I kind of owe an article on how I would cut government health care expenditures in half by improving the efficiency of care. Unfortunately if I write everything I want to write, that article is going to be a long time in coming. So here's a "short" article on why I think a large fraction of all medical expenditures in the U.S. - like about half - are probably waste and could be cut out completely without affecting the quality of care, even with no other improvements in efficiency.
The first thing to understand is that the large majority of medical expenditures are not on things that happen to the young - where "young" means roughly "under 50". In fact, 75% of U.S. medical expenditures are on chronic diseases, which primarily affect old people - like my mother, who at 77 is on 7 different kinds of medication for things like diabetes, high blood pressure, and the side effects of others of the medications. And she's relatively healthy for her age! So if you're 30 or 40 years old, the vast majority of health care expenditures are on things that you've likely never even had to think about.
The second thing to understand is that the medical establishment has little knowledge of what causes these diseases. Rather, they treat symptoms. As one doctor puts it:
What would be the typical treatment of cardiovascular disease? First they check the cholesterol. High cholesterol over 200, they put you on cholesterol lowering drugs and what does it do? It shuts off your CoQ10. What does CoQ10 do? It is involved in the energy production and protection of little energy furnaces in every cell, so energy production goes way down.... One of the best treatments for a weak heart is CoQ10 for congestive heart failure. But medicine has no trouble shutting CoQ10 production off so that they can treat a number (cholesterol figure)....
As Gary Taubes puts it, doctors want to do something for their patients now, not do years of research to figure out what the right thing to do is. Regarding metabolic syndrome, which accounts for most of that chronic disease expenditure, he points out:
The urge to simplify a complex scientific situation so that physicians can apply it and their patients and the public embrace it has taken precedence over the scientific obligation of presenting the evidence with relentless honesty. The result is an enormous enterprise dedicated in theory to determining the relationship between diet, obesity, and disease, while dedicated in practice to convincing everyone involved, and the lay public, most of all, that the answers are already known and always have been - an enterprise, in other words, that purports to be a science and yet functions like a religion.
So basically most of that 75% of U.S. medical expenditure on chronic diseases doesn't work and likely makes things worse. If we could just resist the impulse to "do something, anything" when a problem came up, we could save that money with no impact on actual quality of health care.
And yes, we could also make things better while still saving the money. But that's a subject for another post. This one is just on saving the money.
 See http://genomebiology.com/2006/7/2/104 figure 1.
 See my previous post at http://psychohist.livejournal.com/40106.html for more information on Gary Taubes.
 Gary Taubes, Good Calories, Bad Calories, pp 451-452.
I find it hard to swallow that metabolic syndrome accounts for two-thirds of chronic disease expenditure. Your reference  shows spending for diabetes, congestive heart failure, respiratory disease, hypertension, and cancer. Even if you claim that metabolic syndrome is responsible for essentially all of the first four (which seems unlikely), spending on those according to the chart adds up to $100-$200 billion of the $1.2 trillion spent on "chronic disease." Where's the other trillion?
Further, even if you cut out the drugs used to treat "metabolic syndrome," you're not going to cut out all the spending on it. Your reference  is talking about how he treats various people with problems that he believes stem from insulin resistance, but he's still treating them and charging for office visits and lab tests to monitor their insulin and various vitamin levels. Their medical spending on these problems may have dropped, but it hasn't dropped to zero by any means.
I don't think wasted treatment is limited to metabolic syndrome; the Taubes quote is about metabolic syndrome but the indictment of the approach of the medical establishment applies more generally.
I do think that if 60% of the top five categories - diabetes, congestive heart failure, and hypertension are the ones I'd include - are attributable to metabolic syndrome, it's likely that somewhere around 60% of the categories not listed also are. Again, I do think other categories also include a lot of money spent on treatment that doesn't actually work.
In this post I'm only talking about saving money and not about improving health care. The saving money part does include dropping the costs of the present useless care to zero by dropping those treatments entirely. Finding out if there are other treatments that are not useless comes under "improving health care". I think that doctor's techniques are likely to be less useless than present treatments, but I don't think they're at all the most cost effective approach.
