American Enterprise Institute

April 5, 2006

[Edited transcript from audio tapes]

8:45 a.m.	Registration
9:00	Presenters:	David Dranove, Kellogg School of Management, Northwestern University
		Jonathan Klick, AEI and Florida State University College of Law
9:40	Discussants:	Randy Bovbjerg, Urban Institute
		Ted Frank, AEI
	Moderator:	John E. Calfee, AEI
11:00	Adjournment

Proceedings:

John Calfee:  Well, good morning and welcome to the American Enterprise Institute.  I’m Jack Calfee, I’m an economist here.  I’ll be moderating the session.  And let me begin with a few announcements, the most relevant of which is that one of our speakers, Jonathan Klick, is not here this morning due to some sort of a glitch last night in catching an airline flight out of Florida, and I’m not sure exactly what happened, but we know for sure that Jonathan is not here.  Ted Frank will more or less pinch-hit for him.  We do have Jonathan’s PowerPoint, which helps quite a bit and, of course, Ted has also read his paper. 

Ted Frank:  And you are still paying for his travel, right?

John Calfee:  We will see.  And a few other things beforehand, first of all I wanted to thank my colleague, Bob Helms, who did most of the work of actually organizing this session but bears no responsibility for airline malfunctions as it were.  Two other upcoming events:  On April 14th, in the morning, we are going to be having a small conference featuring Paul Offit of the Children’s Hospital of Philadelphia.  Paul Offit is the author of a book called The Cutter Incident, which is about the early days of the polio vaccine, which is the mid-1950s and some very important litigation that ensued, which in turn had a powerful influence in the liability system generally and especially the role of the liability system in vaccine development and usage. 

And so Paul will present the results, basically [will] summarize his book in that event, and a couple of others of us will be commenting on that.  And then on the afternoon of April 24th, my colleague Ted Frank, who is sitting next to me, is organizing - I would not quite call it impromptu, but it is a session that we will be looking at and thinking about and talking about discussing and possibly debating a medical malpractice legislation, which is expected to be debated in the Senate shortly after April 24th, specifically the first week of May.  And so Ted is organizing that, and that will be in the afternoon of April 24th and that should be of interest, at least to quite a few of the people who are here today. 

And so we move on to today’s session, which is on medical malpractice, specifically on the impact of liability and changes in the liability system on access to care.  We have two papers, one author, to talk about this.  And we are going to start off with David Dranove, who teaches in the Kellogg School at Northwestern University.  Just to give you an idea of how this will flow, after David talks, Ted Frank will give his interpretation of Jonathan Klick and then Randy Bovbjerg - I better get that pronunciation correct.

Randy Bovbjerg:  Bovbjerg like iceberg.

John Calfee:  Bovbjerg, Randy Bovbjerg of the Urban Institute here in Washington will present his take, his critique of both papers, and then Ted Frank will come back and presumably critique his own earlier presentation and also talk about Dranove’s paper, and then we will have a discussion amongst the authors and discussants and then we will have questions and answers from the audience.  So David, I think we are ready for you to start.

David Dranove:  Great, thanks, Jack.  A pleasure to be back.  I think this is my fifth time, having an opportunity to present here over the years, and I would like to thank Bob Helms who has been kind enough to invite me back each time.  We are going to be talking today about a second paper in a research agenda that is still ongoing.  The first paper was co-authored by Anne Gron, and we published that in the journal Health Affairs about a year ago and it presented some preliminary, just kind of raw data about patient access and incidence of high-risk procedures in Florida through 2003. 

The paper I will be presenting today, which is also co-authored with Andrew Sfekas, a new colleague of mine at Northwestern, extends our work out to 2004, does some additional comparisons and does some more rigorous statistical analysis, including hypothesis testing.  There we go. 

High malpractice premiums have obviously been a concern of policy analysts for a long time.  There has been a lot of work on defensive medicine and whether that is driving up the cost of care, and there is also considerable concern about whether exit by physicians might harm access to care.  There has certainly been a lot of television and newspaper reports, a lot of anecdotal reports about physician exit.  There has not been a lot of systematic study on the issue, however, and this paper contributes to the research on access. 

The prior research such as it is, I think, begins with a GAO report, which investigated anecdotal reports of physician exit through 2002, and they were able to find few confirmed exits that were linked directly to malpractice concerns, and they even question whether the rate of exit by physicians from the market had increased over the pre-2000 rate, 2000 really being about the dividing point when the current malpractice “crisis” begins. 

There were some concerns about the study, however.  They only examined a limited number of anecdotes.  It is not a systematic study by any stretch of the imagination.  They only look at physician exit from the markets.  They do not consider another possible outcome of the crisis, which is, physicians simply reducing the number of high-risk procedures they perform.  That could have very serious implications for access as well. 

And as I stated, their study ends in 2002, and if you think about physicians deciding to relocate to another state in response of the malpractice crisis, a couple of years’ worth of premium increases might not be enough to drive that kind of major decision.  So it would be very important to look at longer-term responses.  Systematic research, however, has been somewhat lacking in this area.  Research by Baicker and Chandra and Berchiaris have found modest relationships between malpractice premiums and physician location decisions, and the Klick paper that we are going to hear discussed today, I think, represents an important addition to this literature, and I will let you hear from not Klick himself but the rest of the panel in discussing the paper.  Okay. 

So this paper that I’m going to present provides a systematic study of Florida, which is by all accounts a malpractice crisis state, and I’ll present more information about that in just a moment.  I’m going to focus on neurosurgery and obstetrics, two specialties that have been especially hard-hit by the crisis.  For access, I’m going to measure travel times for high-risk procedures within these specialties – craniotomies and high-risk deliveries.  I’m also going to assess physician activity levels, including but not limited to exit.  So I’m also going to look at downsizing of physician practices. 

I will engage in several statistical comparisons.  I’ll be doing a pre-post comparison, what was the trend like in the malpractice years 2000 and 2004 versus the previous time period.  I’ll compare trends for high-risk procedures with trends for low-risk procedures, and finally I’ll compare Florida with California, which by all accounts, is a non-crisis state.  And lastly, I’ll present the results of a couple of case studies of markets that experienced physician exit.  These case studies are representative of a series of case studies that we performed. 

Let me summarize the main results, and then I’ll go into more detail into each of the topics that I just mentioned.  We find little evidence of increased travel times.  What we do find is for craniotomies and the travel times for craniotomies seem to be going up for urban patients.  And what you might expect the crisis to have bite for rural patients, the results there are mixed. 

We do not find any evidence of increased travel times for patients having high-risk deliveries.  We do find fewer providers performing high-risk procedures.  The decline is especially concentrated amongst low-volume providers, although we find broader evidence amongst the neurosurgeons that is happening a little bit more across the board.  And lastly, the case studies are unable to identify very many markets that were devastated by exit, and in fact in the few markets that were able to hone in on, there are a variety of responses in those markets suggesting that in most of these markets, the patients were able to cope just fine. 

