drug, productivity, and health risk assessment data for millions of employees and their dependents enrolled in employer-sponsored health plans. The MarketScan Health Risk Assessment data include employees’ self-reported responses to a health risk assessment survey administered by their employer that contains questions that are primarily about employees’ health and health-related behaviors. Used primarily for research, these databases are fully compliant with US privacy laws and regulations (ie, the Health Insurance Portability and Accountability Act). Marketscan data have been used in more than 1100 peer-reviewed publications to date.17
The study sample included active employees who met the following inclusion criteria: (1) were aged 18 to 64 years and (2) were continuously enrolled in health coverage through their employer in 2016 or 2017. The productivity analyses were limited to employees who had completed a health risk assessment because we used responses to the survey to measure productivity.
We examined four outcomes: health-related absences, presenteeism, absences due to family member health, and annual health care costs. The first three outcomes were measured using responses to the following health risk assessment items: “In past 12 months, how many work days were missed due to illness or injury?”; “In the past 4 weeks, how many days did your health problems affect productivity at work?”; and “In the past 6 months, how many work days were missed due to care of a sick family member?” Some employers used health risk assessment questions with different look-back time frames. In all cases, we normalized responses to 12 months to create annual figures. Further, when relevant, we replaced categorical responses with mean number of days. Total annual health care costs were the sum of the employer and employee payments for all incurred medical and pharmacy claims in the calendar year.
Our primary measure of interest was OUD, which the Centers for Disease Control and Prevention, the US Department of Health and Human Services, and the Centers for Medicare & Medicaid Services define as “A problematic pattern of opioid use that causes significant impairment or distress.”22 We reviewed all-listed diagnosis codes on outpatient, emergency department, and inpatient claims for an OUD diagnosis. To identify employees and spouses who were receiving pharmacotherapy for OUD, we scanned (1) pharmacy claims for evidence of buprenorphine-naloxone and naltrexone and (2) procedure codes for evidence of methadone and buprenorphine service administration in the outpatient setting.
We also flagged employees and dependents with other behavioral health and chronic conditions. We scanned diagnosis codes for evidence of nonopioid SUD including misuse of alcohol, cocaine, amphetamine, cannabis, and combinations of drugs excluding opioid type drugs. We also looked for mental health diagnoses including adjustment disorder, anxiety disorders, attention-deficit disorder, conduct disorder, disruptive behavior disorders, mood disorders, personality disorders, and schizophrenia and other psychotic disorders. The codes used to identify behavioral health and pharmacotherapy for OUD are provided in Appendix A, http://links.lww.com/JOM/A711. We used the Agency for Healthcare Research and Quality Clinical Classifications Software to measure for the presence of chronic conditions.23 Chronic conditions included chronic nervous system disorders, headaches, rheumatoid arthritis and related disease, other non-traumatic joint disorders, osteoarthritis, and other back pain and disorders.
We applied exclusion criteria to identify (1) employees with 1 year of employer-sponsored health care enrollment and health risk assessment data and (2) employees with 1 year of employer-sponsored health care enrollment data with or without health risk assessment data. We tabulated the distribution of all study variables across these two samples. We estimated regression models that measured the association between the employee and the employee's family member health on self-reported days of reduced productivity at work, health-related absences, family-related absences, and health care costs. By necessity, we limited the sample estimating reduced productivity and absences to the sample of employees with health risk assessment data because health risk assessment data were used to measure these outcomes. Health measures included as independent variables in the model were OUD, nonopioid SUD, mental illness, and the presence of chronic medical conditions. Other model covariates included employees’ sex, age, health plan type, industry, and salary type (salaried vs hourly wage). When we estimated regression models examining the association between employee health and health care costs, we estimated models including and excluding measures of employee family member health. In this study, we report on the more parsimonious models excluding measures of family member health.
Table 1 provides characteristics of the study samples. A total of 16,886,347 study subjects were identified on the basis of the inclusion criteria. Of these, about 6% (972,936) had completed the health risk assessment survey and were included in the analyses of absences and work productivity.
