Cigarette Daydreams

Last week, U-Haul announced that beginning February 1 of this year it would no longer hire nicotine users in the 21 states that permit employers to take that information into account when hiring. To be clear, this covers use outside of the workplace, and it could affect former smokers who use nicotine patches or similar delivery systems.

If you didn’t hear about this, you have my envy and my respect, as it probably indicates that you’re employed and/or don’t spend hours on Twitter, where the topic was briefly trending.

From what I can tell, the reaction has mostly been cynicism about U-Haul wanting to cut healthcare expenses. I think that’s probably true; smokers can indeed be charged higher premiums than non-smokers under the Affordable Care Act, and presumably U-Haul will cover at least some of this.

But since I spent a small chunk of last year’s blog posts talking about selection effects and signalling (most notably here, here, and here), I feel obliged to point out that selecting against smokers is also an effective way to screen for other undesirable qualities in employees. As smoking rates have plunged, smoking has become an increasingly good proxy for a bunch of socioeconomic factors.

(Pardon the use of 2013 data in the following charts; I happened to have this version of the BRFSS on my laptop, and went with it for convenience. I don’t think there’s a good reason the general idea should have changed much in the interceding years.)

For example, smokers, and to a lesser extent, former smokers, miss more days due to their physical or mental health than non-smokers:

Smokers miss more days
The examples given in the BRFSS codebook of “usual activities” are work, self-care, or recreation.

They’re also more likely to have less education and to have lower incomes:

Educational attainment by smoking status

Income bracket by smoking status

Notably, U-Haul is pitching this as a step in “fostering a culture of wellness”:

“We are deeply invested in the well-being of our Team Members,” said U-Haul Chief of Staff Jessica Lopez in a press release. “Nicotine products are addictive and pose a variety of serious health risks. This policy is a responsible step in fostering a culture of wellness at U-Haul, with the goal of helping our Team Members on their health journey.”

So here we have a company policy that is ostensibly for the benefit of employees’ health, but the actual consequences of which will be to save money for the employer and disproportionately preclude the candidacies of many low-SES applicants. Mind you, this is for a job at U-Haul during a period of supposedly record-low unemployment.

I think what bugs me most about this is a feeling (which could be wrong but is at least widespread) that there are diminishing opportunities—especially for low-SES Americans—to participate in the productive side of the economy, which by all indication is something that gives people a sense of meaning and self-respect.

I also can’t shake the feeling that a more powerful or organized constituency would be able to generate some public sympathy in a similar situation. If U-Haul made it policy to deny employment to obese people, presumably to similar effect, there would have been a cascade of outraged NYT opinion pieces and an ACLU lawsuit.

Still more hilarious is the contrast with growing corporate, public, and governmental acceptance (or even endorsement) of marijuana use. A new, publicly announced prohibition on employee use of marijuana would come across as horribly retrograde and likely receive more negative attention.

Smokers, alas, are nearly universally reviled, out of the graces of the upper classes and on the wrong side of demographics.

Obamacare’s Complicated Relationship with the Opioid Crisis

The opioid epidemic, having evolved into one of the greatest public health crises of our time, is a contentious aspect of the ongoing debate surrounding the Republican healthcare proposal.

Voices on the left worry that the repeal of Obamacare’s Medicaid expansion and essential health benefits would worsen the crisis by cutting off access to treatment for addiction and substance abuse. Mother Jones and Vox have both covered the issue. Former President Obama’s June 22nd Facebook post stated his hope that Senators looking to replace the ACA would ask themselves, “What will happen to the Americans grappling with opioid addiction who suddenly lose their coverage?”

On the other side of things, there are theories that the Affordable Care Act actually helped finance–and perhaps exacerbated–the opioid epidemic. It goes something like this: Expanded insurance coverage was taken as a primary goal of the ACA. Two of its policies that supported that goal–allowing adults up to 26 years of age to remain on their parents’ insurance and expanding Medicaid to cover a greater percentage of the population–unintentionally connected at-risk cohorts (chronically unemployed prime-age men and young, previously uninsured whites with an appetite for drugs, to name two) with the means to obtain highly addictive and liberally-prescribed pain medications at a fraction of their street price. (Some knowledge about labor force and other social trends helps paint a clearer picture on this.) Once addicted, many moved on to cheaper and more dangerous alternatives, like heroin or synthetic opioids, thus driving the growth in overdose deaths.

This is a really interesting, if tragic, narrative, so I decided to take a look. I focused on state-level analysis by comparing CDC Wonder data on drug-induced deaths with Kaiser Family Foundation data on expansion status and growth in Medicaid enrollment. The graph below plots states based on their rates of growth in Medicaid rolls and overdose deaths from 2010 to 2015, and is color-coded for expansion status: blue for states that expanded coverage before or by 2015, yellow for states that expanded during 2015, and red for states that hadn’t expanded by the end of 2015. (A note: this isn’t an original idea; for a more in-depth analysis, check out this post.)

What’s interesting is the places where overdoses have increased the quickest. The fastest growing rates of overdose deaths were mostly in states that had expanded Medicaid by 2015; the only non-expansion state to grow by more than 50% since 2010 was Virginia, which in 2015 still had a relatively low rate of 12.7 fatal overdoses per 100,000 population. For some perspective on how bad things have gotten, that rate would have been the 19th highest among states in 2005; today Virginia ranks 42nd in terms of OD rate.

