02/28/2018
Aaron CarrollThe idea that spending more on preventive care will reduce overall health care spending is widely believed and often promoted as a reason to support reform. Unfortunately, that doesn’t pan out in real life.
This episode was adapted from a column I wrote for the Upshot. Links to further reading and sources can be found there.
02/28/2018 Aaron CarrollThis post is part of a series in which I’m dedicating a month to learning about periods in history this year. The full schedule can be found here. This is month one. (tl;dr at the bottom of this post)
This month was a bit surprising in that I the parts I thought I would enjoy the most and the parts I thought I would enjoy the least were somewhat reversed. Let me start by saying that I wasn’t ignorant about Rome before I read this month. I took a gazillion years of Latin in middle school and high school, and I’ve even read Virgil’s Aeneid in the original Latin. Not well, mind you (I’ve got a great story about the AP Latin test where I mixed up a mountain and a tree, and… forget it).
I’ve also always loved the architecture of Rome, and am well versed on the period of Cicero, to Caesar, into the emperors. That said, there was still plenty to learn.
I started the month with Mary Beard’s SPQR, which is every bit as good as everyone says it is. She begins the book by focusing on Cicero, which is a great idea, and then backs up to the beginning (Romulus and Remus!) and moves forward through the emperors. That’s a lot of ground to cover, and she does it fairly well. One of the things I appreciated the most was how she spent time talking about how we know what we know. There are no videos or news reports, obviously. Much of what we get is through what was written down and survived. It’s critical to remember that there were no printing presses. Things has to be hand copied or written many times.
One of the reasons we know so much about Cicero is that, as a politician, he would print up many, many copies of his speeches. He knew not everyone would come to hear him, so he had some infrastructure to write those things down and distribute them to people. Smart. Also, good for history.
A couple key players in history also spent a lot of time writing their memoirs. That’s why we know so much about them. If you didn’t take the time to do that… well, then you have even less control of who tells your story.
Beard’s book gives you a pretty good sense of the time, and does a good job of making you realize that life in ancient Rome, while certainly more precarious than now (politicians get killed way too often for comfort and you spent a lot of time fighting in wars), it was also reasonably comfortable at times. More on that in a bit. She also reminds you that the periods we focus on (Caesar +) are only a blip in the history of Rome.
Next I read Mike Duncan’s The Storm Before the Storm, which I wish I’d read first. That’s because his book focuses on the period right up to Caesar et al, and is a much more in depth history of Rome before what we know. I mean WAY more in depth. There were a hell of a lot of wars, and a hell of a lot of political intrigue before the Republic.
Side note – one of the things I also got from these books was that Kings and leadership didn’t pass down through family (ie sons) the way we seem to assume they do now. The ancient Romans would have thought that was ridiculous. How do you know kids can rule as well as their parents? That idea came along much, much later, and it was somewhat surprising to me.
I was also stunned at how interesting the whole setup of government was as the republic took shape. Very corrupt, but also very stable.
On the other hand, the discussion of political norms, and how those were slowly chipped away (in both books) hits a bit close to home at times. The political maneuvers ring true in a number of ways, and there’s a lot to make you uneasy. Still, you are much less likely to get killed while trying to vote or get elected today than you were then.
Next up was Decline and Fall of the Roman Empire, by Peter Gibbons. I didn’t like it. Too dry. I tried and tried, and then decided to move on and come back later. So I moved on to The Fall of Rome: And the End of Civilization by Bryan Ward-Perkins. I was somewhat unprepared for how interesting the fall of Rome would be.
I used to think of “civilization” as meaning big buildings, big armies, and a stable government. But it’s the economic stuff I took for granted that mattered. Because roads were so good and trade was so robust, people would stop trying to specialize locally. The would get amazing pottery from far away, for instance, so why bother to make it close to home. Food moved all over, so you didn’t need to farm as much. But as the empire collapsed, so did its trade. You couldn’t get that pottery anymore. You couldn’t get cheap and easy food. The quality of life of pretty much everyone dropped dramatically, not just in how much money they had but in what you could actually obtain. People had to go back into farming or starve. I wasn’t thinking about the fall of civilization properly.
And THAT’s why I’m even more excited for March. I want to read more about that, and how people pulled themselves back up. It was also good to sit and think about how the world’s superpower that hung around for like 1000 years completely fell apart. Why? I want to know more about that and what came after. On to March.
