Congress is finally taking action to combat these nationwide challenges. Most recently, on May 26, HR 5189, the Opioid Abuse Crisis Act of 2016 (sponsored by Ann M. Kuster, D-NH-2) was referred to the House Subcommittee on Health. Additionally, grants toward addressing opiate abuse have passed the House and the Senate, with differences between the chambers currently being resolved. Other legislation has been introduced but has not yet met passage this session.
Chronic pain is far from trivial, and when not managed well can contribute powerfully to depression, which is a leading cause of disability and can coexist with a number of other health problems. Moreover, some setting out to responsibly treat pain can potentially inadvertently find themselves addicted. Yet the way that we have worked to address severe and chronic pain over the years has changed, such that despite not reporting marked changes in the level of pain experienced, prescriptions for opioids have increased dramatically since the early 1990s with the introduction of MS Contin and Oxycontin. The long history of pain management and the more recent onset of this addiction epidemic suggests that biologically-based interventions will be only limited in their efficacy, and looking at the underlying socioeconomic aspects will be essential for an effective medial and policy intervention with respect to the addiction to opioids (and other substances).
The fact that the face of opioid addiction has changed over the years, affecting diverse populations of different ages, makes complicated the determination of the most effective interventions. However, I decided to peruse the data (it’s what I do…). I began by looking at the ProPublica database of doctors and drug prescriptions participating in Medicare Part D, with the database tracking all prescriptions from these providers. I focus on the prescription of Hydrocodone-Acetaminophen, otherwise known as Vicodin, though I examined also the prescription of Oxycodone, another opiate. Of the fifty states and the District of Columbia, 17 states had Vicodin as the #1 most prescribed drug in 2013, and 42 states had Vicodin as one of the ten most prescribed drugs (out of over 1500 drugs enumerated). Admittedly, pain medication does have a more general applicability than the more specific application of a thyroid drug or one aimed at lowering blood pressure or cholesterol, though these numbers are nevertheless quite striking. The absolute lowest rank that Vicodin had was #37. Oxycodone-HCL, or Oxycontin, earned lower ranks in prescribing, as did Oxycodone-Acetaminophen, or Percocet, and there was only a modest relationship between the prescribing of those drugs and the propensity with which Vicodin was prescribed.
I then looked more systematically at the relationship between opiate prescription and health and economic factors. I used as the dependent variable the prescription drug rank that Vicodin took in state level prescriptions. Thus, the value of 1 indicates that Vicodin was the most common medication prescribed, while higher values indicate a lower prevalence of the medication among other prescriptions written in that state, based on the data provided by ProPublica from 2013. This value ranges from 0 to 37.
I looked at a handful of independent variables. To identify the parts of the country most likely to be affected by opiate addiction based on geographic factors, I include the state’s rank with respect to population density (ranging from 1 to 51, with higher values associated with sparser populations), as well as the state’s rate of unemployment, which ranges from 2.5 to 6.6. To account for differences in the ideological orientation of the state and thus potential variation in the extent and nature of services provided, I account for the state government ideology using the measures provided by Berry et al (2011).
The extensive prescription of opioids does not necessarily lead to addiction. In theory, one can prescribe opioids to a large number of people who use them responsibly and do not become addicted. The Centers for Disease Control and Prevention (CDC) tracks the extent to which state health care expenses are attributable to opioid addiction or abuse, and I control for the percent of abuse-related health costs, which ranges from 0.1% to 17.1% (with California being the highest). Curiously, the states ranking in the highest on this dimension were not necessarily the states prescribing opioids to the greatest degree (California and Texas both prescribed Vicodin the most, but Florida and Ohio had Vicodin only in the top 10, and New York ranked at only 20).
Finally, to examine the effect of health spending on doctors’ propensity to dispense opioids, I looked at the extent of overall health spending as well as the extent of health spending specific to mental health. Given the close relationship between substance abuse and mental health, I focus on the results when controlling for mental health spending, though the results are consistent in both specifications.
While I examine the effects using ordinary least squares (OLS) given the large distribution of the dependent variable, the results are consistent when utilizing instead a negative binomial specification, which is appropriate to count data that exhibits overdispersion (as is the case here). I do not find that state ideology is associated with any difference in states’ doctors’ propensity to prescribe opioids. I find that a standard deviation increase in mental health spending is associated with a 2.7 standard deviation decline in the extent to which states prescribe opioids relative to other medications (p<.01). While, as I noted above, one can provide much access to the responsible use of opioids for the effective management of chronic or severe pain, I do find that a standard deviation in the percent of state health costs attributable to opioids is associated with a 1.6 standard deviation increase in the extent to which opioids are prescribed relative to other medications (p<.05).
Consistent with expectations, I find that a standard deviation increase in population density of a state is associated a 4.3 standard deviation increase in the extent to which opioids are prescribed relative to other drugs in the state (p<.01). I find also that a standard deviation increase in the state’s unemployment rate is associated with a 2.3 standard deviation increase in the extent to which opioids are prescribed relative to other medications in the state (p<.01). Replacing the state unemployment rate with the state poverty rate also produced a significant test statistic with a large substantive effect.
It goes without saying that these are simple models comprising only 48 observations (Florida and New Mexico did not report state mental health funding data) and a small number of controls. However, it is noteworthy that even with all of the demand on the data, there are clear patterns that emerge that are consistent with expectations: that we see greater opioid prescribing in more sparsely populated regions and those facing greater economic downturn, and that there appear to be strong associations between rates of opioid prescribing and the extent to which health expenditures are related to opioid abuse. Moreover, rather than being a fruitless effort to “throw money at the problem,” investing more in healthcare overall and mental health specifically may potentially be an effective intervention in working to combat these pervasive problems of abuse and addiction, whether by treating current addiction or by managing better peoples’ physical and psychological challenges leading one from responsible prescription use to abuse. While a number of the states investing larger amounts of money toward mental health care are wealthier and more populous (e.g., New York, New Jersey, Connecticut, Pennsylvania), that is far from the rule, with Alaska, Maine, Vermont, and Montana also ranking high on this dimension.
The data here cannot disentangle whether greater funding allocation helps to prevent the onset of major abuse by facilitating individuals’ reaching out to other resources for help, or whether that funding helps to intervene in reducing existing problems of abuse. Likewise, the data cannot tell us whether unemployment and poverty create conditions that become ripe for opioid abuse, or whether there are common factors underlying both unemployment and opioid abuse. However, what they can show is that given the strong associations that appear in these cursory regressions, we would be remiss in treating opioid addiction specifically as opposed to examining it in light of other economic factors as well as physical and mental illness to which it may be tied. For example, the highest rates of opioid prescriptions also occur in those states with higher rates of serious mental illness (though there does not appear to be an association with alcohol in general or binge drinking specifically). With high rates of co-morbidity between drug abuse and mood and anxiety disorders, given both genetic and environmental vulnerabilities and with economic downturn associated with increases in substance-use disorders, it is imperative that in addition to restricting the prescription of these substances in lieu of less (or non) habit forming medications such as NSAIDS, we look holistically at both the biological and socioeconomic factors underlying opioid abuse and its effective resolution, lest one abuse epidemic be replaced with simply another.