So why should we study access?  Well, physician responses to the malpractice crisis, such as exit and downsizing, could affect incidence and access.  If there is a substitute lower-risk intervention, even watchful waiting, then the incidence of a high-risk procedure could fall, and so you might want to look at incidence effects and we did so in the Health Affairs and found no incidence effects whatsoever.  And the reality is for craniotomies and especially for deliveries there really is not a very good alternative.  Watchful waiting does not work very well for deliveries.  If a patient’s preferred provider does leave the market or downsize however, then that doctor’s patient is going to have to seek an alternative provider for care, and that is really the access problem. 

Now, the access problem can manifest itself in many ways.  You do not get to see your preferred provider is the big picture, how you would measure that is kind of an interesting problem and we are going to use a standard measure, which is travel times, which is kind of nice.  It certainly captures what a lot of people have been saying in the anecdotes about rural patients now having to travel out of their own zip code or their own county in order to find care, but certainly there is much more to it than that. 

We all have idiosyncratic preferences, loyalties to specific physicians, and if we cannot access those physicians even if we could find another local provider who can pick up the slack, that does not necessarily mean we are just as well off as we used to be.  We are not going to discuss the latter at all.  We do not have the tools right now to measure that, so we are just going to look at travel times. 

Okay.  Why are we studying Florida?  Florida is a good study state for several reasons.  According to the American Medical Association, it is one of currently 20 crisis states.  There is a variety of criteria for identifying crisis states, but perhaps the most significant and the most obvious one is that the increases in malpractice insurance premiums in Florida since 2000 have been – I think it is now 60 percent over four years, from 2000 to 2004 – well exceeding the median for the United States.  California, in contrast, the premiums have gone up, something like 10 or 15 percent in the same four-year time period.  They have a number of caps on malpractice litigation and you will hear about that in the Klick paper. 

Another important reason to study Florida, it is not the only crisis state, but Florida has unique data, which lends itself to the study.  It not only offers patient-level hospital data, which several other states offer, it also includes the license number of the operating physicians who are able to track physician activity levels over time and find out whether physicians are performing procedures or even leaving the market altogether.  Lastly, Florida is amongst the most up-to-date of all the states.  We have all the data through 2004 and within a few months, all of 2005 will be available, and this project can be updated at that time. 

So how are we going to identify high-risk procedures?  That turns out to be a little bit of a tricky question, so we talked to a lot of doctors to try to nail this down.  First thing we did was focus on neurosurgery and obstetrics.  Those are the two specialties, along with emergency medicine, that have received the most attention in the media. 

Within neurosurgery, we define the high-risk procedures to be craniotomies.  Anytime they are opening your skull, it kind of suggests that it is a high-risk procedure.  These do have the highest mortality risk and the highest expense amongst the neurosurgical procedures.  For obstetrics, you would like to identify high-risk patients prior to the delivery.  You do not want the situation where the doctor discovers at the delivery that the patient is high-risk, because there is not much the doctor can do about it at that point.  They are not going to say, “Well, I’m not going to take the baby out at this point in time because I think you are too high-risk.”  So what we try to do is find some secondary diagnoses such as whether or not the mother had diabetes or whether it was multiple-gestation, things that are known to the physician early on in the pregnancy and the physician can decide whether or not they want to take on the patient. 

A problem we encountered, however, was that the coding of these diagnoses in the medical records that were then given to the state of Florida and given to us was inconsistent [with] a great increase in the number of secondary diagnoses being coded over time and this could create some problems for our empirical research.  So instead we were forced to use the diagnosis-related groups corresponding to deliveries with complications.  A drawback of this is, if some of these complications arise during delivery, so we are seeing some physician activity that would have taken place for physicians who are trying to cut back on these types of patients but did not see the complications were there until it was too late to decline the care. 

Okay.  Well, let me move now into the methods.  You understand why we are doing Florida versus California, why we are doing neurosurgery and obstetrics, and how we chose the high-risk procedures that we did.  It is not perfect but it is the best we can do with our data. 

Let’s talk about the research methods now for the regression analysis for travel times.  The unit of observation is the patient and we have the travel time from the patient’s zip code, centroid, to the hospital location and we used Map Quest travel time calculator – it is not great, but it is probably as good as any of the others that are available.  We estimated what are known as “difference in difference” regressions.  This is just fancy public finance speak for we are going to compare the trend for the experimental group against the trend for a control group and see if the experimental trend differs from the control trend.  Then if it does, then we have got an effect. 

What are the comparisons that we are going to look at?  We are going to look at the trend in travel times for high-risk deliveries, versus the trend in travel times for low-risk deliveries.  We are going to look at the trend in travel times in 2000 and 2004, which is the crisis period to the trend in ’96 to 2000, which is the pre-crisis period.  We are going to look at the trend in Florida versus the trend in California, a crisis state versus a non-crisis state. 

We are going to get very consistent results in all of these examinations, so I’m not going to present all of them here.  We are going to use regression analysis to control for other factors that might affect travel time, such as age, race, and insurance coverage that might be changing over time or cross markets.  So these will be controlled for when we do the trend analysis. 

I do not know how well we could see this here.  I’m going to try to focus your attention with a laser pointer.  These are the results or some of the results for the neurosurgery travel times.  What you see in, for example, this number here, 33.46 is the travel time for the average patient with the average characteristics in the sample who receives a craniotomy in 1996 in Florida and it is 33.46 minutes.  This is the rural patients and their average travel time is 55.11 [minutes].  This includes rural and urban.  You could see that by 2000, the same craniotomy patients have experienced a travel time increased to 37 minutes and it jumps up to 42 minutes.  Here are the travel times for the low-risk neurosurgery procedures and here it kicks up from 23 to 24.5 to 25.

John Calfee:  David, can I interrupt, just to clarify an inquiry.  The travel times, you are talking about travel times in 1996, but they were calculated from Map Quest in 2005.

David Dranove:  That is correct, an obvious limitation.  That is right.  However, if road conditions have changed, for example, and people are moving further away from the central city, the control would be the travel times for the other neurosurgery group. 

What we have here is what we call the difference in difference estimator.  This one in particular is taking the increase for the craniotomy patients between 2000 and 2004 - that is about five minutes - and subtracting the increase for the low-risk patients in the same time period, which is about four-tenths of a minute, hence we get 4.6 minutes as the difference in difference estimate and it is statistically significant at less than 0.001, so the travel time increase in 2000 to 2004 for the craniotomy patients is much bigger than the travel time increase for the other neurosurgery patients.  That suggests that there is something going on that is associated with access. 

Another way of doing it is to compare the trend in 2000 and 2004 versus the trend in the previous period when there was no crisis, and now the difference in difference estimates are smaller.  It is only 1.6 minutes, but again it is statistically significant. 

There is something going on here.  People are traveling further and if we do the same comparison against California, we get, again, about a four-minute bigger increase in Florida than we do in California.  This is our strongest evidence of access effects and let me try to put this in perspective.  A typical craniotomy patient is in the hospital for about 15 days.  If we allow for visits by family members everyday, roundtrip visits, so we take the craniotomy patient, one roundtrip is now about eight minutes longer, about 15 visits by family members – one a day, we were thinking – so we are adding now about another two hours’ worth of travel.  We could even double that, allow other family members to come visit and make it four hours’ worth of additional travel. 

