Good and Bad Arguments Against the Obamacare-Opiate Effect

The Health Affairs blog entry disputing that the Affordable Care Act worsened the opioid crisis makes three main claims:

First, trends in opioid deaths nationally and by Medicaid expansion status predate the ACA. Second, counties with the largest coverage gains actually experienced smaller increases in drug-related mortality than counties with smaller coverage gains. Third, the fact that Medicaid recipients fill more opioid prescriptions than non-recipients largely reflects greater levels of disability and chronic illness in the populations that Medicaid serves.

The first point is the one that’s being repeated on Twitter most triumphantly.

But this is actually a quite weak argument. As I said in yesterday’s post, it’s not true that the diverging trends by Medicaid expansion status predate the ACA- they predate the ACA’s full implementation in 2014-2015. From 2010, when the law was passed, through 2013, the whole country was obsessed with the law, state health care officials most especially. The idea that this couldn’t affect insurance rates and drug access seems silly, especially given that we see the trends for Medicaid expanding and non-expanding states diverge at this time. For example, we can compare the state-level change in uninsured rates from the average of 2007-2009 between expansion and non-expansion states: it’s clear that already by 2010-2012 the non-expansion states are kicking people off the rolls and the expansion states are getting more people signed up, even though the non-expansion states already have more uninsured: the ACA was already making more of a difference in the expansion states by 2010-2012:

uninsuredchange

Looking at individual states, this is even more clear: most “Red” holdout states had increased their uninsured rate between 2007-2009 and 2013, most “Blue” Medicaid expansion states had decreased their uninsured rate; it’s hard not to think ACA implementation (and decreasing the ranks of the uninsured being a national priority of the Democratic party) had to do with this:

ratechange

As Bob Dylan said, you don’t need a weatherman to know which way the wind is blowing.

The Health Affairs authors also seem similarly to misunderstand a National Review article that presents evidence that Medicaid contributed to the opiate crisis since the early 2000s, well before the ACA, responding: “In contrast, most states that expanded Medicaid began offering benefits in January 2014. (Six states expanded early but limited coverage.) The claim that Medicaid expansion is “largely responsible for starting the epidemic in the first place” is clearly false simply given this timeline.” No, the idea that Medicaid began contributing to the opiate epidemic decades ago and then continued to fuel the epidemic when expanded in 2014 is not inconsistent with the timeline of expansion.

The Health Affairs authors’ second argument is also weak for multiple reasons. First of all, the reason that states with the largest coverage gains weren’t the ones with the largest increases in overdoses was because states with the largest increases in overdoses already had almost complete coverage- they couldn’t increase coverage that much because there wasn’t that far to go. Several of the states with the largest increases in overdoses already had over 85% of their population insured by 2013:

Sandoe-Exhibit-2

As I’ve said before, it’s very likely that as you give insurance to a larger and larger proportion of young people, the marginal insured person has a larger propensity to abuse. This is why a relatively modest increase in coverage (from 85% to 90% in the case of some New England states, for example), might have a disproportionately large impact on overdoses, because the last person to get coverage might be the most likely to use that coverage to get or sell drugs.

graph666 (3)

This is a general pattern in American public-private ventures, like our subsidies for homeownership and our medical insurance system, wherein the opportunities for fraud and abuse increase dramatically as the percent of the population covered by subsidies increases, since the marginal person subsidized becomes higher and higher risk.

Their third argument is interesting but relatively irrelevant. It may be true that Medicaid recipients are less healthy and therefore more likely to receive opiate prescriptions and to overdose. But this is only relevant to an estimate of the impact of the Medicaid expansion if potential Medicaid recipients were already more healthy in non-expansion states than in expansion states prior to the expansion. (In Econometrics 101-speak, the omitted variables only bias the impact estimate if they are correlated with the treatment indicator.) As I noted in the posts on obesity and alcohol and suicide, there is little evidence that the “omitted variables” here of individual health prior to coverage expansion are actually biasing our estimates of the impacts of coverage expansion. And across age groups, there was an uptick in 2015 in Medicaid expansion states but not in non-expansion states, suggesting this isn’t simply about an aging population becoming more susceptible to chronic pain, for example:

ageovertimemedstatusrecent (1)

So, the three main arguments the Health Affairs authors present are pretty weak. What are some stronger arguments against the idea that Obamacare made the opiate crisis worse?

First, let’s note that the “Obamacare effect” is completely driven by a regional confound. The Northeastern quarter of the United States had a huge increase in overdose rates, and this quarter of the country (apart from Maine) also expanded Medicaid. But this might be almost entirely a process of regional diffusion from the historical centers of the problem in Eastern Ohio and West Virginia, sped along by a new center in New England:

cruderate2015 (1)recentdrugsstate (5)

Second, as Twitter users like the pseudonymous data analyst Xenocrypt and John Tucker (who I believe is a chemist with some expertise in pharamaceuticals) have pointed out to me, there’s a very ambiguous relationship over the last few years between prescriptions and overdoses.

This is possibly because reducing prescriptions or introducing abuse-resistant pharmaceuticals pushes addicts into using ultimately more destructive heroin or fentanyl. This is what this Rand paper suggests, somewhat plausibly. Personally, I think the brute fiscal impact of increasing subsidies in certain states has poured gasoline on the flames and, regardless of the specific mechanism, has made things worse- I think this is evident by the huge overall differences between ACA and non-ACA states.

Unfortunately, while Republican politicians might be tickled pink to show evidence that the ACA made things worse, many of the people who are more expert than me in discussions of Medicaid, insurance, and the opiate crisis are equally eager to see the ACA exonerated- the enthusiasm about the Health Affairs blog entry over the last day seemed a little overeager. (The head of Medicaid, for example, admonished “shady anonymous sources” yesterday for “slowing the debate.”)

What we really need is someone who is familiar with the prescription process and insurance regulation, who has a genuine commitment to objectivity and impartial weighing of evidence, perhaps someone who has treated drug addicts directly and understands how illegal and legal opiates interact, to do a deep dive of the evidence and give an unbiased opinion. Who could possibly fulfill all these criteria?

 

 

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s