This is such an important question to resolve! Especially now, with more than a dozen states yet to decide on Medicaid expansion, while studies—some pro, some con, fail to provide an answer.
So many factors are in play, and the statistical analysis involved is intensive.
Background
The Patient Protection and Affordable Care Act (ACA) became law in 2010, permitting states to expand Medicaid coverage for most low-income adults, up to 138% of the federal poverty level. States that chose to do so are called expansion states.
The ACA requires that mental health and substance use disorder (SUD) services be treated on an equal level with other medical services.
Expansion States
States can opt in or out of the Medicaid expansion portion of the ACA. To date, 37 states and Washington, DC have adopted expansion; 14 states have not.
Whether or not a state has opted for expansion, residents can qualify for Medicaid, based on income, disability, and other factors.
Residents of states that have expanded Medicaid coverage can qualify based on income alone.
https://www.healthcare.gov/medicaid-chip/medicaid-expansion-and-you/
The Medicaid services include enabling people of low income and relatively higher rates of chronic disease to access low-cost prescription medications, including opioid pain relievers. Because the Medicaid population has higher rates of chronic disease and disability, some speculation has been circulating that Medicaid expansion might actually increase the risk of opioid-related harms.
Changes in Expansion States Since Passage
According to a study by Kravitz-Wirtz et al, papers published since expansion have found an increase in Medicaid reimbursement for prescriptions. But they have not found a significant change in rate of prescriptions written for opioid pain relievers, before and after expansion, in expansion versus nonexpansion states.
And some studies have suggested that the use of medications for opioid use disorder (MOUDs) has gone up in expansion states, compared to nonexpansion states. MOUD medications include methadone, buprenorphine, extended-release naltrexone, and naloxone, the agent that reverses ODs.
What About Opioid OD Mortality?
The Kravitz-Wirtz team wondered if ACA-related Medicaid expansion is associated with a change in county-level opioid OD mortality rates. Studies have been conducted at the state level, but the investigators wondered if state-level analyses might fail to reflect within-state differences in policy implementation, and the degree and rate of increase in OD deaths.
So, the team designed a study to find out. In January they published their findings online in JAMA Network Open.
The Study
Objective:
To examine whether Medicaid expansion is associated with county × year counts of opioid OD deaths overall, and by class of opioid.
Key Points
- Design: A serial cross-sectional study (data are collected in the study population at different times)
- Study area: 3,109 counties within 49 states and the District of Columbia
- 3,109 x 17 years = 52,853 county-years
- Study dates: January 1, 2001 ─ December 31, 2017
- Analysis dates: April 1, 2018 ─ July 31, 2019
Primary Exposure:
─State adoption of Medicaid expansion under the ACA.
Using data from the Kaiser Family Foundation, the team measured each state’s primary exposure as the proportion of each calendar year Medicaid expansion was in effect in that state.
Outcomes of Interest
Annual county-level deaths from ODs involving specific drugs:
- Any opioid
- Natural and semisynthetic opioids
- Methadone
- Heroin
- Synthetic opioids other than methadone
The OD data were derived from the National Vital Statistics System multiple cause-of-death files. A secondary analysis examined fatal ODs involving all drugs.
Results
─ Opioid OD fatalities totaled 383,091 across the observed counties from January 1, 2001 through December 31, 2017. Deaths per county averaged 7.25; range, 0 ─ 1,145.
─ The overall opioid mortality per 100,000 rose from 2.49 deaths in 2001 to 11.41 in 2017.
─ Counties in expansion states had a death rate involving heroin that was 11% lower than the rate in counties in nonexpansion states.
─Expansion was associated with an 11% increased rate of OD deaths involving methadone.
─ Rates were generally higher in expansion states than in nonexpansion states.
─ Yet, counties in states that had adopted expansion had a total opioid OD death rate 6% lower than the rate in counties in nonexpansion states.
(These findings, some of which seem contradictory, are discussed below.)
Secondary Analysis
Analysis of OD deaths involving all drugs showed that the OD death rate decreased 2% after expansion, compared with the death rate in counties in nonexpansion states.
Additional Findings
When the study ended in 2017, synthetic opioids other than methadone were involved in most OD deaths.