Oh, I tried to go back to Gibbons again at the end of the month, and still couldn’t fininsh the book. I know it’s a classic. Sorry.
tl;dr: If you want to focus on history up to the fall of the republic, read Mike Duncan’s The Storm Before the Storm. Mary Beard’s SPQR, is more of a broader review of Rome from the beginning through the emperors, but also great. More Fall of Rome next month.
02/26/2018 Aaron CarrollI’m going to spend March learning about the Fall of Rome and the Dark Ages. You’ve already also given me some great ideas. I want to post them here, so you can help me prioritize what to read. If you think I’m missing something, please go tell me. I’m opening comments, or you can tweet me.
What do you all think? Any thoughts on the order? Am I missing something good?
02/22/2018 Nicholas BagleyI’ve got a new piece at Vox digging into Idaho’s decision to flout the Affordable Care Act. If you want to learn something about Idaho administrative law (I know, I know, pure clickbait), this is the place to look. I also kick the tires on the analogy between what Idaho’s doing and marijuana legalization.
The upshot is that I wouldn’t count on the courts to intervene, even though what Idaho is doing is patently illegal.
If Idaho moves forward and other states follow its lead, what will emerge is a gray market in noncompliant insurance coverage, not unlike the gray market in legalized marijuana. Indeed, the marijuana analogy fits neatly. In both cases, state officials have purported to legalize conduct banned by federal law; in both cases, federal officials have been reluctant to enforce a law they disagree with.
And as with marijuana, what starts in one state will spread. As floutings of federal law go, Idaho’s approach is pretty measured. Under its rules, insurers that sell noncompliant plans must also sell compliant plans, the unhealthy can “only” be charged 50 percent more than the healthy, and insurers must cover preexisting conditions (unless there’s a gap in coverage, in which case they don’t).
But other red states that follow Idaho’s lead may not be so restrained. They might allow insurers to ditch their ACA-compliant plans, to exclude any and all preexisting conditions, or to jettison coverage for mental health care. Red states could take us back to the harsh pre-ACA state of affairs, and all without the need for congressional action.
So this story isn’t really about Idaho. It’s about every Republican-controlled state that’s waiting to see what happens next.
02/22/2018 Bill GardnerThe data on mass shootings show that the frequency of these killings has been increasing. Many of the killers have used military-style rifles, including the AR-15. The use of these rifles has prompted some people to call for a renewal of the federal ban on assault weapons, which was passed by Congress on September 13, 1994, and expired on September 13, 2004.
Critics of the proposed ban argue that the definition of an ‘assault rifle’ is vague and arbitrary. There is merit in that claim, although I am not sure how much we should care if the ban saves lives anyway. To the latter point, however, critics claim that the 1994 ban had little effect.
I can’t evaluate whether the ban worked. That would require time, data sets, and expertise that I lack. But I decided to look at my data set of shootings where more than four people were killed and see if there was any evidence about the effect of the ban.*
Below is a graph of the logarithm of the days between successive massacres, as a function of their dates. I have also plotted a loess-smoothed curve to look for trends in the data. The fall in the curve means that the time between events is decreasing, which is another way of saying that the frequency of massacres is increasing. The vertical lines mark the beginning and end of the assault weapon ban.
The accelerating frequency of mass shootings appears to have slowed while the assault weapons ban was in effect.
Taking the logarithm makes the increase in frequency look less dramatic: the untransformed data are much scarier. However, because all you see in the untransformed data is the speedup in the rate of massacres, it’s hard to see anything else. What this graph shows, however, is that the frequency of mass killings was increasing before the ban started and after it ended, but that it paused while the ban was in effect. Christopher Ingraham makes a related argument here in the Washington Post.
There are so many caveats needed here. First, this is a small dataset. Massacres are, thanks be to God, uncommon events (albeit becoming more common). Second, the coincidence of a slowdown with the assault weapons ban proves nothing. These data can’t tell us whether the ban caused the slowdown; it could have been something else in that decade. Finally, the slowdown in the increase in massacres means that lives were saved. That’s good and worth doing, but what we want is for the curve to trend up.
Bottom line: I can’t tell you that the assault weapon ban worked. But it may have had a small effect.
*Thanks to friend-of-the-blog Dr. David States for the prompt to look at this.
02/21/2018 Aaron CarrollEvery once in a while, we like to take a moment and focus on health systems around the world. Today, we’re looking at Taiwan, which made the transition to a single payer system kind of suddenly, and pretty recently.
This episode was adapted from a column Austin and I wrote for the Upshot.