If you put the going rate on travel times onto these numbers and the going value of an hour of time is somewhere in the neighborhood of about $15 an hour and obviously varies by who you are and what your income is and so forth.  You could see that the amount of dollars at stake is a very, very small amount, in the order of about $100 is the dollar equivalent of the travel time increase, and these are procedures with costs running to the tens of thousands of dollars.  So while we are documenting an effect, effects that you might find from other problems, such as defensive medicine, ordering one additional test, that cost will dwarf what we are finding in the access effect here. 

Not only that, if you focus on the rural travel time, you get inconsistent results.  It is certainly the case of the increase for craniotomies in the crisis period is bigger than the increase for other neurosurgeries, but is no different than the pre-crisis trend and it really was not any different from what was going on in California.  So we would expect to see the biggest effect, we actually did not see anything consistent. 

We move along and show you very quickly for high-risk deliveries, the magnitude of the increase in travel time between 2000 and 2004 is trivial, it is less than one minute and it is not any different from the trend for low-risk deliveries, or from the pre-crisis trend, or from the increase that we observed in California, and there is also nothing going on in the rural markets.  So for high-risk deliveries, there does not seem to be any increase in travel times whatsoever. 

Okay.  Catch our breaths and move along to the next part here.  Very quickly now, I’m going to talk about the physician activity levels.  We are going to look at a couple of things.  We are going to look at the distribution of physicians who are performing virtually no high-risk procedures all the way to lots of high-risk procedures, and these are the volumes we use for these definitions.  One to three procedures is very low volume; four to 10 is low volume, 11 to 24, medium.  I think actually 25 to 51 is high and we may have very high in the way the next table is set up, which is on average of at least one a week. 

And then we defined “exit” to be not necessarily a complete exit, but substantial withdraw and mind you, a lot of these exits could simply be retirement.  We have no way of knowing why they are exiting, and so low-volume exit as you go from one to three in, say, year 2002, and two years later you are doing none.  For a low-volume exit, you are doing four to 10 and two years later, you are doing none or one.  For the higher volumes, we call exit at least a 50 percent reduction in the number of high-risk procedures you are performing.  We have tried other definitions.  We get the same general results. 

Obviously, these are kind of arbitrary, but a lot of arbitrary alternatives that give you the same results.  The first thing is here, as the tables are showing you, how the physicians break out.  And one of the things that is startling here is, even for craniotomies, about half of all the physicians who do craniotomies in a given year only do about one or two of them.  That is a rather scary proposition. These could very simply be emergency procedures and the neurosurgeon is just not around to do them.  We could see that for high-risk deliveries, about 20 percent of the physicians who do high-risk deliveries are doing only one or two of them.  Of course, the vast majority of procedures are concentrated in the high-volume and very high-volume providers. 

Of note in this table there seems to be a slight change in the make up here for craniotomies, the percentage of the doctors who do only one or two in a year seems to have fallen just a little bit, but those who are doing three to ten a year seems to have increased and there is no other substantial change, so it is kind of wash with craniotomies, not much happening there.  For high-risk deliveries, however, there is a statistically significant reduction in the percentage of physicians who are only doing one or two high-risk deliveries a year. 

So there seems to be a concentration increase instead amongst higher-volume providers.  The same type result is going to show up when I look at the actual exit rates, and here we see that in any given year that providers have only done one or without high-risk procedures in that year, say, high-risk craniotomies, 80 percent of those providers will do none two years later.  That is what that column means. 

So between 2000 and 2002, 80 percent of the docs who did one or two procedures in 2000 did none in 2002 and then you see that between 2002 and 2004, the exit rate has gone up for the very low volume providers.  Eighty-five percent are the ones who did one or two have exited the market.  That might be encouraging but what is disturbing here, this is the other disturbing result for neurosurgery, is that those who are doing between 25 and 50 procedures have exited the market, twelve percent of them have exited during that time period.  Only four percent have exited in the previous two-year window and only two percent exited during the previous two-year window, so there was a sharp acceleration in the exit rate amongst this group of physicians and it is statistically significant. 

For the high-risk deliveries, however, the picture is much more consistent as the low-to-medium volume providers who are exiting the market in bigger numbers, and you could see the trends here are all going up the percentage who decide to drop out the market is much higher in 2002 to 2004 then it was in previous years and again, statistically significant.  I will summarize all these results in just a moment. 

[I] just want to very briefly talk about a couple of the case studies that we looked at.  We looked only at rural markets, figuring that if you are in an urban market, your provider leaves the market, you will be able to find a local alternative.  We looked for rural markets that had at least five high-risk patients in 2002, after which exiters accounted for a large percentage of the procedures done in 2002, at least 50 percent.  So these are markets that had a fairly large number of patients and there was at least one doctor in that market who was exits the market who has taken care of most of them. 

We are only able to identify six such markets for craniotomies, and in all of those markets there are exactly five patients who had received surgery in the year before the exit.  We are able to identify 14 such zip codes for high-risk deliveries.  They had 119 patients, so the total number of patients who are in markets that you might think were substantially affected by exit is just 149 for the entire state.  It is a very tiny percentage. 

Let us walk you through a couple of the case studies.  In one zip code, there are 18 high-risk deliveries performed in 2002 by 11 different doctors.  There was one doctor who did six of those procedures and that doctor exited the market.  All of those patients traveled less than 40 minutes to get their care, two years later, there were 29 high-risk deliveries, so this is a very busy zip code.  There was one new doctor who had not previously been in the market who now performs seven procedures, so he is a new doctor who entered the market to pick up the slack.  But we noted in this market, they are five of the 29 patients who are now traveling considerably further than anybody had traveled before, so this was a market in which the exit appears to have had some kind of negative impact. 

More typical was this kind market, [a] zip code that had 12 high-risk deliveries in 2002, on exiting physician that performed six of them and all of the patients traveled at least 85 minutes to get their care.  In 2004, there were fourteen high-risk deliveries, and there was a new physician who had entered the market who is now performing six of those.  And the local hospital had opened a new obstetrics unit that was performing quite a few of those high-risk deliveries, reducing the travel times. 

In each of the case studies that we did, we found a different pattern.  There was no consistent conclusion to draw [about] what happens to a rural market when the doctor leaves town.  Our conclusions, we find mixed evidence that the crisis has had harmful effects through 2004.  We find higher travel times for urban craniotomy patients, although the dollar implications of those higher travel times are somewhat small.  Otherwise, there was, I should not say really apparent decrease from trend for rural craniotomy patients.  It really depends on which comparison you are looking at.  No change in travel times or high-risk deliveries.  Reductions across the board in activity levels of low-volume providers, and I should have added her, a somewhat disturbing reduction for some of the high-volume craniotomy providers.  Exit by neurosurgeons in high-risk and obstetricians doing high-risk deliveries, especially again at the low-to-medium activity levels and a few case studies of rural markets affected by exit with no consistent harmful effects. 

So yes, we can document some harmful effects, but they are kind of just in a few places.  The overall results seem to point towards this being very consistent with what the GAO reported, which is, Shakespeare might have said, “Much ado about nothing.”  The end.