Employees were geographically dispersed and tended to be male, hold salaried positions, and have preferred provider organization health plan types. Seven in 10 employees in the sample had at least one chronic health condition. About half of employees in the sample had a covered spouse aged 18 to 64 years, and about one-fifth of employees had at least one covered dependent. Characteristics for the subset of employees with a completed health risk assessment were generally similar. A comparison of industry type is provided in Appendix B, http://links.lww.com/JOM/A711.
Looking at the all employees sample, less than 1% of employees (0.6%), their spouses (0.4%), and their dependents (0.2%) had a diagnosis of OUD. Nonopioid SUDs were higher among employees (1.1%) than among their dependents (0.7%) and spouses (0.5%). Mental illness was prevalent among employees (16.2%), their spouses (7.4%), and their dependents (6.5%). These patterns were similar for the subset of employees with health risk assessment data.
Appendix Table C, http://links.lww.com/JOM/A711 provides results from a bivariate analysis of sample characteristics and outcomes. In this unadjusted analysis, we find that all sample characteristics are statistically significantly associated with outcomes (P < 0.001).
Table 2 provides the results of the regression analyses for presenteeism and absenteeism for the subset of employees with health risk assessment survey data. In the regression model predicting presenteeism, employee health—particularly OUD, nonopioid SUD, mental illness, and chronic conditions—were associated with more lost productive days. Employee OUD and no evidence of medication-assisted treatment (MAT) had a greater impact on presenteeism than employee mental illness (14 vs 5 lost at-work productivity days). The association between OUD with evidence of pharmacotherapy and presenteeism was not statistically significant and was much lower in magnitude then the relationship between OUD and no evidence of MAT and lost productive days. None of the indicators for family member health were positive predictors of lower at-work productivity.
Absences were positively predicted by employee health, in particular OUD with no evidence of pharmacotherapy (0.67, P < 0.0001), OUD with evidence of pharmacotherapy (0.32, P = 0.0004), nonopioid SUD (0.24, P < 0.0001), mental illness (0.22, P < 0.0001), and chronic conditions (0.08, P < 0.0001). The only family health indicator that positively predicted absences was spouse's mental health (0.06, P = 0.0005). Family-related absences were positively predicted by employee health (mental illness, opioid and nonopioid SUDs, and chronic conditions) and spouse's mental health (0.03, P = 0.0001).
For both the full sample and the subset of employees with health risk assessment data, annual health care costs were positively predicted by employee health (mental illness, opioid and nonopioid SUDs, and chronic conditions) (see Table 3). OUD with evidence of pharmacotherapy predicted costs, but the magnitude was much lower than OUD without evidence of pharmacotherapy ($5427 vs $19,178 in the model not limited to health risk assessment respondents). Employee OUD without MAT was the most expensive employee behavioral health condition. In comparison, employee mental illness was associated with $3921 in additional health care costs and nonopoid SUD was associated with $10,876 additional health care costs. In the model focusing on all employees and including family member health measures, spouse OUD and spouse mental illness, and dependent mental illness were significantly associated with employee health care costs. Full model results including parameter estimates associated with all covariates are provided in Appendix D, http://links.lww.com/JOM/A711. We found that plan type had a variance inflation factor over greater than 5 in the presenteeism model thus removed this variable and re-estimated the model as a sensitivity analysis. Model estimates did not change meaningfully (Appendix E, http://links.lww.com/JOM/A711) providing confidence in the original specification.
We based our estimate of the incremental costs associated with an employee with OUD who did not receive pharmacotherapy on the regression results using the sample with health risk assessment data because they had complete outcomes, and we used December 2018 data on the mean hourly wage of $27.53/hr from the Bureau of Labor Statistics.24 These additional costs came out to $24,858 a year, which included $3105 for presenteeism, $148 for sick days, $35 for family leave, and $21,570 for health care expenses.