Opioids (%) Involved in County-Year Opioid OD Deaths
- 40.9 natural and semisynthetic opioids
- 25.3 heroin
- 24.0 synthetic opioids other than methadone (eg, illicitly manufactured fentanyl)
- 17.1 methadone (used mostly for treating pain, rather than OUD) )
Methadone and OD Mortality
The authors commented that the finding that methadone-related OD mortality increased 11% in expansion states deserves further investigation. They did not believe that increased access to MOUDs contributed to the increased mortality. Research has found that high rates of methadone are used to treat pain—rather than opioid use disorders—in Medicaid beneficiaries, and “the drug is disproportionately associated with overdose deaths among individuals in this population.”
Lives Saved
The authors pointed out that in the 32 states that expanded Medicaid between 2014 and 2016, 82,228 opioid-related deaths occurred in the 2015 to 2017 period. They calculated that, if expansion did not lower the death rate in expansion states, between 83,906 and 90,360 deaths would have occurred. Therefore, they reasoned, expansion may have prevented between 1,678 and 8,132 deaths in these states during those years.
ACA Inclusion of Mental Health and SUD Benefits
Partially explaining the study’s findings, the authors believe, is the ACA’s coverage of mental health and SUD services. People of low income now have better access to MOUDs and naloxone, and prescriptions for these drugs have increased in expansion states.
Study Limitations
- Using ICD-10 coding may have underestimated opioid OD mortality
- Including deaths from the entire population may have understated some outcomes
- Unmeasured confounding—a possible cause of bias—may have occurred
- Not examined were expansion provisions that could be associated with changes in opioid-related deaths
- Focus was limited to the link between expansion and fatal doses
Authors’ Recommendations for Future Studies
- Examine the mediators and moderators possibly involved in Medicaid expansion, and the possible association of Medicaid expansion with risk of opioid OD
- Consider “the association of expansion with the spectrum of opioid-related harms, including prevention of SUD and nonfatal overdoses”
Authors’ Conclusions
The authors wrote, “Medicaid expansion was associated with reductions in total opioid overdose deaths, particularly deaths involving heroin and synthetic opioids other than methadone, but with increases in methadone-related mortality.” Noting that expansion may play a critical role in states’ efforts to address the opioid OD epidemic, they suggested that states pay attention to the role health coverage expansion can play, potentially by increasing access to medications for opioid use disorder.
Reference
Kravitz-Wirtz N, Davis CS, Ponicki WR, et al. Association of Medicaid expansion with opioid overdose mortality in the United States. JAMA Netw Open. 2020;3(1):e1919066. doi: 10.1001/jamanetworkopen.2019.19066
Commentary
Revisiting the Results Summary: More Deaths? Fewer Deaths?
Some comments that may help clarify the authors’ summary in the Results section, above:
A statement in the Results section of the Abstract associates Medicaid expansion with a 6% lower rate of total opioid OD deaths, compared with the rate in nonexpansion states. Yet a statement in the Results section of the full paper notes that opioid mortality rates were generally higher in expansion states than in nonexpansion states (italics added).
The authors reported that the overall opioid OD mortality began to rise in 2001, and continued to rise after 2014—the year the expansion option became available. Yet, toward the end of that period, rates were generally higher in expansion state counties.
We contacted the study’s lead author, Nicole Kravitz-Wirtz, PhD, MPH, and asked for clarification. She explained in an email that “the rate of change in opioid overdose mortality in Medicaid expansion states was lower than expected, given the prior trajectory of opioid overdose deaths in those states and in non-expansion states.” (italics added)
With that finding, Dr. Kravitz-Wirtz said, these two statements could “exist in tandem: On average, Medicaid expansion states do indeed have higher rates of opioid overdose mortality, but in expansion states, our findings suggest that the rate would have been even higher in the absence of the expansion.”
We mulled this over. On this point, the team focused on how fast the opioid OD mortality rate was increasing. And, after 2014, the rate of the rise of mortality—in other words, how fast the death total was rising—began to decrease in counties in expansion states. To put it another way: the death toll was rising more slowly in expansion-state counties than in nonexpansion-state counties, so proportionally fewer lives were lost.
By distinguishing between the mortality rate and how fast that rate was changing, the authors could forecast likely future mortality rates.
A similar situation was occurring with heroin OD deaths. The rise in the death toll was slowing even more with heroin than with opioids as a group. The result: a forecast for fewer heroin OD deaths than would be expected without expansion. In fact, even fewer OD deaths for heroin than would be expected for opioids as a group.
Other aspects of the Kravitz-Wirtz study left us seeking clarification as well.
The eFigure: Did Death Rates Rise? Did They Fall?