I still don't see how you can say, for example, that diabetes treatments "don't work." I might be willing to concede that hypertension drugs cause other problems in order to avoid a risk that it primarily statistical, but millions of diabetics are kept alive daily by appropriate and timely administration of insulin. Even if this is not the best possible treatment for them, it seems incredibly irresponsible and cruel to say that we should just save money by not treating diabetics at all, because we're only treating a symptom instead of a root cause. Do you really feel that your own mother should not be treated for diabetes, even if it hastens her death?
I still don't see how you can say, for example, that diabetes treatments "don't work."
Allow me explain further, then. Note that there are two kinds of diabetes, now called type 1 and type 2.
Type 1 diabetes is generally developed in childhood and results from the patient's inability to produce insulin. Insulin has been shown to prevent death from diabetes in controlled experiments, so insulin treatment does, in fact, work. Thus, insulin treatment for type 1 diabetes should not be discarded as one of the treatments that "don't work". Fortunately this would have little financial impact, as only 5-10% of diabetes cases in the U.S. are type 1, not to mention that insulin no longer has patent protection and is thus cheap. As far as I know, type 1 diabetes is not related to metabolic syndrome.
Type 2 diabetes is generally developed in adulthood and is due, not to lack of insulin, but to insulin "resistance": the muscle, fat, and liver tissues that normally absorb blood sugar in response to insulin no longer do so.
Now here we get into how medicine becomes religion. We actually do know what causes type 2 diabetes, and how to resolve it, or at least we did in the 1950s. The cause is excessive blood sugar, which comes primarily from dietary carbohydrate, with a small contribution from gluconeogenesis of protein. The solution is a low carbohydrate, high fat diet. Exercise also helps, as it expends blood sugar stored as glycogen in the muscles, but diet is the most effective treatment.
Unfortunately, such a diet runs heretically contrary to the AHA ex cathedra wisdom - unsupported by the evidence - that fat is evil. As a result, doctors end up recommending extremely low calorie diets to get at least some carbohydrate restriction with a politically correct diets. And since those low calorie diets are nearly impossible to adhere to, they end up prescribing a bunch of expensive medicine that might help reduce blood sugar, but may also hasten complete insulin dependence, along with other deleterious side effects.
So my answer in the case of my mother is this: I do not believe that dropping her diabetes medication would hasten her death. Indeed, if anything, I think the medication itself is what's hastening her death. When I saw her a couple months ago, she had weird bruises all over the the backs of her hands, because that's one of the side effects of one of the diabetes drugs if you accidentally get sun exposure. And she still has to restrict herself to 1000 calories per day, which is probably the only part of the regimen which is actually helping her.
I will admit to further empirical evidence that the medicine doesn't help, though. Last year, she was seeing a doctor who evidently understood what Rosedale understood, and had her on a 1000 calorie carbohydrate restricted diet - and no medication. She was more healthy, more active, and didn't have the weird bruises.
Unfortunately, that doctor isn't available any more - so now she sees a doctor who probably makes his Mercedes payments by making sure his patients have to come back to see him regularly, even if that's not the way he thinks of it.
Edited at 2009-10-22 02:18 am (UTC)
Well, of course there are actually more than two types of diabetes - I had the third most common, gestational diabetes, which generally acts like type 2. Insulin was effective for it, as insulin is effective for many type 2s. Insulin is a somewhat scary medication, as it is easier to experience dangerous blood-sugar lows on it than on some of the oral drugs, plus of course most people don't like giving themselves injections. But it does work. (There's some evidence that it can even send early Type 2 into remission.)
And even if your mother went off the drugs and just followed a low-carb diet, she would still require monitoring and testing to be sure that she was achieving adequate control with whatever regimen she was using. If she wasn't achieving good control, perhaps because her diet wasn't good enough or her disease was just too advanced, we'd still have to pay for her complications. Amputations aren't cheap. You just can't drop those costs to zero, even if you shift them around. I agree that more appropriate diet interventions could probably reduce them, but I'd be surprised if you could reduce them by more than half, and cutting half of the costs associated with 60% of chronic conditions (assuming arguendo that your numbers are right) isn't going to get you to a 50% reduction in medical spending.
