June 02, 2003

MEDICAL MALPRACTICE....Atrios links today to a summary of Time magazine's cover story this week about the medical malpractice crisis. Both the summary and the Time story itself claim that malpractice premiums actually rose faster in states with damage caps than in states without them.

This is true, but it turns out it's only part of the story, which gets even more surprising when you see the whole picture. The report they cite is here, and it actually says two things:

  • States with caps did have lower growth in malpractice payouts: a growth rate of 37% between 1991 and 2002 compared to 71% for states without caps.

  • However, exactly the opposite was true for malpractice premiums. States with caps saw premiums grow 48% between 1991 and 2002, while states without caps saw premiums grow only 35%. So in states with caps, payouts went up more slowly but premiums went up more quickly.

UPDATE: OK, hold on a second. I just deleted the rest of the original post.

On reading the report more carefully, I now note that all these figures are for median payouts, which wouldn't be affected at all by a few huge awards. I'm not sure what's going on, but this is so obviously the wrong statistic to use in this case that there must be some kind of axe to grind here. I'm now very skeptical that this report is, as it claims, "not driven by a political ideology or industry-driven self-interest."

What's more, although Appendix 2 does indeed show that premiums have gone up only 35% in states without caps, there are several states with enormous increases and several with tiny increases, and it's hard to see any correlation at all between median payout increases and median premium increases. This might be due to the use of medians, it might genuinely be due to different regulatory regimes, or it might be something else. Who knows?

Until further notice, I no longer believe anything in this report and I don't think Time should have cited it. I'll have more later if I can find any additional explanations.

UPDATE 2: By the way, this is why I hate this whole subject so much: it's practically impossible to get reliable statistics. These guys had access to the raw data and all they had to do was calculate average payout per doctor on a state-by-state basis and compare it to premiums, which would have been even easier than calculating meaningless medians. But they didn't, and no one else does either. Arggh.

Posted by KEVIN DRUM at June 2, 2003 10:13 PM | TrackBack


Only in these Alice in Wonderland times we live in could the right wing claim that the problem with medical malpractice lies in... the people who defend its victims. The problem is that doctors have to pay too much when they fuck up. The problem is not spreading carelessness, mistakes, and poor treatment in our health care system, but an outbreak of venality that causes otherwise unharmed people to sue, sue, sue!

`Now, I give you fair warning,' shouted the Queen, stamping on the ground as she spoke; `either you or your head must be off, and that in about half no time! Take your choice!'

Posted by: Realish at June 2, 2003 10:21 PM

What kevin! You want people to use statistics honestly!?! :)

Posted by: Sebastian Holsclaw at June 2, 2003 11:02 PM

Actually, any statistician will tell you that for many purposes a median is a better measure of central tendency than an average. It gives a much better feel for what is typical (in this case, a typical payout) precisely because it is not as easily thrown about by the occasional atypically high (or low) figure. This is especially true when, as in this case, the distribution is very skewed. The average is then a poor guide to what a 'normal' payout is.

OTOH, if your focus is not on what a typical payout is but rather on how high premiums should be, the average is a more useful figure (tho of course you'll also want to know the incidence - ie how many payouts are made).

So its horses for courses - but no way should you just dismiss the median as a 'meaningless' statistic.

Posted by: derrida derider at June 3, 2003 04:00 AM

But dd, the point of this exercise is to see if a program, caps, to curtail the outliers is working or not.

Posted by: theCoach at June 3, 2003 05:55 AM

I'm usually upset when I see stats that fail to use both the average and the median; both tell you different things, and if you can calculate one, you can generally calculate both. A serious news outlet (especially in print) should always give you both.

Posted by: Crank at June 3, 2003 06:29 AM

Kevin wrote: "However, exactly the opposite was true for malpractice premiums. States with caps saw payouts grow 48% between 1991 and 2002, while states without caps saw payouts grow only 35%. So in states with caps, payouts went up more slowly but premiums went up more quickly."

Since this is dealing with premiums, shouldn't "payouts" be changed to "premiums", except in the last sentence? Or am I reading this wrong?

Posted by: Goof Beyou at June 3, 2003 07:01 AM

If I may hijack the comments board, I would like to hear people's opinions about the root issues behind malpractice suits in general.