We, and others, found the eFigure confusing. In a comment on the electronic version of JAMA Network Open, Michael Ball, MD, wrote: “Nonsense! Please open the paper clip above and look at eFigure. Since 2013, in every category, including total opioid death rate, the non-expansion counties have a lower death rate than the expansion counties!”
The eFigure was also cited in a publication of the Texas Public Policy Foundation, “Is Medicaid Expansion Worth It?” This publication noted that the Kravitz-Wirtz study used numerous controls and a one-year lag to show that death rates from opioids fell in expansion states—a result the Texas authors said was heavily influenced by the control variables selected. They also commented, “The authors did not publish their results without the one-year time lag, so it is unknown how their decision to include a one-year time lag influenced the results.”
(In the Statistical Analysis section, however, the Kravitz-Wirtz team indicate that the results were pretty much the same, whether they included the lag or not.)
Counties Versus States
Another issue of initial confusion was the distinction between counties and states. The data analyzed were from counties, not states—in fact, that was in the study design, as we mentioned earlier. So the results reported were not from “expansion states versus non-expansion states,” but from “counties in expansion states compared with counties in non-expansion states.”
Reaching Out
Statistics and advanced math—not our strong points—play a heavy role in expansion studies. So we reached out to James A. Swartz, PhD—an expert statistician, and lead author on a paper recently published in Addiction. His paper looked at Medicaid expansion from the perspective of prescription opioid availability and opioid OD-related mortality rates.
We asked Dr. Swartz about several issues we were still mulling over.
Opioid OD deaths in expansion-state counties versus counties in nonexpansion states—“The key,” Dr. Swartz explained, “is the rate of increase versus the absolute rate.” In other words, fatalities in expansion states in 2017 were higher than in nonexpansion states, but the rate of increase was falling. The way it was falling indicated that the projected rate would likely become less than the rate in nonexpansion states. If the states hadn’t approved Medicaid expansion, the number of fatalities expected would have been 6% higher.
Although Dr. Swartz used different words, what he said agreed with the lead author’s explanation.
The eFigure—He, too, found the eFigure somewhat confusing. It shows, he said, unadjusted rates. The figure doesn’t adjust for many factors that could influence fatality rates—age, income, unemployment, population density, and others. He likened this to comparing a very rural state with a high level of white residents and a low population density (Nebraska), with a more urban state—one that has a higher minority population and a high population density (New York), “and not taking into account how those population differences influence fatality rates.”
In Short— although we needed help to understand some of the findings and details, the Kravitz-Wirtz study appears basically sound. And it adds information about the effects of expansion; in particular, data showing that in their study, counties in Medicaid expansion states had an overall higher rate of fatalities, but a lower rate of change in fatalities, compared to the nonexpansion states.
So, Did Expansion Increase Opioid OD Deaths? Or, Did It Save Lives?
That remains the key question.
Dr. Swartz says that, unfortunately, we have no answers, only speculation. “You could speculate there was a lag between increases in prescription opioids and fatalities,” so, someone could become addicted within a year after starting a prescription opioid, and could overdose several years later—but the lag time is longer than a few years.
Dr. Swartz’ own published study concluded that increases in Medicaid-reimbursed prescription opioid availability do not appear to have mediated differences in mortality rates after expansion. Yet, he said, there is “still a possibility of lagged effects.”
The Bottom Line
It seems that where the effects of Medicaid expansion are concerned, it may be quite a while before we have a solid answer to this key question.
In other words, despite the studies, despite the data collected, despite the intricate statistical analyses, we still don’t know why mortality rates are higher after expansion in expansion states than in nonexpansion states. What we do know is that they are higher—but apparently not as high as they would be without expansion.
Additional Reading
Swartz JA, Beltran SJ. Prescription opioid availability and opioid overdose-related mortality rates in Medicaid expansion and non-expansion states. Addiction. 2019;114(11):2016-2025. doi:10.1111/add.14741
Blasé B, Balat D. Is Medicaid Expansion Worth It? A Review of the Evidence Suggests Targeted Programs Represent Better Policy. Texas Public Policy Foundation. April 2020.
§ § §
Acknowledgement
We’re deeply grateful to Dr. Swartz for guiding us through some difficult concepts and helping us understand this topic.
James A. Swartz, PhD, is a Professor at Jane Addams College of Social Work, University of Illinois at Chicago. His current areas of research include the epidemiology of opioid use and misuse, and opioid use disorders.
Dr. Swartz serves as a Peer Reviewer for 18 scientific journals.