20 October 2015


"A woman’s lifetime risk of dying from breast cancer is 2.7 percent without screening. Kerlikowske calculates that a woman who follows the new USPSTF guidelines could drop that risk to 2.0 percent, and one who follows the ACS guidelines may reduce the risk a few decimal places more, to between 1.8 and 1.9 percent. To get these benefits, the USPSTF program requires 13 total mammograms in a woman’s lifetime, and the ACS regimen will result in 20 breast X-rays.

In exchange for these risk reductions, 61 percent of women who have annual mammograms and 42 percent of women who have biennial mammograms will be called back at least once for follow-up tests that reveal they do not have cancer, researchers write in an accompanying paper. The anxiety and stress of such a false alarm is the most common harm, but it’s not the only one.

In its own analysis, the USPSTF calculated that if 1,000 women follow its advice and have a mammogram every other year from age 50 to 74, 146 of them will be subjected to unnecessary breast biopsies and 18 of the 1,000 will be diagnosed and treated for a cancer that would have never harmed them (a problem called overdiagnosis and overtreatment). Women who have mammograms more often, as they would under the ACS guidelines, will experience more of these downsides."

- Notice this is roughly the number of women who are dying of breast cancer at all with this regimen. And the unnecessary treatment rate is way more than the number that are being treated successfully to avoid lethal developments. All of that treatment comes with its own complications and issues, and stress and fear associated with possible diagnosis (even before considering the enormous financial costs it burdens us with to deal with all these unnecessary procedures).

What does this mean "scientifically"?
- We need much better detection methods. Mammograms appear to be little better than junk science as they have way too high of a false positive rate and way too small of a detection rate. It's possible this is a problem of there being a fairly small risk, but the lifetime risk here is significant enough that we should be doing something. This feels a lot like a "this is something" solution rather than a very useful one. You might as well flip a coin if it's close to 50% false positive rates to decide whether your doctor thinks you might have something that needs checked out.

This is often over 50%. A coin flip might be a better option. The author's decision to just forgo them entirely may be the wiser course in the meantime. (I personally like Austria's recommendation of "whatever").

- It might also be that doctors don't have any idea how to use them, or more likely use the data they get in order to help patients making informed decisions about their care and health. This is a widespread problem in medicine and it likely encourages over use of procedures and treatment, particularly of scary things like cancer. Cancer is scary. Thinking you have it, or could get it, is scary. Therefore, as much medical treatment to prevent it as we can afford is what most people think is reasonable in response. This is not actually reasonable as a response to the actual risks of cancer that most women will be afflicted with. Doctors should have a better idea how to discuss this problem sensibly. Also they may need to have some incentive to do so. The incentive right now appears to be closer to "generate breast cancer patients" rather than "prevent/detect/treat actual breast cancer when it appears in my patients".

- Looks a little better at detection relative to false positives after age 50 (not into the "good at it" rates, but significantly better than what we do now). This is essentially when the rest of the developed world even starts bothering with these questions. What might be one reason why they delay is it reduces the excess costs of unnecessary treatment without significantly reducing the number of women who are put at risk (possibly 1/1000).

More comprehensive studies have pointed out that this does not account for the number of women dying of any cancer total. Which is effectively unchanged by mammograms or no mammograms. All the focus on breast cancer may be obscuring other cancers that need to be attended to for the health of women (cervical cancer for instance has similar problems with our current detection regimen).

Or it maybe obscured here because our actual treatment options are quite poor.

- The biggest problem here is that threads in response to these changes in guidelines are inherently based upon anecdotal evidence. "I found a lump and I was 32" is treated as a data point against changing the current system. There are potentially higher risk factors as well (genetics for example). Patients don't really understand these, usually (many women will, but not everyone on an open comment thread does). Doctors might, or at least should, and can screen based upon them. Usually the recommendations allow for these deviations. The political economy problem is that many people will then worry if insurers will pay for it if it is "not recommended".

08 October 2015

Don't just stand there do something.

I've noticed a common quotation for (some) libertarians. It comes from "Yes Minister." We must do something, this is something, lets do it".