02/21/2018 Austin FraktThe following originally appeared on The Upshot (copyright 2018, The New York Times Company).
Even before any proposed cuts take effect, Medicaid is already lean in one key area: Many state programs lack coverage for dental care.
That can be bad news not only for people’s overall well-being, but also for their ability to find and keep a job.
Not being able to see a dentist is related to a range of health problems. Periodontal disease (gum infection) is associated with an increased risk of cancer and cardiovascular diseases. In part, this reflects how people with oral health problems tend to be less healthy in other ways; diabetes and smoking, for instance, increase the chances of cardiovascular problems and endanger mouth health.
There is also a causal explanation for how oral health issues can lead to or worsen other illnesses. Bacteria originating in oral infections can circulate elsewhere, contributing to heart disease and strokes. A similar phenomenon may be at the root of the finding that pregnant women lacking dental care or teeth cleaning are more likely to experience a preterm delivery. (Medicaid covers care related to almost half of births in the United States.)
“I’ve seen it in my own practice,” said Sidney Whitman, a dentist who treats Medicaid patients in New Jersey and also advises that state and the American Dental Association on coverage and access issues. “Without adequate oral health care, patients are far more likely to have medical issues down the road.”
There are also clear connections between poor oral health and pain and loss of teeth. Both affect what people can comfortably eat, which can lead to unhealthy changes in diet.
But the problems go beyond health. People with bad teeth can be stigmatized, both in social settings and in finding employment. Studies document that we make judgments about one another — including about intelligence — according to the aesthetics of teeth and mouth.
About one-third of adults with incomes below 138 percent of the poverty level (low enough to be eligible for Medicaid in states that adopted the Affordable Care Act Medicaid expansion) report that the appearance of their teeth and mouth affected their ability to interview for a job. By comparison, only 15 percent of adults with incomes above 400 percent of the poverty level feel that way.
Some indirect evidence of the economic effects of poor oral health comes from a study of water fluoridation, which protects teeth from decay. It found that fluoridation increased the earnings of women by 4 percent on average, and more so for women of low socioeconomic status.
Other evidence comes from a randomized study in Brazil. In that study, investigators showed one of two images to people responsible for hiring: pictures either of a person without dental problems or with uncorrected dental problems. Those with dental problems were more likely to be judged as less intelligent and were less likely to be considered suitable for hiring.
The relationship between oral health and work has gained new salience in light of Kentucky’s recently approved Medicaid waiver, which permits the state to impose work requirements on some able-bodied Medicaid enrollees. It’s a step that some other states are also considering.
Medicaid takes different forms in different states, and even within states, different populations are entitled to different benefits. Though all states must cover dental benefits for children in low-income families, they aren’t required to do so for adults.
As of January 2018, only 17 state Medicaid programs offered comprehensive adult dental benefits, and only 14 of those did so for the population eligible for Medicaid under the Affordable Care Act. More typically, states offer only limited dental benefits or none.
Dental coverage under most private health care plans isn’t comprehensive, either — people who want it have to buy separate dental plans. But compared with those enrolled in private coverage through an employer or on their own, the population eligible for Medicaid is much more likely to need dental care and much less likely to be able to afford it or coverage for it. People with incomes low enough to qualify for Medicaid are twice as likely to have untreated tooth decay, relative to their higher-income counterparts.
Kentucky offers limited dental benefits to Medicaid enrollees, including those on whom work requirements would be imposed. Those benefits exclude coverage for dentures, root canals and crowns, which could challenge some enrollees’ ability to maintain good oral health and lead to greater emergency department use.
One study found that after Kentucky’s Medicaid expansion in 2014, the rate of use of the emergency department for oral health conditions tripled. Another study found that about $1 billion in annual emergency department spending was attributed to dental conditions, and 30 percent of emergency department visits for dental problems were made by people enrolled in Medicaid.
Other states that have proposed imposing work requirements as a condition of Medicaid eligibility include Arizona, Arkansas, Indiana, Kansas, Maine, New Hampshire, North Carolina, Utah and Wisconsin. Of these, only North Carolina and Wisconsin offer extensive dental benefits, while Arkansas, Indiana and Kansas offer limited benefits. (The definition of “limited” varies by state, but in all such states benefits are capped at $1,000 per year and cover less than 100 of 600 recognized dental procedures.) Maine, New Hampshire and Utah offer emergency-only benefits. Arizona offers none.