I was going with the view that gestational diabetes is a subset of type 2 ("adult onset") diabetes, but I actually agree with you that it should be treated separately. One of the reasons for separate treatment is that intervention can be justified even if that intervention would not improve the health or lifespan of the mother, because the fetus may be at risk when the mother is not. The other is that gestational diabetes is not really a chronic disease - it normally goes away after childbirth. The overall health care expenses associated with gestational diabetes are negligible compared to normal type 2 diabetes, though.
It's true that diabetics would require some monitoring, such as self administered blood sugar tests, which my mother does a couple times daily. Those are very cheap, though. With respect to the costs savings that could be obtained, I'd bet on more than a 50% reduction even if we didn't take people off medication, just by replacing all diabetes medication with straight, cheap, unpatented nonanalog chemically human insulin costing maybe $30 a month rather than the order of magnitude higher costs of my mother's insulin medications. My mother could spend $40 a month on insulin and aspirin instead of the over $1000 a month she spends on all her medications now. Any savings from removing unnecessary drug treatment would be over and above the savings from the cheaper insulin and the simpler doctor care.
Basically, I think we could get much more than a 50% reduction in medical spending in the areas that like diabetes. I also think those areas include more than the 60% of areas that are covered by metabolic syndrom - for example see jaedian's comment and my response regarding cancer treatment. So I'll stick with my opinion that more than a 50% overall savings can be achieved, especially with respect to government expenditures, which are less efficient than private expenditures.
Incidentally, we did try to control my GDM with a low-carb diet before adding insulin. I could get good blood sugar control during the day, but my fasting sugar was always too high in the mornings. I had to add a long-acting insulin at bedtime to get that under control.
I am not clear that so many treatments don't work at all. The problem we have is that so many treatments work some of the time and we don't know how to distinguish who they will work on. Like cancer treatments - even if the success rate is low, there are some people treatments do help and we don't know how to tell who. I would like to see more gene testing for this, so we can give treatments to the people it will do good. We are moving in this direction.
And it is hard to NOT do anything, either as a doctor or as a patient. Maybe better education on that front would be useful? I don't think it is doctors being evil and trying to turn a profit. I think doctors were trained to heal/help people and it is so frustrating to say that you can't do anything for someone, so you try treatments that aren't likely to work just to be doing something. (doing nothing is a hard skill)
We also have a lot of issues performing medical tests. It takes many studies to really see if something works. And it is really hard to tease out the variables the way we do the studies. In addition I think the outcomes we measure are not always the right ones. Sometimes that is for ease in designing the trials. Instead of measuring if something reduces the mortality rate, it is easier to see if it shrinks the tumor, or improves 5 year survival rates. Neither of these is as good. I was looking at some research on prostate cancer that suggested the 5 year survival rate was bogus. It makes it look like we are doing so much better in the US because our 5 year survival rate is better than Britain. But that is because we diagnose (and often treat) earlier than in Britain. But the data suggestion people are overall dying at the same ages whether they get early treatment or not. In this case the "5 year" survival rate is an artifact produced by early diagnosis (the 5 year clock starts ticking earlier in the US)
I do think some of the science has merit. (not sure if it all does, but I did read the 2nd article which is interesting) The problem is, it is very hard to get people to change their diets. (we have made it even worse with all the lame info out there, like the politically motivated food pyramid) So I am not sure what you do then. Certainly it would be awesome if we can design medications that go after the underlying insulin resistance. (and I think we are with reservatol and others like it) But diet changes has proven to be a hard medicine to administer.