John Calfee:  Thank you, David.  We will go directly to Ted Frank, who would do his imitation of Jonathan Klick, who is at Florida State University as far as I know, as far as I know is at Florida State University at this very moment, and we have a new PowerPoint coming up.  And we are all set? 

Randy Bovbjerg:  How many people here are medical…    

Ted Frank:  Okay, so pinch-hitting for Jon Klick.  Pretend I’m taller and more hirsute and better looking… he’s got a fantastic beard.  Jon, with Tom Strathman, has done a study on Medical Malpractice Reform In Physicians In High-Risk Specialties. 

As you know, there is a controversy; the AMA saying there is a med mal crisis, some lawyers arguing that there is no evidence of this.  What do we know about the relationship?  Well, there have been studies but here have been problems with the studies.  We have limited data even when you do have single-state studies.  The limits on the data create some potentially misleading results.  Jon would also have some criticisms of the Baicker and Chandra study here, which would be really phenomenal criticisms, I’ll tell you.  But what one can do is indirectly test the doctor location.  Do tort reform effects have impact on where doctors decide to locate and there have been previous studies on this, looking at direct and indirect effects. 

I won’t go over that in detail.  Those are discussed in the paper itself, but there are also problems with these indirect studies, if you are taking a look at a state, and if you are looking at the number of doctors and you are comparing it to tort reforms, there are other effects going on also.  You have the political economy, you have health care effects, the population is growing, the population is getting older, so over the course between 1996 and 2004 in Florida, you have the population increase by over a million people, you are going to expect more doctors coming into the market even as you have these other effects from the medical malpractice crisis, to the extent that there is a crisis and it is hard to separate these effects out. 

Klick and Strathman did do an earlier study, they were unhappy with their ability to isolate these, but I think what they have done here is they have come up with a clever way to separate these effects out to the extent doctors are concerned about malpractice liability.  They are concerned because they are in high-risk areas, obstetrics, neurosurgery, emergency surgery, whereas other doctors had very little effect.  Medical malpractice has very little effect on them, perhaps pediatrics, dermatology, other minor medical areas.  You can compare these low-risk specialists to the high-risk specialists and the low-risk specialists will have all the same effects within the state that the high-risk specialists have theirs, facing the same economy, they are facing the same population growth. 

They are facing the same changes and aging population.  What changes between them is the malpractice environment that they operate in.  That is changing for the high-risk specialists, changing much, much less for the low-risk specialists.  So to the extent, the world is a better place for doctors or a worst place for doctors, it is the same of high-risk and low-risk, except with respect to the malpractice environment, to the extent caps make a difference, to the extent reforms make a difference, we should see different effects for the high-risk doctors than for the low-risk doctors, and that is what the study attempts to do. 

It performs the regressions on a group of high-risk doctors, compares it to the low-risk doctors to the extent that the low-risk doctors are indifferent about malpractice reform, to the extent because the doctor does not really face liability concerns, the doctor does not care about caps, does not care about joint and several liability - this strategy will work.  You will be able to isolate the effect on high-risk doctors.  However, if low-risk doctors do care about this, it will dampen the effect.  If for some reason low-risk doctors are more likely to practice in a state that has a terrible malpractice environment, then we might see the reverse problem: that the effects would be amplified within the study if the reaction is opposite. 

But assuming for the moment that doctors who face low risk of malpractice liability are indifferent about malpractice reform - we can do this model.  The dependent variable is the number of physicians in specialty I in state S during year T, per 100,000 state S residents and takes individual variables controlling, if there is a reform in place, binary variable 1, 0, or otherwise, with control for various dummies and specialty years and fixed effects. The high-risk specialties versus the low-risk specialties — these were taken from, it says in the paper where they were taken from, with the most claims awarded per doctor.  Neurosurgery, thoracic surgery, obstetrics, GP, emergency room – these are the doctors that have the most claims against them; whereas for whatever reason, oncologists, psychiatrists, doctors dealing with diabetes have fewer claims. 

The reforms tested various demands.  These are the ones that the AMA has identified as the reforms they most want to see – damage caps, collateral source reform, joint and several liability, caps on contingent fees, and so on.  These are what were modeled, and the data was taken from the American Tort Reform Association’s listing of when state legislatures have actually passed reforms. 

If you turn to table three in the paper, you do see positive statistically significant effects for several of these reforms, that it does have an effect on where high-risk doctors practice.  When you have non-economic damages caps, damage caps in general, collateral source reform, contingency fee caps, and it does have - if you pass all four of these reforms - it has an actual effect of 11 percent on the number of doctors who practice in the state, whether or not these reforms take effect.

John Calfee:  You are getting more doctors…

Ted Frank:  You get more doctors if these reforms happen.  Interestingly, there are reforms out there that have the opposite effect that have an effect that decreases the number of doctors that reduces access.  Joint and several liability was the one that, while the least significant of these, though it is significant at a two percent level, was the most problematic.  Klick theorizes that the problem there is that when you eliminate joint and several liability, plaintiffs are more likely to go after doctors’ assets individually rather than against deep-pocket hospitals, and that might have an effect on where the doctors choose to practice or might cause them to decide that they need additional coverage at greater expense. 

Other effects, a bit smaller, but they also had negative effects, periodic payments and victims’ fund.  Victims’ fund is obvious why that would happen.  Because when states establish no-fault victims’ fund, as a couple of states have for birth injuries, that is usually funded by a tax on doctors, so there is a direct income effect on doctors when those happen. 

Periodic payments - that is the reform, whereby courts are given authority to establish damages in form of annuities.  Where this comes into play most often is injury, birth injuries, because there is always a debate over the life span of the child, the plaintiff wanting to claim that this child is going to need an extensive expensive medical care for 80 years going into the future, the defense claiming, no, this child is going to be dead by the age of five and therefore, you need fewer damages.  By establishing periodic payments, the court moves the risk outside the province of the jury.  Why this has a negative effect, it is not clear, but it does within the study. 

And I have gone the wrong direction.  Here we go.  It does not matter.  Klick calls this the great unexplored question because what we are talking here is just the number of doctors, but just the number of doctors increase access to health care.  Klick and Strathman did find effects for infant mortality rates.  Rubin and Shepherd found effects for accidental death rates and certainly Kessler and McClellan study on defense of medicine shows a great expense is there, but what impressed me the most is another Kessler study, that if you adopt the findings of that study to the Klick and Stratmann study here, to get the same effects in doctor population, in number of doctors, as you could get by tort reform, you would have to raise doctors’ salaries by 37 percent.  That seems a remarkably high number, but that is the impact of tort reform, according to the study. 

There you see that on the first conclusion, medical malpractice reforms may affect access to care.  The suggestion is that the effect is likely to be concentrated geographically and by specialty, the high-risk specialties versus the low-risk specialties.  That might imply that a better reform would be to focus the reforms on the doctors most at risk for malpractice.  Of course, I’ll leave that side … well, Klick says, by targeting the reforms you can get the most bang for the buck.

John Calfee:  Thank you, Ted, a very good job, considering that an hour or two ago, he had no idea that he would be presenting this paper.  We are fortunate in having Randy Bovbjerg and with any luck, his single slide, is that right, Randy?  Randy is from the Urban Institute, and as in the case of David Dranove, this is far from his first appearance here at AEI.