One intended aspect of this position is to say that in times of crisis or suffering or peril, there is a strong compulsion to be seen to act on a problem (and presumably to take credit for having successfully solved that problem), whether or not those actions have anything to do with the causes of strife under which we are laboring. Libertarians, at least those concerned with public policy, are fond of deriding this variety of common and very public thinking I suspect because it implicitly suggests we are running around doing a lot of foolish things and probably not fixing things via the public policy routes that are often preferred. Sometimes this is undoubtedly true. It is one of the reasons libertarians are potentially quite useful to the political process is they can quickly point out where the system sucks, finding low-hanging fruits that need to be plucked out.

The impulse of this insight and thinking is pretty strong and there are often sound objections to a great many popular responses; from opposition to same sex marriage, to demands to drug test welfare recipients, to protectionism in borders or trade, to some forms of gun control, and so on. This does not make these objections automatically correct over and above potential gains, but it often helps clarify the proposed gains to be something other than the stated goals. So for instance, much complaining about immigration turns out not to have much to do about jobs, which are not really impacted anyway. But instead vague worries about culture. As though either a) historically "American culture" has dramatically changed via massive waves of immigration far larger than we have had in the last 3-5 decades and b) getting more "foreign" restaurants around is a bad thing, suggesting that if there are changes, they're likely to be pretty good things. This is one of those scenarios where the impulse is a very good one. Leave it alone at worst, and if anything, make it easier to get here and live here. If we must do something, doing the wrong thing is worse. Pushing against doing the wrong thing, restricting immigration further, is simply a nice side benefit.

One of the stronger logical basis for this kind of thinking is to point out not merely that the proposed idea may not do much about a problem, or will identify the wrong thing as a problem in the first place, as with the case of immigration reforms demanded by anti-immigration groups, but also to suggest that a proposed idea has "unintended" consequences that will be harmful and present themselves as substantial costs over and above what we are intending to do. Much anti-immigration policy would fall under this category. Much anti-trade policy likewise. Terrorism policy post 9-11 largely falls into this category as well.

More paranoid positions suggest that these "unintended" consequences are intended. I'm more of the opinion that they are often exploited by people who have their own agendas within the bureaucratic chain than that they are intended policy shifts. I don't think governments are fundamentally evil. But they're not fundamentally good and decent either.

A more challenging subject.

I've written quite about skeptically about gun control proposals. I have maintained relative agnosticism about background checks and their effectiveness. I think they are possibly helpful, but there are a lot of reforms I would propose within them. I'm not sure a felony conviction should be indicative of anything on its own as we have over-criminalized a lot of behavior that is not fundamentally violent (drug distribution for example). Many other popularly proposed ideas, I'm less positive on still even from this rather tepid position of endorsement of the status quo, or what amounts to it, to say that I'd rather do nothing than try them as they seem very much less tied to the proposed problem and very much more in the category of "we must do something" thinking that often leads to unintended consequences. "Assault weapons" seem more to do with cosmetic features of weapons than the causes of violence, as rifles in general are rarely used in the commission of violent crime and death and fundamentally a gun that propels a shaped fragment of metal at high speed and potentially into or through human tissue and bone in a lethal way isn't that different from any other type of weapon. Cosmetic adjustments aren't the issue. "Mental health" doesn't have much to do with anything we can put into policy either as it is too hard to determine who might be violent without reliance upon who has already been violent, which is probably a more reliable marker. Violent video games have been around for 20 years or more in mass markets, and that's precisely the same span that we have seen a large decline in violent crime (this is also true of pornography and rape). All this stuff always comes up and swirls around. Most of it smells like an awful backed-up bathroom toilet by now as it's been so heavily recycled and overused. It sucks up valuable attention and time to discuss these questions.

One of the problems with this system seems to be that it is frequently tied in public debate to the question of what to do about these horrible shootings we have that make national news for killing a handful of people at one time at a school or a church or some other public place. But it is not abundantly clear that backgrounds checks, the more popular suggestion of something to do about gun violence in general, have much of anything to do with helping us prevent these. People who become deranged or angry enough to want to try to casually kill possibly dozens of strangers don't have to have had any outward signs of mental disorder and distress beforehand. In part because not everyone has sought treatment, but also in part because I'm not convinced this is directly related. It's not a universal truth that everyone who kills other people in these events was determined to be some variety of what we have popularly termed as crazy. They don't have to have had a criminal record. They don't have to have had these as problems at the time they acquired the weapons perfectly legally, possibly years before.