Our study using data from a convenience sample of large employers found that about 5 out of 1000 employees have evidence of OUD in health care claims. Employees with OUD who did not receive pharmacotherapy cost employers about $25,000 per employee per year including lost productivity and increased health care costs. Employees who received any pharmacotherapy for OUD incurred significantly lower lost productivity and health care costs. Employees with spouses or dependents with OUD did not have increased sick days or family-related absences, productivity, or health care costs. OUD was associated with greater losses in productivity and larger increases in health care costs on a per employee basis than mental illness but was less prevalent.
These results are consistent with previous research that found that employers incur significant costs from OUD. Our findings add to the literature by providing evidence that employers would benefit financially from expanding access to pharmacotherapy for their employees with OUD. Employers should work with other payers to tackle important barriers to treatment for OUD by supporting efforts to expand provider education and licensure requirements to include MAT and increasing insurance coverage for these treatments.
Employers should be attentive of ongoing research related to the most effective ways to implement MAT. For example, research has shown that ongoing pharmacotherapy treatment is more effective than tapering patients off after a short course.25,26 Also, initiating buprenorphine treatment when a patient is admitted to the emergency department, such as for an overdose, has been found to be a more effective way to engage a patient in treatment than referral or brief intervention.27 Employers should partner with health plans and providers to review care provided to their population with OUD to take steps to ensure that these best practices are being adopted.
Prejudice and discrimination against people with substance abuse and mental disorders has remained high over the period from the 1950s to 2006 according to the General Social Survey stigma modules.28 Although public awareness and understanding of the neurobiological drivers and treatments of these disorders has increased, stigma levels have not decreased.28,29 Employers can help change stigmatizing attitudes and beliefs and implement changes to help employees and their families navigate to and obtain SUD and mental health services.
The opioid crisis lies at the intersection of two public health challenges: reducing the burden of suffering from pain and containing the increasing toll of the harms that can arise from use of opioid medications. On the one hand, meeting the needs of tens of millions of US residents suffering from pain (including acute pain, chronic pain, or pain at the end of life) requires access to a broad array of therapies for pain management. On the other hand, harms associated with use of prescription opioids, including misuse, OUD, and overdose, affect not only patients with pain but also their families, their communities, and society at large. Employers can reduce the likelihood that employees become addicted to opioids by partnering with health plans and pharmacy benefit plans to ensure access to opioids when appropriate and limiting access when the risks outweigh any benefits. Because prescribing guidelines may be most effective when accompanied by education, an evidence-based national approach to pain education, including pharmacologic and nonpharmacologic treatments and materials on opioid prescribing, is needed. Employer and insurance policies aimed at reducing the use of specific prescription drugs such as limitations on opioid days’ supply should be used intelligently to ensure access to medications for individuals for whom they are appropriate. Further, employers should consider coverage for and access to comprehensive pain management that includes both pharmacologic and nonpharmacologic options such as acupuncture, massage therapy, and other alternative approaches.
This study had limitations. First, we used a convenience sample of large employers that contributed data to the MarketScan Research Databases. Findings may not be generalizable to employees who work for small- or medium-sized employers. Second, outcomes measured using health risk assessment data were based on self-report. Only employees who voluntarily completed the surveys were included in the models of these outcomes, and, as shown in Table 1, these employees are systematically different from employees who did not complete the surveys. Third, questions referred to earlier time periods may have been subject to recall bias. Fourth, we required employees and their dependents to have 12 months of continuous enrollment. Employees and dependents with fatal overdoses or whose disorder led to employment termination within this time frame would have been excluded from this study. A fifth limitation is related to our examination of family member health and employee outcomes. Employees may provide a range of support to family members who are ill—from no support to routine support of activities of daily living. We did not have the data to distinguish the level of caregiving and involvement an employee had in a family member's care. Sixth, we used claims data to identify behavioral health conditions (OUD, nonopioid SUD, and mental illness), but individuals who never seek or receive treatment for these conditions would not have been correctly identified. Similarly, because some pharmacotherapy providers do not accept insurance, we could not detect in claims some individuals who may have been receiving pharmacotherapy for OUD.