Randy Bovbjerg:  I guess this will constrain me from waving my arms around very much, at least my right.  I will not be able to bash on the right, only to the left.  I know there are some doctors in the audience.  I know when you go to a medical convention, if you do not have slides, you might as well stand up without wearing pants.  It is not serious.  So I have a slide and the important things are how to pronounce my name, which we have already covered. 

But I’m at the Urban Institute, which we have already covered, but you need to notice that that is the dagger you see before you and what the dagger means is the standard disclaimer applies, which means the Urban Institute wants you to know that I’m not representing the Urban Institute, its trustees, its funders, any of my colleagues at the current time or the past time, or anyone who has ever gone into the door at 2100 M Street.  Then last, I think these are really good papers.  There is an awful lot of nonsense on both sides of this tort reform debate.  I give you that one citation.  I think there are handouts that you can get and you do not have to write down the URL, but Bob Berenson and I did this thing about myths on both sides that you might find interesting.  It is only eight pages or something like that, but for eight pages, you do get 100 references.  We limit ourselves to 100, and a couple of things forthcoming about different reforms that are not discussed here.  By my watch, I have minus nine minutes.  What do you want?  All right, I will start the timer.  It gives real-time feedback that actually does have a deterrent effect one may ask whether legal incentives, regulatory or liability do, but this one does, so I will have a peer help me remember when I get past 10 minutes, but it is counting up.  That is just easier for my brain to deal with. 

I like these papers a lot, so let me do two parts to this.  Usually I would do something simplistic, like good news and bad news, but I think here, it is sort of what I liked a lot and what I liked less.  I think these papers really are an improvement over a fairly simplistic bunch of stuff that has been out there in the past.  Things that I like a lot, number one, these are good issues.  This takes on the central doctor myth and that is that not only do doctors get hurt by lawyers, but patients get hurt by lawyers.  Who cares whether the doctors are hurt by lawyers, besides the people they give campaign contributions to and play golf with?  But patients, we do care about.  That is a much better issue.

Second, I think that the overall impact, what one takes away from a thing like this is sort of what is the big picture?  The big picture here is it is a pretty small picture.  Okay?  There are impacts of crises.  There are impacts on access, but they are not big, and that fits my preconceptions.  One jarring note was the thing about having to raise doctors’ salaries by 37 percent to get the same impact.  I think if we raise doctors’ salaries by 37 percent, we could reduce travel time by more than two minutes, so I’m very suspicious about the 37 percent number.  I do not know where it came from.  So this does match with common sense. 

Third, although I’m not an econometrician, I can tell you these are fancy analyses, and they are better.  Mainly, I think the thing is that the underlying model is better.  Instead of just looking one state compared to another at a single point in time, they are really comparing multiple states across multiple time periods and this business of looking at high risk versus low risk within the state and looking at travel times is really good because that captures some of the other things that are going on. 

Payment rate changes will be the same for the high and low-risk within the same state, one presumes, not necessarily for these high-risk procedures.  There might be something very specific to a high-risk procedure, especially in obstetrics, where Medicaid is a very dominant pair and if you get a big change in Medicaid payment, you have affected, what, 40 or 50 percent of deliveries and that is a big deal. 

There can also be changes like high-cost case management, actually quality-oriented things where you actually want people to travel.  You want them to find a good high-risk provider.  So it is more or less more complicated than this, but this analysis is more complicated and more sophisticated than its predecessors. 

I guess I want to say that the AEI deserves congratulation.  I think a lot of people in the world, if you ask them what AEI’s position was, they would say tort bad, reform good.  And here is a much more sophisticated analysis than that, saying tort, a little bit bad; reform a little bit good, and it is a mixed picture, which I think is really what is going in the world. 

How that will play in advocacy is what I will end with in seven minutes.  What I like less is the underlying model is still not super-sophisticated, compared with what I think really is going on in the world.  Considering, what it is about tort reform that might do good?  Well, the main thing it does and the main reason that doctors are marching is premiums.  Premiums are a cost.  This is an oil price shock and doctors, unlike the last couple big crises, are price takers.  They just cannot pass it through.  So a lot of this may be related to premiums. 

Premiums vary by state, except and I think it is 17 states where there are sub-state areas, and of course any group of physicians can self-insure or do next to self-insure and change their premiums in different ways.  That gets to be complex.  But tort reform also does something different, especially caps.  There is not just the premium effect in February whenever you write the check for the year.  You may have to do it twice a year because the insurance companies want to raise rates more frequently. 

There is also the cost of being a litigant.  There are psychic costs, and so on and I mentioned that the high-risk people are more affected than low-risk people really kind of picks up on that.  Being a litigant is crap on either side.  Being lawyer on either side is different.  Being a litigant is really pretty bad, and particularly in the case of caps, not for any of the others, collateral source offset applies to every case.  But caps directly apply only to the big cases, and they give a defendant, particularly the individual entrepreneur doctor defendant, something that you cannot get from premium support.  They had New Jersey, other places give doctors more money to cover premiums in lieu or raising the Medicaid rate. 

That is a purely fiscal effect, but what the cap does is it gets rid of somewhat irrational but very real fear of losing everything.  You cannot buy liability insurance coverage that will cover the big cases they read about all the time that are rare, but they hear about them all the time.  You just cannot do it.  London will not sell it to you; Zurich will not sell it to you.  Hartford will not sell it to you, particularly not in the aftermath of 9/11.  You cannot get such a big-ticket thing and if you could, it would be enormously expensive. 

So there are these two effects.  One is the premium one, which is kind of blunderbuss and one is this fear of the tort system and the cost of the tort system alone, which would apply to high and low-risk doctors both, and the premium effects also spill over because this is an imperfect business (setting premiums) and everybody bears a share.  The actuary is set.  What they need for all the specialties at once and then they share them out across specialties.  So there is a lot of spillover across specialty of premium effects.  I think that the time for care is a really clever measure.  It looks at the individual procedures and so on but it does not capture anything. 

One thing I said was for a certain high-risk, things like high-risk pregnancy, you want people to travel farther to a better place.  Okay, I grew up in Iowa, in Iowa City, where it was within 10 minutes of three hospitals, one of them was the largest, at that time, medical school in the country.  I spent summers at Okoboji.  Who knows where Okoboji is?  We are about to have all kinds of correspondents in Iowa, getting ready for the caucuses.  If you want to zap them on local color, ask him how it is playing in Okoboji and take a look at their face, okay - mainly rural resort area in the northwest part of the state.  It was 40 minutes to get to kind of a glorified nursing home hospital across the lake, so you had to go around the north end, but it saved my dad’s life for a low-risk gallbladder procedure.  And that is another thing. 

There is a lamppost effect here.  Everybody is captive of their data.  It is like the drunk looking for his keys, where does he look?  Where the light is, okay?  Where does an econometrician look?  Where the data are.  Now these data are better than previous data, but they still have the limitations of being their data.  I will tell you the number one thing the trial lawyers will say:  You measure the hospital procedures.  You do not get the guy who died in the ambulance on the way to the hospital.  So there are limitations like that in here. 