It seems to me that malpractice suits serve three different purposes: (1) to compensate victims and their families for their losses (future medical costs, plus loss of income due to disability, etc.), (2) to encourage future medical care professionals to do their jobs better, and (3) to satisfy the need of the victims and their families for justice.

It's not clear to me that malpractice suits are the best way to accomplish these goals. Goals (2) and (3) are related to punishing the people who did the bad thing. But it seems to me that for an insurance company to pay for the damages, rather than the responsible party, lessens the punitive effect of the suit. As for goal (1), I don't see why a malpractice suit should be necessary. Why is it necessary to establish fault before a victim can be compensated for his losses?

I'd like to know to what extent punitive malpractice awards actually improve medical care. I guess it would be hard to study, since the effect is pretty indirect. The complaint by doctors is that the threat of malpractice encourages defensive medicine, where the aim is more to protect the doctor than it is to protect the patient. (These aren't necessarily the same thing.)

Posted by: Daryl McCullough at June 3, 2003 08:26 AM

They use medians for both premiums and payouts. For premiums this probably doesn't make too much difference (although who knows for sure?), but for payouts the whole point of the exercise is to find out if there are a small number of huge outlier awards. This gets hidden by the use of medians.

Medians aren't *generally* useless, they're just useless for this particular exercise. What's more, it's so obvious that they're useless, that I suspect some kind of axe to grind.

I wrote an email to the company that produced the report, but I haven't gotten a reply yet.

Goof: yes, you're right. I've corrected the text.

Posted by: Kevin Drum at June 3, 2003 08:36 AM

I was going to write something snarky about wanting the median tax cut but the average malpractice payout, but then I realized Kevin's argument makes sense.

The engineer in me would love to see an actual probability distribution for tax-related issues, with income being the independent variable. But then I realized there really isn't an independent variable for the distribution of malpractice; per state, it's just how much per capita awards there have been, and the per capita premium. Well, maybe in some states there's a huge range of premiums, I doubt it.

Anyway, I agree. Let's get the raw data.

Posted by: Stoffel at June 3, 2003 09:13 AM

Hmmm that's an interesting case; one should use the mean precisely for the reasons one WOULDN'T use the mean if this were an ordinary analysis. I'll have to keep it in mind.

Incidentally, crank's right - both mean and median should be reported when there is any substantial difference between the two.

I'm probably being naive, but is it possible that it's just "recipe" style statistical analysis (as opposed to an analysis informed by common sense - something that is surprisingly uncommon) that is the culprit here, rather than ideological axe-grinding? Meaning: is the analyst thinking "skewed distribution ---> better use the median," or is (s)he rather thinking (less black-box, but more cynically) "the median tells the story I want to tell"?

Either way it's depressing.

On the other hand: just how skewed is the distribution of payouts? It seems to me that there is something of a drumbeat out there implying that payouts are uniformly too large. Perhaps the caps are explicitly meant to curtail the magnitude of outliers, but their overall frequency is important too - if they really out "outliers" in the classic sense, and the great bulk of the money in malpractice payouts actually comes from the more common and reasonable settlements, then the median may be defensible.

We need to see the data or at least get a look at the means.

Posted by: ryan at June 3, 2003 09:32 AM

"Only in these Alice in Wonderland times we live in could the right wing claim that the problem with medical malpractice lies in... the people who defend its victims. The problem is that doctors have to pay too much when they fuck up."

You see no connection between the fact that doctors get to pay too much (The very problem those evil right-wing caps are designed to curtail btw.) and the system that determines the payouts?

And the people who defend the wronged have never been the problem - it's rather the ones who fight for those who merely claim to have been wronged that are problematic.


Posted by: GulGnu at June 3, 2003 10:24 AM

I think the real problem is that we don't have a good handle on what an appropriate level of payout would like.

There is also the issue of what ought to qualify as malpractice. Medicine is an inherently risky art, and over-defining the standard of care can stifle medical innovation. My life was quite probably saved by a vascular surgeon trying a new technique. Would he have risked doing so under a very rigid 'standard of care' system.

The problem with the term 'malpractice' is that it encompasses unwarranted risks to the patient, and some cases where the risk levels are uncertain. The current malpractice system is such that doctors have to spend a huge percentage of their time justifying their decisions instead of making them. I'm not convinced that the proper balance has been found between protecting patients from truly bad doctors and allowing for the riskyness of a risky medical world.