Seventh, our analysis of the association between OUD and lost work productivity and costs includes data from one year—any costs that began in the study year and spanned to the next year would not have been included making our estimates more conservative. Finally, enrollees with OUD may have been more likely to have unobservable preexisting medical conditions or medical events such as injuries and surgeries, associated with higher costs. These variables were not available in our dataset and have a complex relationship with OUD and costs worthy of further study.
Opioid misuse is problematic across all demographics, including employed individuals and their families.30 Employee OUD increases employer health care and productivity costs but has a much lower impact if employees receive treatment. Employers should consider efforts to reduce the incidence of OUD and improve treatment uptake because of its large impact on productivity and health care costs. Future research should examine whether employees with OUD who receive maintenance pharmacotherapy for OUD adherent to treatment guidelines have greater improvements in productivity and reductions in health care costs than employees who receive short-term treatment.
We thank Paige Jackson (IBM Watson Health) for editorial assistance.
1. Goetzel RZ, Long SR, Ozminkowski RJ, Hawkins K, Wang S, Lynch W. Health, absence, disability, and presenteeism
cost estimates of certain physical and mental health conditions affecting U.S. employers. J Occup Environ Med
2. Goetzel RZ, Hawkins K, Ozminkowski RJ, Wang S. The health and productivity cost burden of the “top 10” physical and mental health conditions affecting six large U.S. employers in 1999. J Occup Environ Med
3. Stewart WF, Ricci JA, Chee E, Hahn SR, Morganstein D. Cost of lost productive work time among US workers with depression. JAMA
4. Bohnert AS, Valenstein M, Bair MJ, et al. Association between opioid prescribing patterns and opioid overdose-related deaths. JAMA
5. Bohnert ASB, Guy GP Jr, Losby JL. Opioid prescribing in the United States before and after the Centers for Disease Control and Prevention's 2016 Opioid Guideline. Ann Intern Med
6. Guy GP, Zhang K, Bohm MK, et al. Vital signs: changes in opioid prescribing in the United States, 2006–2015. MMWR
7. Shute N. Poll: Doctors Are Still Prescribing Lots of Opioids for Low Back Pain
[National Public Radio Shots Health News from NPR Web site
]. May 19, 2017. Available at: https://www.npr.org/sections/health-shots/2017/05/19/528516985/poll-doctors-are-still-prescribing-lots-of-opioids-for-low-back-pain
. Accessed March 3, 2018.
8. McCabe SE, West BT, Boyd CJ. Leftover prescription opioids and nonmedical use among high school seniors: a multi-cohort national study. J Adolesc Health
9. Kirson NY, Shei A, Rice JB, et al. The burden of undiagnosed opioid abuse among commercially insured individuals. Pain Med
10. Johnston SS, Alexander AH, Masters ET, et al. Costs and work loss burden of diagnosed opioid abuse among employees on workers compensation or short-term disability. J Occup Environ Med
11. Florence CS, Zhou C, Luo F, Xu L. The economic burden of prescription opioid overdose, abuse, and dependence in the United States, 2013. Med Care
12. Rice JB, Kirson NY, Shei A, et al. Estimating the costs of opioid abuse and dependence from an employer
perspective: a retrospective analysis using administrative claims data. Appl Health Econ Health Policy
13. Van Hasselt M, Keyes V, Bray J, Miller T. Prescription drug abuse and workplace absenteeism
: evidence from the 2008–2012 National Survey on Drug Use and Health. J Workplace Behavioral Health
14. Roland CL, Joshi AV, Mardekian J, Walden SC, Harnett J. Prevalence and cost of diagnosed opioid abuse in a privately insured population in the United States. J Opioid Manag
15. Meyer R, Patel AM, Rattana SK, Quock TP, Mody SH. Prescription opioid abuse: a literature review of the clinical and economic burden in the United States. Popul Health Manag
16. Kirson NY, Scarpati LM, Enloe CJ, Dincer AP, Birnbaum HG, Mayne TJ. The economic burden of opioid abuse: updated findings. J Manag Care Spec Pharm
17. Mazanec SR, Daly BJ, Douglas SL, Lipson AR. Work productivity and health of informal caregivers of persons with advanced cancer. Res Nurs Health
18. Ganapathy V, Graham GD, DiBonaventure MD, Gillard PJ, Goren A, Zorowitz RD. Caregiver burden, productivity loss, and indirect costs associated with caring for patients with poststroke spasticity. Clin Interv Aging
19. Gingrich N, Kennedy, PJ, Jones V. Medication-Assisted Treatment Critical to Fight Addiction
[The Hill web site]. October 4, 2016. Available at: https://thehill.com/opinion/op-ed/299275-medication-assisted-treatment-critical-to-fight-addiction
. Accessed October 28, 2018.