I think also that I’m very nervous about this high-risk versus low-risk, it kind of makes sense but I think it needs a lot more thinking. 

Let me give you another example from Iowa.  Family practitioners, it is said, are dropping out of delivering babies because it is not just specialty but it is procedures that determine your malpractice rate.  And if you do 20 deliveries a year, in Milford, Iowa, near Okoboji but on the south end, you will actually be lucky to do 10, I think.  You are going to pay the same rate as that guy in Iowa City who is doing lots.  And so family practitioners are dropping out of delivering babies.  That’s important. 

Access to primary care is incredibly important.  How do you learn that you are a high-risk pregnancy?  By going to a doctor who is willing to see you and determine that you needed to travel and not just show up at the hospital and have it determined there. 

So I’m a little nervous about this high-risk/low-risk in that way, and I think an awful lot has to do with the intra-state high-risk/low-risk of the probability of being sued.  Huge difference between south Florida and north Florida.  Enormous, okay?  And partly these analyses control for that and partly they do not.  I think that the model needs to be more complex and proved over time.  I would be very happy to see some labor economists for people who do work force analyses take this issue as one little piece where you put the premiums together with other things.  And if you think about it, what is it that over the long run, longer than 2000 or 2004, what is it that drives this stuff?  Well, it is choice of specialty in medical school.  Oh my goodness.  Choice of specialty in medical school, choice of where you first hang out your shingle, choice of what procedures you do once it is hung up, choice of whether you move, all kinds of things affect choices. 

When my wife and I, before we had kids in school and we were willing to consider moving, one of our screens was is there an NBA team in town, okay?  There are lots of things.  I saw an analysis published once of a work force that had average temperature, and it was statistically significant.  You could go play golf, okay?  Lots of things affect, I would be happier to see some of this analysis come out of that world rather than out of the malpractice world.

 Skipping right ahead, the biggest issue of all, now that I’m at minus 48 seconds, I think is it is not what is wrong with these papers but what would be better with the next round.  The lawyer’s myth is the really big one here.  The lawyer’s myth is that, of course, the legal system is a very powerful tool.  The powerful tool does not affect doctors and where they locate and stuff like that.  What it affects is how careful they are.  That is the thing that really needs analysis. 

Then let me end with what I think will happen in the political debate.  I’m a terrible politician but I like to observe.   What will the trial lawyers say about this?  And advocacy tends to radically oversimplify everything, whether it is courtroom advocacy, you have got to tell your story.  Or whether it is the legislative advocacy where to a certain extent, you can tell different stories and different cloakrooms but you still got to have a pretty simple story.  I think the trial lawyers will pound very hard; they are very good at pounding very hard; that not capturing the deaths in the ambulances is a fatal flaw.  That’s the big deal here. 

They will also say that thing that I said before, “Hey, this does not show that we are bad.  It shows that we are good.  We are making people travel farther and think more carefully.  We are driving out those people who only do two procedures a year.  That is great.”  So there you go.  You just helped the trial lawyers a lot. 

What will the doctors say?  “Oh, come on.  This is too short-term.  Four years?  Give me a break.  All those young guys are not going into obstetrics.  Old guys tend to leave obstetrics.  That is how it is.  You cannot stay up all night all the time and so forth.  This is terrible.  You are not truly capturing the effect of a bunch of people in obstetrics-gyn practice, where half of them go to GYN only so they do not have to pay any more premiums for those guys and the other guys do more volume but their premium remains the same because for the market for premiums is kind of mucky.”

And finally, I think that the doctors will say, “Listen, you are just not focusing on this one little effect for these high-risk specialties.  You are missing the big picture.  This really does affect every doctor.  It really does affect access to primary care; it really does affect trust and all those other things.”  So I think that is what will happen in the political process, but I think in the commentary process and the academic process.  These are very big improvements.

John Calfee:  Thank you, Randy.  We did not mind at all your taking a little bit of extra time.  Ted Klick was originally recruited as a commenter on these papers and he presented Jonathan Klick and Strathman’s paper, but I imagined he had something to say about David’s paper also. 

Ted Frank:  Thanks, Jack.  If I can get the… oh, I have it.  The disadvantage of going second is Randy stole all my really good points, but I’ll do what I can.  The question of access is of course important.  The idea of malpractice liability is that it is intended to improve patient safety.  Does it have incentive effects on doctors?  The lawyers say that it does, but to the extent that doctors respond to incentives, it works both ways.  They are not only deterred from bad practice, but they might be deterred from practice at all, and if at the margin the adverse effects of the malpractice status quo outweigh the beneficial effects, that is an argument for reform. 

Again, I think I could have just given the slides to Randy, I will not repeat it.  So, let’s go straight to the Dranove paper.  It is an attempt to measure all these things and my first concern, like Randy, with both of these studies, is what I liked and what I liked a little bit less.  My concern is that the study dampens effects to a certain extent.  And I see the nodding. 

There is probably some agreement there but I do not know if you have ever driven from Los Angeles to Las Vegas, but Barstow is where you stop for the bathroom on the way.  And if you go to the Barstow Community Hospital in the 92311 area code, if you are traveling within 92311 to Barstow Community Hospital, it is countered within the Dranove paper as a one-minute travel time, but that zip code’s over 50 miles long north to south, and it is California and there are much larger zip codes in California than 92311. 

In comparison, if you go to Cedars-Sinai where the celebrities go, you are on 90211, that is seven blocks from Olympic to Beverly Boulevard in that zip code, and that is going to be the same minute.  And it is probably more than a minute to travel those seven blocks, but it is still probably less time than the average time and so travel within 92311. 

Another concern I would have is that over time you would expect the trends to reduce travel time.  We heard about the hospital opening.  I imagine if you looked at that case study, what you would find is that you had a small town growing into a much larger town and suddenly was able to economically support hospitals, whereas before they had to travel a county over.  They are now able to travel to a hospital within that [county].  So one would expect in both California and Florida is reduced travel time and we are seeing the exact opposite.  We are seeing increased travel time so that the effects that we are seeing are more than just the four or five minutes, more than, say, the $30 at $15 an hour we would see, and I think Randy hit upon this, that travel time just does not affect travel time but it affects whether or not the patient dies in the ambulance. 

Another concern I had is that in terms of high risk versus low risk, the procedure studies are an imperfect proxy. Let’s start with the craniotomies.  One thing I learned from Randy’s previous presentation here at AEI is that visual aids are very important, so for craniotomies…

Randy Bovbjerg:  That is me on the right.

Ted Frank:  With craniotomies, there are two types of craniotomies:  There is the scheduled craniotomy.  There is the x-ray, and “Hey, you have a problem here.  We need to have a craniotomy.  Let’s schedule it.”  And you have the other kind of craniotomy, which is you are in a bad trauma, car accident, and the ambulance needs to take you somewhere, either by helicopter or by ambulance.  In the case of the scheduled craniotomy, if you have that rural hospital, where it goes from one full-time doctor to one part-time doctor or two doctors to one doctor, the doctor is not on-call a particular day.  You re-schedule the craniotomy from Monday to Thursday.  No difference in travel time, no real difference in convenience. 