Posted by: Sebastian Holsclaw at June 3, 2003 10:29 AM

Actually, malpractice premium rates have surprisingly little to do with payouts. Insurance companies make most of their money on the float - the interest they collect on the money between the time it is paid in and the time they have to pay it out. Hence, when interest rates go down, premiums go up, and vice versa. This always happens. Politicians would much rather not confuse us with this fact, though.

Posted by: Emma Anne at June 3, 2003 11:01 AM

For those with an interest in immunizing themselves against statistical flummery, the old classic "How To Lie With Statistics" is very useful and approachable.

Somewhat dated in the examples (1st published just after WWII, I think) but that does not invalidate the teaching points, and to me gives it a kind of funky charm...

Posted by: Parker at June 3, 2003 11:27 AM

Emma - It's true that insurance companies make most of their money on the float. It's also true that payouts that are 20% larger than expected would more than eliminate any profit from the float. The base premium (from the entire risk pool) should about equal the expected liabilities. If the base premium is much larger than the expected liabilities, another company will bid the away the business. If the base premium is much smaller than the expected liabilities, the company will go bankrupt.

The problem? Actuaries are finding it much more difficult to predict the actual liabilities associated with a given risk pool. It's not just the size of the awards that is the problem; it's also the variability of the size of the awards. Suppose one case settles at $50,000 and another, nearly identical case, ends with an award of $1 million. If you are the actuary, do you treat the $50,000 settlement as if a similar case in the future will settle for that amount or do you ignore the $1 million as an aberration. Remember, if you set the premium too high, your company will go out of business. If you set it too low, it will go bankrupt. (The answer, of course, is that the attorney representing the client who settled for $50,000 has learned from the outcome of the other case. In the future, $50,000 settlements will be hard to come by.)

Posted by: David Walser at June 3, 2003 11:31 AM
Actually, malpractice premium rates have surprisingly little to do with payouts. Insurance companies make most of their money on the float - the interest they collect on the money between the time it is paid in and the time they have to pay it out. Hence, when interest rates go down, premiums go up, and vice versa. This always happens. Politicians would much rather not confuse us with this fact, though.
This is true, however the fact that the increase is disproportionately large in some states versus others indicates that there may be some other variable other than the prevailing interest rates at work. The premium increase from interest rates should be constant across the board. Posted by: Paladin at June 3, 2003 11:37 AM

Well, I make these decisions about whether or not to enter a market to write med mal. No cap, you don't get my capacity. Put a cap in place, you get my capacity. Period. All else is basically irrelevant.

So what premiums are being tracked, admitted and non-admitted? Non-admitted is tough to track. What happens is the use of admitted, trackable paper goes up after a cap, and the amount of E&S, non-admitted paper goes down.

and please show me how I can make a decent float in todays market, umm? and we do not ignore the $1,000,000 cases, not in med mal.

I love it when people start talking about this, for the level of knowledge (lack thereof) gets revealed! His document does not address E&S, surplus, etc..

signed med mal u/w

Posted by: TK at June 3, 2003 12:06 PM

Palidin: "This is true, however the fact that the increase is disproportionately large in some states versus others indicates that there may be some other variable other than the prevailing interest rates at work. The premium increase from interest rates should be constant across the board."

Agreed, but since the people complaining about malpractice premiums neglect to tell us that most of the increases are a result of dropping interest rates, I am uninclined to believe their claims about what does cause the increases.

They want the change, so they they have to make their case. Well, actually, they dont, but they should have to. . .

Posted by: Emma Anne at June 3, 2003 01:23 PM

Disclaimers: (1) I am a practicing physician and (2) I have no idea what E&S is or are.

After reading the executive summary of the report, I thought their conclusion made a lot of sense: "There are other, far more important factors driving the rise in med mal premiums".

Caps appeal to most physicians because they resent having to be defensive, and often have undergone some relative financial hardship (please note the relative) because of insurance. More importantly, they are getting less and less reward -- both emotional and financial -- from actual one-on-one doctor-patient medicine. I think caps are a "feel good" issue -- like going to war, it's something concrete one can do to address a big source of frustration and concern. Also like going to war, it doesn't really work.