Very interesting discussion though. I do need to read Taubes book. (on my list) I do think the whole low-fat diet fad really messed up a lot of people, and caused a lot of weight gain. (since it was low-fat, high sugar, carb heavy)
It's true that some cancer treatments do seem to work on some people, and I wouldn't claim it's possible to drop them all without affecting health. However, that doesn't mean they all work, especially when you factor in the health costs that many of them have. In addition, I think it's questionable as to whether we should be doing treatments based on what just seems to work; it would be better to do some trials to identify what actually does work.
I agree the doctors are not consciously being evil; it's more a case of the road to hell being paved with good intentions. I also agree that it's psychologically difficult not to do anything. I do think that the psychological part of the problem could largely be dealt with by good science: that is, well designed experimental trials, as you suggest, rather than poorly designed trials or nonexperimental observational studies. If we had solid science providing a better idea of what did and did not work, I think it would be easier to stick to the "does work" category.
Though it goes beyond the topic a bit, I think a properly designed health insurance system has a part to play here. Right now part of the problem is that the only actors who have an incentive to do experimental trials are the drug companies, and they have incentives to set up the trials in such a way that they can only be used to prove that drugs work, not to prove that other methods might work. What we really need is a health insurance system tied to individuals rather than corporations, which would give the insurance companies an incentive to look at all options and pick the best ones for their patients. Of course it would have to be competitive so patients could abandon insurance companies that simply chose to minimize treatment irrespective of effectiveness.
Diet really gets into the area of improvements, and not just cost reduction. For now, I would just point out that the low fat diet "fad" is not really a fad - it's people paying attention to that bogus food pyramid the government promulgated.
Yeah, there is definitely a lot of treating symptoms and using new drugs before well-designed studies have shown that they are actually useful. I've seen a number of cases where drugs that have been properly controlling symptoms (such as atrial fibrillation) turn out not to actually improve mortality or other health aspects once the more detailed studies are done. If I remember correctly Obama was pushing for several billion dollars for such studies and the Republicans (probably pushed by the drugmakers) were opposing it.
While I do think better science would help, I think the government is exactly the wrong entity to do it. Don't forget that the government is responsible for the food pyramid that gave us the obesity and metabolic syndrome epidemic.
In fact, I don't think any actor with power over the necessary amounts of money has an interest in good science in this area. As noted in my response to jaedian, health insurance companies' interests could be aligned with the patients, but that would require a change that's quite unlikely to happen.
Well, at least we agree that better science would help. You just don't like the government doing anything :-)
My experience from doing both government-funded research (in grad school) and commercial work is that the government-funded research is actually one of the better ways to get good scientific results. Corporations are very good at marketing their own products, but you can't realistically ask a company why its product might be bad and expect to get a reasonable answer.
I don't know much about the origin of the food pyramid, but I'm guessing that the main problem was lack of good science.
I like the government to stick to things it's good at. Science is not one of those things.
I do agree that corporations don't do science that doesn't have a chance at promoting their products. This is the problem with the drug companies: they are required to do experimental trials, which do prove cause and effect, but the effect may not be something that actually contributes to health, and of course they'll only test drugs as causes, and will not test competing nondrug solutions. This is why health insurance companies' interests would have to be aligned with the insured, rather than with government or corporations, for them to do good science.
The food pyramid is actually a good example of how government makes science bad. It was based on the McGovern committee hearings, which in turn was based on months of interviews of the top nutrition scientists. Those scientists had done lots of government funded science, and lots more government funded science has been done since. However, as you guessed, it was not good science. The key issue here is farm states have disproportionate power in the Senate - McGovern was from a farm state - so the government has an interest in "science" that will promote farm products like grains. It works the same way drug company research does: it promotes the interests of the researcher, not the subjects of the research.
I think you are intermingling government policy decisions based on science with the actual research itself. The government policy decisions are clearly very influenced by different interest groups. It sounds like the McGovern committee hearings were a good example of this.
On the other hand, actual research funded by the government seems quite free of government influence. I'm probably biased because my parents both worked for NIH for many years, but I certainly never saw any government influence in my work or my parents work and never heard of any stories of attempted government meddling from the people I worked with.