We also don't have a very good model that most of these are things that would help us either. Mental illness is virtually impossible to note as having a direct link to violence and mayhem. If it has one, it's more likely as victims of violence and mayhem than as a causal agent itself and there doesn't appear to be any meaningful way we can screen out who is or is not a threat, leading a lot of misguided fear-based policies and potentially stripping millions of people of basic civil rights and decency if policies are not done very, very carefully. The types of policies that are unlikely to be crafted in the wake of major national news stories. As pointed out above, a criminal record doesn't have much either to suggest much in many cases that we will be dealing with a violent person who shouldn't be trusted around firearms. Finally, I'm not sure there's a good and safe way to go in and by some means of force remove someone's firearms they have purchased legally on the idea that they had developed some kind of noticeable mental defect that could lead to trouble. There are undoubtedly good ways to approach this as a problem, but they are less likely to do with requiring people to surrender weapons/property they have legally purchased in times of personal crisis en masse and more likely to do with people discussing their problems with a doctor and friends or family who could do things like help to secure such weapons privately (and voluntarily) while someone seeks help. That isn't something that is easily done through national policy. Which is to say: it might be that some one should do something about these as a terrible problem, but it might not be the politician who gets to take credit for it if it works. It would be ordinary people instead who stepped forward to help and they who achieved some successful change in society.

Spending a lot of time focusing a policy agenda upon these events seems like a rather poor use of the politicizing attention. Spending a lot of time focusing on that there's a whole huge mess of other gun violence that goes on (still), or that there are thousands of suicides with firearms every year. That seems like a better use of the time. Bring that up and talk about it. Spin it somewhere we can more effectively do something or discuss different political solutions. That's where the conversation on policy still has ground to cover for now.

Incomplete list of complicated things that get discussed with vaguely described solutions

1) Gun Control - "common sense gun control" means? So far I've heard this used to describe a lot of things, but mostly it gets used to mean "people shouldn't have guns". That's a sentiment I agree is preferable for a modern civilisation is that it not rely upon mass armament or fear of crime and that the general public can go about their business freely and safely without any concern for whether or not they are armed or not (a condition which I believe already exists in probably 90% of American towns and suburbs, and most major cities). I think we are going to disagree about what methods are appropriate for most people not to have guns. It helps to lay out a specific idea here and not just assume everyone knows what you mean. Strictly enforced and expanded background checks? Gun safety training? Gun buy-backs? What exactly are these common sense regulations everyone supposedly agrees we should do?

I wrote about this problem several years ago and unfortunately little has changed in the tenor and tone of the debate in any way, and very little of the debate focuses on the largest sources of problems (violence often related to black market drug distribution, and suicide) while an enormous amount focuses on these (still) fairly rare events that become national news. Enacting the sorts of policies that would significantly reduce the former problems will probably look very different than anything that would deal with the latter, if there is anything that can be done at the policy level. They look like very different issues at this point. But the former problem is way, way worse.

A brief addendum to the things I wrote before would be that the types of policies likely to impact gun violence in a meaningful way, in my view at least, would be things that would take some time to see some results. Alcohol taxes could be raised. Mental health training for police could be dramatically overhauled and mental health care improvements would probably start to cut into suicide rates. Various narcotic street markets could be priced out by providing safer and legal markets for the substances being traded. Background checks may take years before they began finding points of sale that in some way meaningfully restricted gun ownership and at considerable regulatory cost as the existing market and pool of available weapons is too high that transfers can be done privately without much monitoring. And so on. Even among the things that are far more radical in the American context and much less clearly common sense (assault weapons bans, gun buybacks, etc), we would probably not see much impact for many years, if at all.

We should not expect that if anything is done it will stop mass shootings from happening.The causes of those are likely to be all over the place that they will be difficult to respond to with single effective policies. For instance, many of these "common sense" reforms have had little to do with the actual shootings (background checks for example seem to have had mixed results, as one should expect). That is not by itself a reason not to do them. It is simply to say that the idea that we should tie policy responses to fairly rare events is probably a bad way to try to sell it as a good and necessary change (as we did with terrorism and as Obama explicitly tried to justify over a trillion dollars of spending to deal with terrorism over the last decade, which I disagree was money well-spent). We should expect that if anything is done we might see reductions in the rate of violent crime and suicide. That would be the yardstick I'd measure it against. Comparatively speaking, measuring the rates of violent crime in other countries is a more interesting study than looking at sensational events in smaller industrialized democratic countries and their responses. It's also very much like the "there are no giants here" theory, in that if we take a very rare situation and adopt often radical policies in response to it as a serious problem, and then nothing happens for a long time, we often hear "well see, we got rid of the giants, you don't see any around do you?" as though the cause of this absence was the radical policy shift. I think this is dubious in the case of terrorism. A more likely cause is that the number of people willing and capable of acting out terrorist deeds is extremely small. Just as the population of people willing to commit mass shootings is.