Caps won't help much to restore patients' access to needed care, which has widely been blamed on the malpractice "crisis." They may help some, but that's just a bandage on a wound. What we need is an organ transplant.

Daryl McCullough's post above details three benefits of malpractice, and I agree with him that it's only effective in his item 1, compensating malpractice victims for suffering and, more importantly, future medical care. That's important, but I think it's awfully Republican (read that as inhumane, in case you need translation) of our society to rely on an adversarial system to assure that.

Those who suffer bad medical outcomes should be compensated, but their compensation should be decided rationally, and without anyone having an interest in denying them that compensation. It should be no-fault. In fact, the best way to assure that these unfortunates get adequate care is to assure that we all do -- a single payer system.

Of course, I've skipped a couple steps here, but this rant has already gone on long enough.

Posted by: JPG at June 3, 2003 02:02 PM

hey guys, no increases across the board, we slice and dice states and territories into all sorts of pieces. PA is like 5-6, even though you have to file rates statewide. So rural places don't get as much as urban, in the same state.

And your blaming all this on interest rates in only a part, a small part. the main is just STUPID underwriting on behalf of all these people like Swiss Re, St. Paul,etc.. reporting bad results.

E&S: Excess & Surplus, not regulated in form or rate by the state, and mostly not tracked.

Posted by: TK at June 3, 2003 02:21 PM

The payout data mentioned in the Weiss Report is available on NPDB's website for free, so anyone can do their own analysis if they like.

Go to and download it. If that link does not work www.npdb.com and click on "public data files" in the left column.

In looking at the NPDB data...claim severity has gone up as the # of claims above $1MM has risen from about 200 to 240 such cases throughout most of the 1990's to around 400 in 2002.

Also not all caps are equal, some states have caps on both economic and non-economic damages. Some states have exceptions to the non-economic damage caps.

Note the NPDB is for healthcare providers, mostly physicians, no good data exists on what hospitals payout and they are the deepest pocket in med mal so they are the ones that take the really big hits, not the physicians.

Posted by: ma at June 3, 2003 03:18 PM

One issue that I haven't seen discussed: the fact that according to the NPDB and the Dept. of HHS the majority of malpractice payouts in this country (54%) result from the acts of 5% of the physicians! One doc in PA had 24 claims against him between 1993 and 2001 with payouts of over $8 million. This same doc was never disciplined by the PA med association. What's up with that?

It is my understanding that the amount of premiums are not experience rated based on a doc's claims history, only by speciality within the profession. If, with a clean driving record over the last five years, my auto insurance premium was the same as the guy who's had five accidents that were his fault over the last five years, I think I'd be pretty mad at my insurer. So why isn't the med profession focusing some wrath at their insurers along with plaintiffs and their lawyers?

If the medical profession did a better job of policing its own ranks, we would see less medical malpractice and lower premiums. That, combined with reforms to the insurance industry to deal with the insurance business cycle, will do more to deal with this situation than caps ever will.

Posted by: Brian King at June 3, 2003 04:33 PM

I think your indignation at the use of medians is misplaced. Far from weakening the conclusion, it strengthens the conclusion. To reiterate the findings.
i) states without caps saw their payouts rise more than states with (If there is any effect due to very large payout outliers in the states without caps...this observation still stands)
ii) states with caps saw premiums rise faster than states without. (unless there are large outlier premiums...in the states without caps...the median is not going to be that different from the mean...I would argue for consistency's sake that you use the same measure)

Suppose the author(s) had used the mean, you could have said that the payout rise in the states without caps is affected by outliers and therefore we can't be sure if there is indeed a secular increase in payouts in those states. The use of medians is a good counter to such arguments.

Posted by: Sam Jackson at June 4, 2003 02:39 AM

umm there is a "non-standard" physician market for those 5% who cause all the claims. My fave was a bad doc at a large hosp. I kept seeing his name next to large suits. All six figures.

But the biggest claim was when the hosp tried to fire him, he filed a "denial of privileges claim", that almost cost more than his dropped sponges and wrong cuts.

So can we talk about the lawyers now?

and those stats are meaningless. Go look at the schedule P for an insurance carrier.