Where the issue rises, and again Randy touched on this, is emergency surgery.  The sort staffing means somebody is not on duty that particular time when you ran into the truck.  The patient travels farther and that could have a significant impact on recovery, not just the $15 an hour but even worst, the patient could die and that artificially truncates the travel times because if somebody needs to get to the operating table within 30 minutes and travel time is 40 minutes, it is not going to show up in the study at all because as Randy noted, they are going to die in the ambulance. 

Obstetrics, it did not really play to me.  The study measured high-risk deliveries versus low-risk deliveries and perhaps I’m wrong here but it is not clear to me that an obstetrician will go to the insurance company and say, “Okay, can I get a lower insurance rate if I say I’m not going to do the high-risk deliveries anymore.”  I think what the real choice obstetricians face is delivery or no delivery.  And so I think if you were to study this, you would find, what Dranove did find, which is that there really was not a difference in the trend between high-risk deliveries and low-risk deliveries because the doctors for both are facing the same set of constraints. 

The GPs are deciding not to deliver.  They are not deciding, “I will do the low-risk deliveries and I will not do the high-risk deliveries.”  They are deciding, “I’m not going to deliver.”  This is especially problematic because to determine whether or not a delivery is high risk, a lot of the cases are co-dependent on [whether] there were complications at the time of delivery.  You cannot know that at the time you had agreed on the delivery months in advance.  Are the doctors who really choose the low-risk delivery practices and say, “I’m not going to do any DRG 372s” - that probably does not happen so the study is not measuring what is really out there. 

How can we fix that?  Well, I think you would need to study what Randy talked about, the GP problem.  I think Klick has done another paper where he studies infant mortality rates or infant health rates.  Randy has another suggestion that I did not quite hear but another place where there has been little evidence is vaginal birth after cesarean.  It is a concern for a lot of women.  There are a lot of doctors, a lot of hospitals out there that just say, “If you have had a cesarean, we are not going to let you do a vaginal birth.  We are going to give you another cesarean.”  The malpractice risks are too high.  There is going to be too much second-guessing if something goes wrong. 

And so the number of cesareans has increased in this country substantially.  I think the numbers are that it has tripled over the last 20, 30 years; whereas the number of birth defect injuries have remained level.  This is almost purely defensive medicine that we are seeing at a great personal cost to women and a great expense to the medical system and no marginal benefits to the health and that is something that can be studied, to sort of see whether or not there are effects.  Something that really struck me in the paper was a discussion of the obstetrics problem in determining high-risk behaviors. 

The original idea within the study, according to footnote seven, was they were going to look at the number of classified births that before the delivery were classified as high risk, and they found that that number had just dramatically increased over time.  And so because of that problem, they had decided to look at it afterwards, and I would like to know why that has dramatically increased over time.  Is that just simple upcoding or doctors trying to rip of their insurance companies or is there something going on there where this diagnosis creep is evidence of defensive medicine, that the health system is just as healthy as it was 10 years ago but now doctors feel the need to paper the record in such a way to protect themselves. 

And if so are they ordering the necessary tests or perhaps as Tom Baker might argue, are they are actually providing better care that they would not have without the malpractice liability?  I will say less about Klick because I have already sort of talk about Klick.  My main concern is the use of the data, which only looks at the legislative passage.  And the unfortunate political reality is that in many states, when the legislature passes reform, that is not a sign that reform is actually going to take place but a sign that reform is more likely to take place because now the battle is going to shift to the courts, and the courts are going to decide whether or not the reform actually comes into effect.  And most insurance companies wait until that battle is done, wait until the court rules whether the reform is going to take effect before they lower the rates or before they assume that they are going to be caps. They assume that there is going to be joint and several liability. 

Ohio passed reform in 2003, but the State Supreme Court has historically been not very friendly to the idea of reform and the insurers have a real concern there that the State Supreme Court is going to reverse it.  They have not lowered the insurance rates yet and because of that, you have that binary variable saying that there is reform in effect in Ohio but on the grounds in Ohio, they probably do not consider that to be a place where there has actually been reform.  The result here is probably to underestimate the impact of reform.  You are creating some noise in the variable because the binary variably says 1 when it probably should say 0.4. 

At the end of the day, you have these high effects and it may be multiplying errors when you take the Kessler study about income effects on doctor supply, and you try to translate it into the effects on doctor supply of tort reform and equate it to, so maybe 37 percent is too big of a number.  But on the other hand, we have a study here that has probably underestimated the impact of reform where that ends up, I’m not sure but it probably still remains to be a big number.  I’ll leave it at that.             

John E. Calfee:  Yes.  We are going to be moving the microphone back and forth here and then when we turn to the audience, which we will do in a few minutes, we will have a second microphone, which alleviates things a little bit.  David, I would be interested in hearing your responses in some of these things.  There are two questions I want to make sure that you address.  One of them is a diagnostic creed that Ted just mentioned.  The other is why is California not a crisis state of Florida?

David Dranove:  Let me answer those two questions.  Am I on?  Okay, great.  The diagnosis creed I do not have a good answer to, but I do suspect it is the hospitals covering their hinnies for any number of reasons.  It may be related to reimbursement changes where more documentations leading to better reimbursements and maybe legal, but we have not explored that. 

California has certainly not experienced anywhere near the rate of premium increases as other states.  They also have some of the purportedly toughest tort reform legislation, CAPS and premiums, and all the like, and certainly a lot of people points to those as the reason why it is considered to be a more friendly state in terms of the litigious environment.  I did hear a lot of terrific comments about research methods.  I think I could respond in an encouraging way to all but one, which I think has given me some pause at how would I address the last one of those. 

One of the comments had to do with the different sizes of zip codes and travel time and so forth.  Actually, I have done a lot of research in my life on geographic markets and travel, and so this is an area that is enough concern of mine.  One thing we did in all of our regressions was to include on zip code fixed effects, which were dummy variables for each zip code which, to some extent, is going to capture what you are talking about.  It would not capture changes over time within zip codes or travel patterns, but it will capture the mean travel differences and zip codes throughout the data.  The use of one minute per travel time was really not very important to our results. 

Both of the discussants raised concerns about comparing high risk versus low risk in our study and the high risk/low risk comparison, as you recall, is just one of several comparisons that we used, and we only did it for the travel time.  We also did pre-trend/post-trend and we also compared high risk in Florida, high risk in California, and the results were quite consistent across the different comparisons.  So that does suggest that there is something there that is not an artifact of some of the difficulties we did admittedly have in defining high risk. 

Also, well, we agreed that if you are a doctor you cannot decide only to do low-risk delivery.  You cannot inspect early in the game to discover if it is going to be a high-risk pregnancy and try to pass it on to somebody else.  Maybe indeed somebody further away and we, in fact, try to remain somewhat agnostic about increases in travel time. 

It is certainly a cost to the patient in the travel component.  But as we discussed, when we get into the fact that we see a lot of the low volume doctors pulling out of the market that might actually be a very good thing.  That might be exactly what the malpractice system is supposed to encourage to happen, assuming that low volume is somehow correlated with low quality.  And besides, we do find in our results this exit phenomenon even when we just focused on the doctors who are doing high risk.  I think what you have suggested is that if we also looked to doctors doing low risk, we might see comparable exit and that might encourage us to do more case studies and not limit them to high risk. 