Shorter-term the change I would make to deal with mass shootings would be simply to have the media cover the shooter much less, cover the victims. The academics and police can poke into the shooter and their background to figure out what's going on there. I don't give a shit anymore who these people are. They don't seem that different from anyone else fundamentally that we have much of a pattern to identify who dangerous and disturbed people are beforehand and thereby make any effective policy change. So we could change the culture instead and start ignoring their deeds and focus on the effects for now. Maybe that will make it less likely someone shoots other people for attention. Maybe not. It's a better use of our time to remember people who lost their lives senselessly and their names than the people who took them. I'm not quite sure why Jack the Ripper is well known for example. Or Ted Kasinzski. Obviously it's not a new problem but it's one that I'd be quite happy to see a shift on.

2) Family Leave - "people should afford their own children" to me means "poor people shouldn't have children". That's not why people have children is a financial decision (if anything, there are relatively cheap things we could do that more poorer people could have fewer children if they wished, but also make it easier for people to have children if they do wish to).

This is barely disguised socially Darwinian thinking. It's very similar to the "we should drug test welfare recipients" idea that I routinely bash upon as idiotic and wasteful, not to mention morally reprehensible, particularly for self-described liberals. Speaking of which, yet another state has failed to materialize any savings or significant effect from doing that.

3) Syria. "Help the Syrians more actively, establish a stable regime" - That's a nice idea. I agree that would be awesome if Syria wasn't involved in a civil war and had a stable moderately democratic government. This would be a lot better for the Syrian people than the status quo of the last several years.

But. This is kind of like the underpants gnome plot as applied to international relations. Resolving multi-factional violence is a difficult problem on its own. It is not resolved easily by having us picking sides or trying to make friends in the midst of it. It is not clear to me that "taking an active role" (which is vaguely defined as "go bomb people we don't like") would be a significant improvement on the ground, particularly if it is not followed up with. Which is the more challenging part still than getting people to stop shooting each other. Often bribing them works pretty well on that front. Following up with that to establish some kind of peaceful resolution with inclusive governments forged from separatist factions that must work out their differences within a country is very hard. In fact, establishing any form of stable democratic regime in foreign countries is extremely hard as an international relations problem. To the point that I'm not entirely sure we have any idea how to do it. Japan might be the only successful example in the last century and that took a massive war against a world power with a fairly developed economy and the institutions to go with it that lasted, in one form or another, over a decade (going from Japan's fighting with China and the USSR), with another decade of military occupation (followed up by decades more of military "cooperation"). Point being we had moderately good conditions in which to forge and impose a democratic rule, a clear international sanction to do it, and it still took decades of work. We would have few of these building blocks in Syria (or Iraq or Afghanistan or Yemen or Kosovo or Libya), and limited international support to do so. Which means we could very well end up right where we started. Basically what happened with Iraq.

I'm unsure why there is still much enthusiasm for this in the international relations policy elite. Yes it is a terrible problem. No it isn't clear to me you have proposed a plan that a) solves it and b) does so at an acceptable risk, for example how your solutions or interests should differ on this from other international relations concerns that might be far more damaging to American power and security (say, the Ukrainian crisis). I still hear people thinking we should have a no-fly-zone. When the Russians are now operating air strikes in country? Really? This is a good idea? Risking war with a nuclear power? To carry out a vaguely defined mission that has little impact upon American security concerns? If the concern is humanitarian, take in a million refugees. We could do that in a year. I'd frankly be even greedier and take in all of the refugees if they want to come here. All of them if necessary. We can do something about the suffering and condition of these people immediately and easily. We cannot do something about the underlying conditions that created that suffering very easily and immediately. That doesn't mean we cannot or should not. It means the vaguely defined missions and interests I've heard being bandied about aren't very compelling cases that the people who would be in charge of conducting such operations have a clear idea how to do it and would succeed, or at worst would not make things worse by running about. Until they sound much less like the Green Lantern theory of international politics or the underpants gnomes, I'm probably going to be very skeptical that someone has a handle on things and should run with it.