Posted by: TK at June 4, 2003 08:27 AM

RE: 5% of the doctors result in 54% of the payouts. I think there is more going on here than the assumption which you guys seem to be working under: the idea that this 5% represents the bad apple doctors. I don't know where we could get statistics on these issues, but there are other factors that could contribute.

1) These doctors could be in high-risk, experimental, or highly emotional specialties. This is probably the biggest non-negligence factor. Your average dermatologist isn't going to get sued much because his work isn't experimental and when failures occur they aren't spectacular. An interventional radiologist suffers much more risk because his specialty is newer, and the failures are spectacular.

2) Some portion of this 5% may live near a particularly agressive med-mal attorney.

3) Re: payouts. I'm not sure whether this includes payment in settlement and judgments only, or if it includes the cost of defending cases. Obstetricians get sued regularly, and their defense costs are very high even if they are never found liable for malpractice. These defense costs increase the premiums.

I don't know where you could get reliable statistics on any of these factors, but they should be considered in the discussion.

Posted by: Sebastian Holsclaw at June 4, 2003 09:26 AM

As a physician who has been sued (and "won" , as if anyone actually wins a lawsuit) , I can speak with experience on this malpractice "crisis". The real problem as I see it is that the system is stacked for the trial lawyers to make as much money as they can and not for any sense of justice or fairness for the patients or doctors.The trial lawyers get "experts" to testify against you, who in most cases are well-paid professional hacks. The trial lawyers also concentrate on the outcome that occurred and not how it happened. The actual facts of a case don't seem to matter as long as the lawyer can confuse and convince a jury that you are incompetent and the plaintiff "deserves" compensation. Patients who are truly injured don't get their just compensation since 30-50% of the award goes toward the lawyer and expenses. It's a terrible system for everyone except the lawyers, who make money on both sides.

Posted by: Tom from Memphis, TN at June 4, 2003 07:30 PM

Blaming the stock market for increases in Pennsylvania's skyrocketing premiums does not stand up to scrutiny.

There is a paper available entitled "Did Investement Affect Medical Liablity Premiums?" that answers the following questions:

1)How did the drop in the equity markets affect insurance company performance? (No)

2)Are companies raising premiums because they lost money on Enron or WorldCom? (No)

3)Did medical malpractice companies have an unusually large amount of equities in their portfolio? (No)

4)Given their level of equity exposure, did they invest prudently in the market or did they gamble by investing in technology or telecom stocks? (Yes)

5) Is 10% the correct amount for medical malpractice insurers to invest in equities? (Yes)

6) If the degree of equity exposure was not unusual, was the investing? (No)

A quote from the paper reads "As medical malpractice companies did not have an unusual amount invested in equities and since they invested these monies in a reasonable market-like fashion, we conclude that the decline in equity valuations is not the cause of rising medical malpractice premiums"

So what is causing premium increase? Another quote "One sees that the losses per doctor have grown at a much higher rate than either medical inflation or premiums per doctor. In order for losses in 2001 to have equaled the build up created by inflation in medical care during the period 1975–2001, companies would have to reduce the amount of paid losses by approximately 60%. Therefore, losses, not inflation, are the problem."


Posted by: Beth at June 5, 2003 01:14 PM

Brian King ... you have many errors in your posting and your comments are incorrect for several reasons. First, the Pa. Medical (Association) Society is not the disciplinary board in Pennsylvania. You've accused the wrong organization, and unfortunately you didn't check the State Board of Medicine's WWW site that lists disciplinary actions. Second, you made the same mistake that Public Citizen did by not understanding the reporting structure from Pennsylvania to the NPDB. Pennsylvania has two levels of liability insurance (primary and MCARE). One payout likely generates two reports to the NPDB. Also, claims history does play a role in the amount of premiums. Check out Act 13 of 2002 for MCARE and talk to insurers for primary.

Posted by: CAM at June 26, 2003 07:16 AM

The Weiss Report (cited above) has been soundly criticized for faulty statistical methods (including use of un-weighted medians).

A more accurate study on this issue was recently released by the acturarial firm Milliman, USA. Milliman used the average cost per doctor (NOT the average or median cost per claim, which is clearly only half the story) to show that doctors have below average medical malpractice losses in states where non-economic damages are capped.


Posted by: AG at July 31, 2003 11:53 AM
Post a comment

Remember personal info?

Powered by
Movable Type 2.63