But, I'm pretty sure I won't convince you that the government should be doing this research since you are claiming from the outset that the government is not good at science. So, alternatively, do you have a suggestion for a better organization to do the research? As you've pointed out the drug companies are extremely biased and you've volunteered that the insurance industry probably won't be changed to the point where it can do the research legitimately. So is there some other organization that can embark on decade-long medical trials with tens of thousands of subjects, not really caring which way the results turn out. That sounds like the government to me, but I wouldn't be opposed to another group doing the work. I love the work that Consumer Reports does, though they are way too small to take on something like this.
Well, I reread your comment and it really doesn't talk about the science in any detail. Nutrition science is probably one of the hardest areas in which to perform properly controlled studies, but it sounds like you are claiming that in general the government is bad at science.
I have to admit that seems like a very strong claim. So none of the aerospace, or astrophysics, or biomedical research that the government has done is any good? Many times the research is done by the same individuals who work for industry, just depending on who they get grant money from. Certainly many of them are world-class researchers.
If you have a chance to explain in more detail that would be appreciated. Alternatively, examples of other groups that do the research better might provide some insight.
What treptoplax said below. Both corporations and governments can do good experiments. Neither corporations nor governments will do experiments that don't have at least a chance of benefiting them.
Edited at 2009-10-23 05:05 am (UTC)
While I do think better science would help, I think the government is exactly the wrong entity to do it.
I am actually interested in what organizations should be doing the scientific research if not the government, though maybe that could be a separate post if this one is running on too long ...
No such organization presently exists. That's why it hasn't been done yet.
The smallest change that could result in such an organization would be tying competitive health insurance to the individual - both in terms of portability and in terms of payment - in which case the interests of the health insurer would be sufficiently aligned with the individual that the needed research would start to be done. This would require federal tax and regulatory changes removing the incentives for and perhaps discouraging employer based health insurance. Limitation of malpractice awards to actual damages or some small multiple of economic damages would also remove a distortion and would result in more effective care. There might need to be some regulatory changes to discourage cherry picking of patients from other insurers, but this would happen naturally under the current state level regulatory framework.
Unfortunately, both parties have ruled out a change from employer paid to individually paid health insurance, and the administration has ruled out a change to malpractice law. That means we'll just have to operate as best we can under the present system.
There are things that individuals can do. Just as you can check Consumer Reports and I can check J. D. Power and Car & Driver to screen which cars are best for us, the internet has made a wealth of health related information available to the individual. It just requires us to screen it carefully, and pick and choose our "test drives" of diet or medication. It does mean that clear cut conclusions will remain difficult to find, and health care costs will largely continue to act as a subsidy from those who take care of their health to those who don't, but I don't see any realistic way around that.
My understanding is that malpractice awards are limited to actual damages - it's just that actual damages can be quite high. Or do you mean that you think we should exclude "pain and suffering" as damage?
Personally, I'd like to see a workman's-comp-style malpractice system, since the current one systematically undercompensates those with modest damages, and randomly overcompensates a tiny fraction of those with high damages, with very high transaction costs. I would expect total malpractice payments to go up, not down, with such a system. Nevertheless, it might remove incentives for certain kinds of defensive medicine.
Perhaps I should have said "tangible damages" or just left it at a small multiple of economic damages.
Workman's comp covers expected accidents, while malpractice requires negligence, I believe. I'd prefer at least the option of lowering my rate through the risk sharing of having only negligence covered; in fact, I think some degree of risk sharing may be important to ensure that the patient cooperates with treatment.
That isn't to say some of the worthwhile aspects of workman's comp could be useful, though workman's comp also has its abuses.
Economically, if the patient is fully compensated for the pain - in the sense of, "here's how much you'd have to pay me to freely accept this amount of pain" - that doesn't count as a net cost, and thus doesn't count as risk sharing.
In that case, I suspect virtually no one is ever fully compensated for pain, because I have never heard of a medical malpractice plaintiff who would not prefer not to have been injured in the first place. Even the legendary McDonald's plaintiff would prefer not to have had her genitals melted and to skip the millions.