Though emergency-only coverage is less than ideal, it is better than nothing, as documented in a recent study based on Oregon’s Medicaid experiment. The study used a random lottery to offer some low-income adult residents eligibility for Medicaid. At the time of the study, Oregon offered dental coverage only for emergencies..
The study found that one year after the lottery, Medicaid coverage meant more people got dental care (largely through emergency department use), and the percentage of people reporting unmet dental needs fell to 47 percent from 61 percent. It also doubled the use of anti-infectives, which are used to reduce gum infections. Another study, published in the Journal of Health Economics, found that Medicaid dental coverage increased the chances that Medicaid-eligible people had a dental visits by as much as 22 percent.
It’s an accident of history that oral care has been divided from care for the rest of our bodies. But it seems less of an accident that the current system hurts those who need it most.
02/20/2018 Bill GardnerThe scale and frequency of mass killings have been increasing, and this is likely to continue. One reason — but just one — is that weapons are always getting more lethal. One of the next technical innovations in small arms will be the use of artificial intelligence (AI) to improve the aiming of weapons. There is no reason for civilians to have this technology and we should ban it now.
By lethality, I mean how many people you can kill in a short period. Lethality depends on many factors, including the weapon’s rate of fire, but not just that. Depending on the circumstances, a shooter with a highly accurate bolt-action rifle may be able to kill more people than one with a fully automatic but inaccurate weapon. Life, as they say, is a trade-off.
However, accurate shooting is hard. Bullets fall as they travel, so you need to estimate the distance to the target and compensate by pointing the barrel up. Likewise, bullets are blown by the wind, so you have to measure that and compensate for it. Finally, it’s difficult to hold a rifle stationary, but if you are even a minute of angle off of true aim when you pull the trigger, at 500 yards range you will be several inches off when the bullet reaches the target.
Here is a rifle that uses AI to increase rifle accuracy (see also here). It automatically carries out drop and wind compensation and times the moment of firing so that the barrel is pointed optimally to hit the target.
Made by TrackingPoint, a start-up based in Austin, Texas, the new $22,000 weapon is a precision-guided firearm (PGF). According to company president Jason Schauble, it uses a variant of the “lock-and-launch” technology that lets fighter jets fire air-to-air missiles without the pilot having to perform precision aiming.
The PGF lets the user choose a target in the rifle’s sights while the weapon decides when it is the best time to shoot – compensating for factors like wind speed, arm shake, recoil, air temperature, humidity and the bullet’s drop due to gravity, all of which can affect accuracy.
To do this, the PGF’s tracking system includes a computer running the open-source Linux operating system, a laser rangefinder, a camera and a high-resolution colour display in an integrated sighting scope mounted on top of the weapon. The user simply takes aim and presses a button near the trigger when a dot from the laser illuminates the target.
The computer then runs an algorithm using image-processing routines to keep track of the target as it moves, keeping the laser dot “painted” on the same point. At the same time, the algorithm increases the pressure required to pull the trigger, only reducing it when the gun’s crosshairs are right over the laser dot – and the bullet is then fired.
The Tracking Point XS-1 precision-guided firearm.
The US and other militaries are developing similar weapons. The Tracking Point rifle appears to be something that Americans can legally own. The technology will, of course, continue to improve and will get cheaper.
At some point, shooters with rifles like these will begin finding positions overlooking busy city streets, arenas, and schoolyards. Although they will fire at slower rates than the Las Vegas shooter did, far higher proportions of their shots will be fatal.
Or, Americans could decide that only the military and the police need these weapons. The time to make this decision is now before the devices get into circulation.
Someone will say that mass shooting are rare. Moreover, if a future schoolyard shooter can’t get an AI rifle, he will use an only marginally less lethal weapon. Thus, preventing civilians from legally owning AI rifles would save only a few lives and only trivially reduce the total of gun deaths. So, really, aren’t you just virtue-signalling?
So what? No one who isn’t serving in the military or police needs an AI rifle more than those future victims need their lives.
02/15/2018 Bill GardnerSome people argue that mass shootings in America result from mental health problem and require mental health policy solutions. Can this work?
Let’s think through the possible mental health policies for preventing mass shootings. I see three: 1) we could reduce the social determinants of mental illness to lower the population prevalence of mental disorders, 2) we could increase the availability of treatments for mental illnesses, and 3) we could attempt to identify the specific individuals likely to kill and get them into treatment.