On the other hand, there isn’t a noticeable correlation between increases in Medicaid coverage and increases in the rate of fatal overdoses. Additionally, the rates of overdose deaths in expansion states were increasing before many of the Affordable Care Act’s key provisions went into effect. Starting around 2010, there was a dramatic divergence between would-be expansion states and the rest. It’s possible that states with accelerating rates were more likely to expand coverage in response to increased frequency of fatal overdoses.

So what’s the deal? Did the Affordable Care Act agitate the opioid epidemic? Obviously I don’t have the answer to that, but here’s my take:

I think it would be difficult to argue it hasn’t been a factor on some level, given the far-higher rates of prescription, death, and opioid use and among Medicaid patients than the general population, as well as the state-level trends in OD rates (with acknowledgement that state-level analysis is pretty clunky in this regard; for many reasons West Virginia’s population isn’t really comparable to California’s). I think the fact that state Medicaid programs are adjusting regulations for painkiller prescriptions is an acknowledgement of that.

But if the ACA had a negative effect, I’d think it must register as a drop in the bucket. There are so many pieces to this story: lax prescription practices and the rise of “pill mills,” declining labor force participation, sophisticated distribution networks of Mexican heroin, bogus research and marketing on pain management, stark disparities between expectations and reality. It’s nice to think there’s one thing we can change to solve everything, but I don’t think we’re going to be so lucky.

Here’s another twist: Even if the Affordable Care Act had some negative impact, it could very well be that ACA repeal could make things worse. Scrapping essential benefits coverage could lead to a loss or reduction of access to addiction treatment for millions of Americans. Moreover, gaining insurance has been shown to alleviate feelings of depression and anxiety. How then, might we guess 20 million Americans will feel after losing their insurance? Given the feedback loop between pain and depression, this question deserves a lot of deliberation.

Insurance Coverage Numbers Are Important, But Not All-Important

Whether you’re into this sort of thing or not, you’ve probably been hearing a lot about healthcare policy these days. Public debate has roiled as Republican lawmakers attempt to make good on their seven-year promise to repeal and replace the Affordable Care Act (ACA). As the debate rages on, one metric in particular appears to hold outsize importance for the American people: the number of Americans covered by health insurance.

Analysis by the Congressional Budget Office, which showed that 14 million more Americans could lose coverage by 2018 under the Republican replacement, caused intense public outcry and was frequently cited as a rationale for not abandoning the ACA. There is immense political pressure not to take actions that will lead to a large loss of coverage.

But here’s the thing: the relevant metric by which to judge Obamacare isn’t insurance coverage numbers. To do so is to move the goal posts and place undue importance on a number that might not be as significant as we imagine.

The ultimate point of health insurance, and the implied rationale for manipulating insurance markets to cover sicker people, is that people will use insurance as a means by which to improve their health, not just carry a plastic card in their wallets.

Health Insurance ≠ Health

The impulse to use insurance coverage as a proxy for health is misguided but understandable. For one thing, it’s a simple, single number that has dropped precipitously since the implementation of the ACA; that makes it a great marketing piece for supporters. For another, health insurance is the mechanism by which most of us pay for most of our healthcare.

And yet in 2015 the uninsured rate fell to 10.5% (down from 16.4% in 2005) while age-adjusted mortality increased for the first time in a decade.

It turns out a nominal increase in the amount of insured Americans doesn’t necessarily translate into improved health outcomes for those individuals. A newly released paper from the National Bureau of Economic Research (NBER) finds that while the ACA has improved access to healthcare, “no statistically significant effects on risky behaviors or self-assessed health” can be detected among the population (beyond a slight uptick in self-reported health in patients over 65).

These results are consistent with other studies, like the Oregon Medicaid Experiment, which found no improvement in patients’ blood pressure, cholesterol, or cardiovascular risk after enrolling them in medicaid, even though they were far more likely to see a doctor. There were, however, some notable-but-mild psychic benefits, such as a reduction in depression and stress in enrollees.

In short, despite gains in coverage, we haven’t much improved the physical health of the average American, which is ostensibly the objective of the ACA.

Why Not?

To be fair, the ACA is relatively young; most of its provisions didn’t go into effect until 2014. It may well be that more time needs to pass before we start to see a positive effect on people’s health. But there are a few reasons to think those health benefits may never materialize–at least, not to a great extent.

A lot of what plagues modern Americans (especially the poorest Americans) has more to do with behavior and environment than access to a doctor. Health insurance can be a lifesaver if you need help paying for antiretroviral medication, but it won’t stop you from living in a neighborhood with a high rate of violent crime. It won’t make you exercise, or change your diet, or stop you from smoking. It won’t force you to take your medicine or stop you from abusing opioids, and it certainly won’t change how you commute to work (that’s a reference to the rapid increase in traffic deaths in 2015).

Here’s something to consider: A lot of the variables that correlate to health–like income and education–also correlate to the likelihood of having health insurance. If we want healthier Americans, there may be more efficient ways to achieve that than expanding insurance coverage, like improving employment and educational opportunities. Maybe something creative, like Oklahoma City’s quest to become more walker-friendly, could yield better results?

Of course, all things being equal, more insurance coverage is better. But nothing comes without cost, and as a society we want to be sure that benefits justify costs. So far, that’s not clear. This poses an existential question about our current pursuit of universal coverage, and, by extension, the relevance of coverage as a metric for the success of healthcare policy: If insurance isn’t the cure, why are we prescribing it with such zeal?