No such organization presently exists. That's why it hasn't been done yet.
The federal government does already do quite a lot of research in this area, though I realize you'd probably prefer that they didn't. For example one large study that provided evidence against a popular but not well tested treatment was the Women's Health Initiative study on hormone replacement therapy. It did a double-blind controlled study on over 100,000 women and determined that some of the popular hormone replacement therapies increased the risk of stroke, heart attack, etc. There are still benefits to the hormone replacement therapy and the risks depend on the hormones used, but at least now women have good data on what the benefits and risks are.
I'm all for a realignment of the insurance industry incentives so that they are more likely to do this kind of research, but in the meantime I'm glad that the government is doing the work.
Also the NIH hormone replacement study is not an example of good science in and of itself. It focused on the downside, just as the original studies focused on the upside. That's just an example of where multiple pieces of bad science can result in a situation where people who do their own research have a chance of making their own decisions, as I suggested.
Interesting. I strikes me the study is less useful if it's not applicable to the primary use of the intervention. Indeed, given the study ended after 5 years, if it's not applicable to 3-4 year use, it seems rather questionably illuminating at all.
Also, I just noticed that one of the conclusions was "All-cause mortality was not affected during the trial." There seem to be slightly more negative outcomes than positive outcomes, but the results are mixed, and the statistical significance is borderline when you consider the probability of chance statistical significance given the number of variables studied.
Well, it looks like we are mostly in agreement. The WHI research provided new scientific data that people can use to make decisions about their health care. The study was government-funded and wouldn't have been done without government funding. I'd certainly endorse changing the insurance incentives so that companies are more likely to do this kind of research. In the meantime I'm glad that the government is working on this type of research.
I just realized from the other comments that I answered a different question than you asked here. You're asking who should do the kinds of research that are presently done. I'm answering the question of who could do the kinds of research that aren't presently done and can't be done under the present system. Those are two completely different questions; no wonder we ended up talking past each other for so long.
No, I think there was plenty of good science.
s. Corporations are very good at marketing their own products, but you can't realistically ask a company why its product might be bad and expect to get a reasonable answer.
Yes, exactly. Drug companies will produce 'scientific research' that shows their drugs work (more if it's true, fortunately). Governments will produce 'scientific research' that gets campaign contributions.
I think it's terribly naive to think "Hey, we can bypass all these utopia-complete political-societal-engineering problems with a simple appeal to INFALLIBLE SCIENCE!" You try to scientize the politics, but you'll end up politicizing the science.
"Hey, we can bypass all these utopia-complete political-societal-engineering problems with a simple appeal to INFALLIBLE SCIENCE!" You try to scientize the politics, but you'll end up politicizing the science.
Uh, I didn't mean to glorify science in that manner. But there are cases where scientific research can provide new and useful information. I think everyone is in agreement on that. Warren's original post was saying that and I certainly agree. The remaining 90% of the conversation was all about whether the government or private companies should do the research.
Take for example the issue of Celebrex versus the old school NSAID's like aspirin. I'm not advocating that the government be in charge of developing new drugs like Celebrex. That seems like something that corporations can do much better. But when it comes to evaluating the effectiveness of new drugs I think the government is less biased than, in this case, Pfizer. Sure, the government won't be completely free of bias, but it's got to be better than letting the company that developed the drug evaluate it. That's all I'm saying. We got sidetracked into a discussion of whether the government was "bad" at research or not.
We have the NHTSA to test crash safety and I'm happy about that. It's not like I'm naïve about government pressures, but asking GM to tell us how safe its cars are has got to be worse.
I would note that the FDA regulatory framework helps keep drug trials honest, and that industry shoulders the actual cost of doing the crash tests; arguably both are examples of government regulated industry science, which while imperfect is probably better than giving either the government or industry complete control.
I would also note that just as the drug testing system leaves no one willing to do definitive studies that could show that, say, aspirin is as effective as expensive statin medications in preventing heart attacks and heart attack mortality, the crash test framework accurately accounts only for accident survival, and not accident likelihood.