20. Substance Abuse and Mental Health Services Administration, Center for Behavioral Health Statistics and Quality. Treatment Episode Data Set (TEDS) 2002-2012: National Admissions to Substance Abuse Treatment Services
[Substance Abuse and Mental Health Services Administration web site]. July 2014. Available at: https://www.samhsa.gov/data/sites/default/files/TEDS2012N_Web.pdf
. Accessed October 28, 2018.
21. Goetzel RZ, Pei X, Tabrizi MJ, et al. Ten modifiable health risk factors are linked to more than one-fifth of employer
-employee health care spending. Health Affairs
22. Centers for Disease Control and Prevention. Commonly Used Terms
[Centers for Disease Control and Prevention web site]. August 28, 2017. Available at: https://www.cdc.gov/drugoverdose/opioids/terms.html
. Accessed October 16, 2018.
23. Healthcare Cost and Utilization Project. Beta Clinical Classifications Software (CCS) for ICD-10-CM/PCS
[Agency for Healthcare Research and Quality web site]. October 2018. Available at: www.hcup-us.ahrq.gov/toolssoftware/ccs10/ccs10.jsp
. Accessed October 28, 2018.
24. U.S. Department of Labor, Bureau of Labor Statistics. Data Retrieval: Employment, Hours, And Earnings (CEs)
[Bureau of Labor Statistics web site]. Last modified September 15, 2015. Available at: https://www.bls.gov/webapps/legacy/cesbtab3.htm
. Accessed October 28, 2018.
25. Woody GE, Poole SA, Subramaniam G, et al. Extended vs short-term buprenorphine-naloxone for treatment of opioid-addicted youth: a randomized trial. JAMA
26. Johnson RE, Jaffe JH, Fudala PJ. A controlled trial of buprenorphine treatment for opioid dependence. JAMA
27. D’Onofrio G, O’Connor PG, Pantalon MV, et al. Emergency department–initiated buprenorphine/naloxone treatment for opioid dependence: a randomized clinical trial. JAMA
28. Committee on the Science of Changing Behavioral Health Social Norms; Board on Behavioral, Cognitive, and Sensory Sciences; Division of Behavioral and Social Sciences and Education; National Academies of Sciences, Engineering, and Medicine. Ending Discrimination Against People With Mental and Substance Use Disorders: The Evidence for Stigma Change
[National Academies Press web site]. 2016. Available at: https://www.nap.edu/catalog/23442/ending-discrimination-against-people-with-mental-and-substance-use-disorders
. Accessed October 28, 2018.
29. Pescosolido BA, Martin JK, Long JS, Medina TR, Phelan JC, Link BG. “A disease like any other”? A decade of change in public reactions to schizophrenia, depression, and alcohol dependence. Am J Psychiatry
30. Davenport S, Matthews K. Opioid Use Disorder in the United States: Diagnosed Prevalence by Payer, Age, Sex, and State
[Milliman web site]. 2018. Available at: http://us.milliman.com/insight/2018/Opioid-use-disorder-in-the-United-States-Diagnosed-prevalence-by-payer--age--sex--and-state/?lng=A=0
. Accessed October 28, 2018.