1. Reduce the population prevalence of mental illness. Mental illness is associated with social adversity. The causality runs both ways: getting ill will hammer your life and, conversely, falling down the social gradient substantially increases your risk of getting ill. Providing more and better jobs and improving the social safety net would raise the well-being of Americans and, plausibly, reduce the population prevalence of mental and substance abuse disorders.
However, the causal association between mental illness and mass killing is weak. Few mentally ill people ever kill anyone. A mentally ill person is, at worst, only slightly more likely to be violent than anyone else. Conversely, it is not clear how many mass shooters were mentally ill. So even a substantial reduction in the prevalence of mental illness would have only a small effect on the number of mass shootings.
2. Increase the availability of mental health treatments. Let’s stipulate that if you are mentally ill and you are at risk of carrying out a massacre, mental health treatment might help you avoid this tragedy. Access to mental health treatment could be increased by training more evidence-based mental health providers, insuring the uninsured, and requiring that health insurance cover mental health treatment.
Unfortunately, the effect of increased access on mass shootings would be limited, because a) it’s likely that many potential shooters are not mentally ill; b) even with improved access, not all mentally-ill potential killers will seek treatment; and c) mental health treatment doesn’t always work.
Policies 1 and 2 are eminently worth pursuing because they would reduce mental illnesses and the suffering they entail. However, they would be expensive, and few of the politicians who talk about mental health as a response to mass killing support these policies. In any event, these strategies would have at best small effects on mass murders.
3. Identify likely mass shooters and deliver mental health care to them. This policy is a non-starter because of the mathematics of prediction. Murderers are too rare in the population. Any conceivable prediction model will generate overwhelming numbers of false positives. There is no Minority Report future world.
In summary: America needs better mental health care. However, the nation is unlikely to make the required effort, and if it did, it wouldn’t have much effect on mass shootings.
I have updated my graphs of mass shootings to include yesterday’s killings in Parkland, FL, but nothing in the overall pattern has changed. This graph plots the number of deaths in shootings that killed more than four people.
This plot only labels shootings that killed 20 or more, so Parkland with only 17 doesn’t get a label. It is, oxymoronically, a routine massacre. As someone noted on Twitter yesterday, it’s a bitter irony that the 1929 Valentine’s Day Massacre involved only seven murders.
The next graph plots the logarithm of the number of days between successive mass shootings against time and adds a smoothed curve. The declining curve starting in about 2007 indicates that mass shootings are happening increasingly frequently.
02/15/2018 Austin FraktLast week, the American Hospital Association (AHA) posted a critique of a recent, NEJM-published study by Sunita Desai and Michael McWilliams examining the effects of the 340B Drug Pricing Program, which has the goal of enhancing care for low-income patients. That AHA critique links to a methodological review by economist Partha Deb of the study’s methods. (Partha Deb was kind enough to speak with me and disclosed that he was compensated by the AHA for his time to prepare his review. At the time we spoke, that financial relationship was not disclosed in his online review, something he said he would try to correct.)
The study used a regression discontinuity design, exploiting a threshold in the program’s eligibility rules for general acute hospitals — hospitals with disproportionate share hospital (DSH) adjustment percentages greater than 11.75% are eligible for the program. The study findings suggest that the 340B Program has increased hospital-physician consolidation and hospital outpatient administration of intravenous and injectable drugs in oncology and ophthalmology, without clear evidence of benefits for low-income patients.
As Desai and McWilliams note in their paper, the study had several limitations. One that holds for all regression discontinuity studies is that the estimates pertain to hospitals close to the threshold. Another is that the study relies on data from Medicare and the Healthcare Cost Report Information System. The AHA claims that these limitations and other issues raised by Deb constitute “major methodological flaws” that “negate” the study’s findings.
In a subsequent post, the AHA suggests that the study was unnecessary because the authors could have just asked hospitals how they were using resources generated from the 340B Program.
The authors’ sent me a response to these critiques, which I have agreed to host here at TIE. I will let that response speak for itself. Read it.
But I do want to make two additional comments. First, the notion that we should only learn how a program works by “just asking” those that participate in it or benefit from it is absurd. To be sure, much insight can be gained from such qualitative work. But independent, objective, quantitative work is also essential to unbiased assessment of programs. I reject the AHA’s dismissal of research on these grounds.
Second, there is something troubling to me about advocacy organizations hiring top academics to critique specific studies. The potential for conflicts of interest is obvious. That is not to say there is anything wrong with Deb’s critique, but it is hard to know, in general, what role the financial relationship plays in these kinds of situations. At a minimum, that financial relationship should be disclosed.
