Comments on: Meds To Cut

By: Overcoming Bias : Against This Med Reform

Overcoming Bias : Against This Med Reform — Tue, 08 Sep 2009 03:04:13 +0000

[...] exclude treatments that are “clinically inappropriate,” but since 46% of treatments are of “unknown effectiveness”, clearly they can’t let plans exclude such treatments, or [...]

By: sheila

sheila — Tue, 11 Aug 2009 03:20:09 +0000

oh i totally agree with jessie riedel, she sounds like a real smart girl!

By: Glen

Glen — Sun, 02 Aug 2009 09:21:48 +0000

I meant to say “makes me somewhat skeptical of the graph as a whole.”

By: Glen

Glen — Sun, 02 Aug 2009 09:20:57 +0000

The “trade off between benefits and harms” category confuses me, and it makes you somewhat skeptical of the graph as a whole, since I would have thought the treatments were categorized by net benefits. Don’t all medical treatments involve both costs and benefits? Maybe they mean that it’s a particularly close trade-off. Or maybe they mean that the trade-off depends on the individual case — although that, too, is surely true of all treatments.

By: Douglas Knight

Douglas Knight — Sun, 02 Aug 2009 01:44:06 +0000

That’s all correct, but the slice that this study labels as ineffective is only 1/5 of the assessed slice. (plus some of the “likely to be effective”) RH is right to say:

Since randomized experiments and cross regional regressions usually find zero correlation between health and medical spending, we should presume that treatments of unknown effectiveness are on average ineffective.

but he was careful not to admit that we should presume this about the 54% studied, as well. (ie, I claim the choice of which 54% to study has only a small bias towards effective treatments. That’s implied by RH’s comment: the alternative is to conclude that unstudied medicine is the harmful part). The main problem is that individual studies are massively biased to showing effectiveness, compared to aggregate studies. It’s not just that the aggregate studies show medicine doesn’t work; they also show that the individual studies are wrong. (One explanation that doesn’t involve corruption is that doctors who do studies are more competent than typical doctors. Also, patients in studies get more attention, raising compliance.)

One effect of focusing on cost-effectiveness is to raise the bar to fight this bias. And it’s not as bad as most instances of fighting bias with bias, since cost-effectiveness is the right thing to do, if it can be done correctly. But there’s a big cost of getting people to accept it at all.

By: Hal Finney

Hal Finney — Sun, 02 Aug 2009 00:19:27 +0000

There are two different issues here which often get mixed up. One is whether a treatment is effective. The other is whether a treatment is cost effective.

It’s one thing to say that we won’t pay for treatments that are not effective, or perhaps equivalently, treatments that do more (medical) harm than good. This seems reasonable. We are aiming to help people, not harm them, and treatments that do harm, or do nothing helpful, shouldn’t be paid for.

But most discussion skips right past this and jumps to the issue of cost effectiveness. It is far more problematic to say we won’t pay for treatments that are not cost effective. This requires weighing the dollar value to the patient of a beneficial treatment. In effect we have to put a price tag on the patient’s life and health.

While there may be good arguments in favor of this, it is important to see the enormous gap between this issue vs the problem of ineffective treatments. If merely eliminating treatments that are not beneficial can already reap great savings, as sounds likely, then we have much less urgency to wade into the murky philosophical waters of weighing human life and health in monetary terms.

By: "Effectiveness" - Maggie's Farm

"Effectiveness" - Maggie's Farm — Sat, 01 Aug 2009 17:11:20 +0000

[...] at Overcoming Bias posts the graph below from the British Medical Journal. It says they looked at 2500 treatments, so [...]

By: Robert Koslover

Robert Koslover — Sat, 01 Aug 2009 16:11:49 +0000

Mario, Just to clarify, I am referring to this issue under the assumption that you are talking about this being a government-imposed rule or law, without input from the affected businesses. Should an insurance company decide to pursue this on its own, then I expect it would: (1) do so for reasons of making money, not simply appealing to political constituencies, (2) do so with full knowledge of the market and realistic expectations for what to expect, and (3) do so with anticipation of the possible frauds and with planned mechanisms to manage them. In that case, however, your idea of "get insurance companies to offer customers..." means that you are simply offering a suggestion to those insurance companies, not a Government-mandated rule. Is that what you meant, Mario? If so, then I have no argument with you. And if so, why not send your suggestion to several health insurance companies and see if they agree? After all, if it really is a good idea, and if they can see how it would help them save money (which is what their business is all about) then they will surely embrace your wisdom (no kidding, and no sarcasm) and will be grateful to you for your brilliant (again, no kidding, and no sarcasm) suggestion.

By: Robert Koslover

Robert Koslover — Sat, 01 Aug 2009 15:58:37 +0000

Mario, with all due respect, I think you are missing the point. The Government's involvement, with its lesser competence in managing money when compared to the private sector, is a key factor in encouraging and enabling fraudulent activities. I gave you a specific example of a major and highly-comparable Government medical program being defrauded. So, given that, you defended adding new Government medical programs by asserting that hey, if fraud is already there, then so what if it continues? Really? The solution is to expand already failing systems? I must confess that I don't find that very persuasive.

By: andrew c

andrew c — Sat, 01 Aug 2009 10:44:52 +0000

2500 treatments studied doesn’t mean they are all applied with the same frequency. Show me the same data with the size of the pie slices proportional to the number of times the treatment is applied.