There is a hazard associated with reading up on something for a few hours and then dashing off an article — you may miss something important, making it obvious to all that you have no idea what you’re talking about.
The upside of making poor arguments for a living is that you do pick up some skill in recognizing them, which leads to a sort of heuristic. That is, it can take ages for me to know for sure that I know what I’m talking about, but it takes no time at all to realize that you do not. Especially if you write for Vox.
In other words, I think that Obamacare — specifically, Medicaid — is killing people, although I can’t be certain that there isn’t another, better explanation of recent overdose and mortality numbers. It’s simply the best explanation available for what’s going on.
However, I can be quite certain that folks arguing the opposite — that Obamacare is saving the lives of the tens of thousands — don’t know what they’re talking about. They are almost all arguing from theory about what ought to happen when Medicaid is expanded, rather than actual data about what’s going on. In reduced form, their argument is that because one thing happened in New York 15 years ago after a Medicaid policy change (mortality fell), we should ignore what is happening around the country now following a Medicaid policy change (mortality is on the rise).
The only exception to that I’ve seen is a study published last week, showing a possible statistical burp downward in cardiac arrests in Portland among the “middle‐aged population exposed to insurance expansion.” There’s the empirical pro-Obamacare case — a tick at the edge of the margin of error, and they’re not even talking about adults who actually gained Medicaid, just adults “exposed” to the expansion, some of whom might actually be insured.
Look at this Vox article, especially the great big chart in the middle presuming to show state-by-state mortality figures if the Senate Republican Obamacare replacement becomes law. (Look out, Indiana — 326 of you are going down!)
The specificity of the claims, not to mention the Harvard credentials of the authors, might convince you there’s something to this chart, but it’s really nothing other than an exercise in long division. The authors simply take estimates of coverage losses by state (which were derived from dubious Congressional Budget Office scoring) and then divide each number by 830.
So Alabama has an estimated coverage reduction of 480,500; 830 into that is 579, your magical “excess death” total. And so on for all the states.
That 830 number comes from the work of Harvard professor Benjamin Sommers, who, like Obamacare architect Jonathan Gruber, is a paid consultant to the assistant secretary for planning and evaluation at the Department of Health and Human Services. It represents the “number to treat,” meaning the number of new patients that he calculates would need to be insured in order to save one life.
Now, I’m not qualified to dispute Sommers’ methods or findings, but I can point out that his work isn’t nearly specific enough to answer the question of Obamacare and drug overdoses. Sommers has done two important studies comparing Medicaid expansion in New York, Arizona, and Maine in the early 2000s to neighboring states that didn’t expand. As Oren Cass has noted, “only one of the three achieved a statistically significant reduction in mortality.”
If you average the three states, of course, you get a number ostensibly showing that Medicaid reduces mortality rates. Do some more math and this turns into your “number to treat.” But as Sommers acknowledged, “the results are largely driven by the largest (New York), so our results may not be generalizable to other states.”
This is the source for Vox’s bloody arguments — tens of thousands will die because of you! And we should call this the Vox school, as it was Ezra Klein who started doing it back in 2009, when he was still just an overconfident blogger for the Washington Post.
If you’re on Vox’s side, in other words, you are basing your answer to whether Obamacare is contributing to the opiate epidemic on the fact of a general decline in morbidity in New York more than a decade ago. If you want to do that, God bless, but there’s a much more interesting analysis done on the question by a pseudonymous blogger called Spotted Toad, which was drawn to my attention after I wrote about this topic last week, mentioning the need for some regression analysis.
My sophisticated technique was to gape at a map and go, “Looky there.” Turns out, when’s something’s obvious even at that level, it gets really obvious with a proper analysis. Ross Douthat of the New York Times has spotlighted Spotted Toad’s work a couple of times, but the work deserves more attention.
Like Sommers did and like I did in my way, Toad starts by comparing states that expanded Medicaid to the holdout states, and finds stark discrepancies. But then Toad goes much deeper. After all, there are cultural differences between states that muddy the waters. So our analyst compared counties within the same state based on their increase or decrease in the uninsured rate, finding that a “county with a one percentage point higher decrease in the uninsured rate had a crude death rate from drug-related causes in 2015 of about 2 per 100,000 higher, relative to other counties in the same state.”
That’s huge. For context, the national overdose rate increased from 12.3 per 100,000 population in 2010 to 16.3 in 2015, according to a study by the Centers for Disease Control. The correlation between expanding access to Medicaid and drug overdoses persisted even when prior overdose rates were controlled for. Same after Toad controlled for the demographic variables most closely associated with overdoses — think unemployed white adults with Oxy prescriptions. In 2015, Medicaid expansion states had 25 percent higher overdose rates than holdout states, after baseline characteristics were controlled for.
Spotted Toad’s conclusion was that Obamacare boosted overdose rates by 4 to 6 per 100,000 in Medicaid expansion states. That would account for the majority of the problem. Toad points to Sam Quinones’ award-winning book, Dreamland, to explain how Medicaid provides opiate prescriptions worth thousands of dollars to poor people for as little as $3 co-pay. Some get abused; some get resold.
Now, the divergence between the two groups of states started in 2010, accelerating in recent years, after Medicaid actually expanded. Toad hypothesizes that the trouble started when young people first became eligible to remain on their parents’ insurance well into their drug-taking years.
There is one fact that changes the picture a bit — fentanyl and heroin are to blame for the overdose spike between 2014 and 2015, not OxyContin and related drugs known as SSIs. There’s still a big problem with fatal SSI overdoses in the U.S. — 12,159 of them in 2014 and 12,727 in 2015. But heroin is surpassing it — 10,574 in 2014 and 12,989 in 2015 — while fentanyl is closing in fast, with a year-over-year boom in its category of 5,544 to 9,580.
Fentanyl, which killed Prince, is apparently quite easy to get online from Chinese suppliers, and it’s extraordinarily toxic. A dose the size of a grain of salt will get you high, but two grains can kill you.
What the numbers suggest to me is that there’s some truth to the old notion of gateway drugs. That’s not in regards to marijuana, of course, as we’ve all known for decades that the vast majority of marijuana smokers never go on to using hard drugs. (Actually, it seems many people would prefer good weed to risky opiates.)
It seems quite possible that restrictions meant to curb opiate abuse are pushing people to seek riskier drugs from other sources. Say what you will about the dangers of Oxy from the factory — at least the standard dosage is reliably non-fatal, unlike the concoctions people are making with their Chinese granules.
I’m not sure what the answer is, but there’s got to be an alternative to subsidizing someone’s addiction to a drug, and then prohibiting him from taking it in safe doses.