On the V-backs currently doing research with my colleague, Yasutora Watanabe , has this wonderful database of basically every malpractice suit that has been filed in this country since time immemorial, and we are studying what happens to the choice of caesareans versus vaginal delivery by a doctor who has been sued or whose colleagues have been sued.  And I hate to state preliminary findings because the referees have not had the chance to tear us to shreds and prove that we are wrong, but we see to be finding, in fact, what you have suggested, which is, once a doctor is sued, there seems to be an immediate and quite measurable increase in the propensity to do caesareans during this time period. 

The one issue that I guess is going to require us to give considerable thought is the emergency patients and specifically the patients who die in the ambulance.  I do not know how big that is, if it is a big number.  It does not take away from anything we have done, but the magnitude of the problem can dwarf the kinds of magnitudes I have been talking about.  The only way to get at this, I believe, would be to go through the Medicare patients who are rather complex combination of Medicare parts A and B, which would give us billings by ambulance companies and combine that with the Social Security death records.  And I think what is nice about that is I think that the grant money involved to do that study would be in the mid six figures at least.

John E. Calfee:  Thank you.  Thank you.  I want to ask a couple of other questions for the panelists preceded by a general observation or two.  And some of the people in this room know, and certainly the people sitting up here know, there is considerable debate about why it is, in some states, the liability insurance premiums have escalated so rapidly.  One possible cause being changes and prevalence of tort liability litigation itself, maybe changes in medical practice.  But the big one that certainly is at least a factor are just the ordinary fluctuations in investment returns and in insurance markets because it is the nature of insurance that premiums are collected and they are paid out much later.  But in the meantime, they are invested and whether or not those investments realize large returns or small returns can have a big impact on premiums. 

And then on tort reform, I think one of the things we know right now is that there are lots and lots of ways to reform the tort system, and some of those ways are very different from other ways.  In other words, you can intuitively expect certain majors to have an impact that might be the opposite of the impact of other measures. 

My question for the panelists right now is looking, I suppose, mainly at David’s paper and some of his implications, are we in the position where we can rule out the possibility of large effects from changes in insurance premiums?

David Dranove:  Well, I think one of the things that motivated our paper was the concerns about the GAO study which only looked at the first two years of the crisis and I think the people are correct to complain that two years in is not a long enough time to really observe the effects.  Four years into the crisis, we still find fairly small effects, and it does raise the question, “Is that long enough?”  And there are actually, I think, has two sides to the answer. 

Obviously, the further at you go, the more people are going to look at their situation and say, “I can be a lot better off somewhere else.”  That might not effect or it might effect incumbent established positions.  It will certainly also affect the choices of new physicians which is something that other people have studied.  But the other thing to note is that for 2005, malpractice premiums in Florida increased by single digits.  I think the median increased across the state.  Across the range of specialties, it was only eight percent.  The picture here is the familiar cycle of malpractice premiums.  They go up steeply, they stabilize, they go up steeply.  This is not the first malpractice crisis.  And if you simply ride it out for a few years, you can expect that things will stabilize. 

One thing we did not talk about also is Medicare builds into its reimbursement rates a kicker for the local malpractice expense.  Now it is based on survey work.  It does not kick in until three or four years after the premiums go up, but you will eventually get paid and, basically, you are just losing about three years worth of that malpractice increase.  And so if it is true that four years after the crisis starts, it seems to be stabilizing in the physicians.  At least those who are treating Medicare patients and those for whom their HMOs and PPOs are using a similar reimbursement rate, they are going to start to get reimbursed for the malpractice expense eventually. 

And so, yes, maybe doctors are just going to wait it out.

John E. Calfee:  And on tort reform, is there any possibility of reaching a very tentative conclusion right now that not only is it true that some tort reform measures have a much bigger impact than others, but the caps may be the most important of all the measures that are usually implemented?

Randy Bovbjerg:  Yes, and yes.  But clearly, caps work in the sense that the people who want caps want, which is to reduce claims payouts, the possibility of running beyond your insurance, lowering¬ insurance.  They take away tort reform and they succeeded in taking it away.  They take away money from the claimants and their lawyers and they move it to the defendants.  That is what it is about.  That is why the doctors want it so much and that is why the lawyers hate it so much.  You can see that politically.  You see it in study after study.  There are an awful lot of press releases with footnotes out there that are called studies that purport to not show that, but it is quite well-established that caps have that effect. 

How well established it is that they really improve patient access to care?  That is what we have just been talking about here.  By the way, the GAO study did have a part 2.  It was not just chasing the anecdotes and doing case studies, they also looked at Medicare data.  I doubt that they had six figures to do it, but they looked at whether the number of procedures went down in one area, which was a very clever thing to do and the number of procedures was not affected.  So affecting the number of doctors and choice of doctors is one thing.  Affecting access to the procedure is quite another thing.

John E. Calfee:  Dave?

David Dranove:  I think the other thing caps do, is they reduce the incentive for lottery litigation.  You have claims out there wholly non-economic damages, and if non-economic damages are untrammeled and they are subject only to the whims of the jury’s imagination, a lawyer is more likely would to be willing to pursue a case that only has a small chance of success for hopes of the large payout.  If on the other hand, there is not a possibility of that lottery payout, that lawsuit will not be brought at all and that also reduces cost, it reduces the litigation effect on doctors. 

With the access question, the argument for malpractice liability at all is that aside from compensating patients, it is needed to deter doctors from performing poorly, and what we have seen is that it is effectively random.  Every neurosurgeon got sued just about every obstetrician gets sued.  There isn’t any basis for saying that five percent of the doctors were responsible for half the claims as the myth goes, other than perhaps if you ignore the difference between high-risk doctors and low-risk doctors because if you are a neurosurgeon, you might be within that five percent.  You would expect it to have an effect on doctors eventually in terms of access.  And then some of the effect on doctors, it has to work both ways, and maybe the positive effects of that incentive, that way the negative effects within that incentive, but that there should be access effects if there are any effects on doctors at all.

David Dranove:  There is this widely held perception of the legal system being somewhat of a lottery and unfortunately, the research on this area is very, very thin.  There is really one seminal paper that I would like to just mention.   It was done by Hank Farber and Shelly White and published in the American Economic Review about 10 years ago.  And they studied what happened to a single hospital that got sued from malpractice.  And the hospital, like most hospitals who gets sued for malpractice, hired an independent consulting firm to evaluate the data and determine whether or not the hospital was likely to be held liable by a rational jury. 

First, they made an assessment as to whether in fact the hospital was at fault because when you are sued, there is a small expense upfront.  A large percentage of the cases are settled or dropped immediately after discovery.  This lawsuit is just to initiate discoveries so the patient could find out whether in fact something screwed up.  ¬That is not very costly to doctors.  What is very costly to the doctors is, if the legal team for the plaintiff discovers there really is some cause, he had really some chance to win and then it is very costly if it goes to trial and, of course, if there is a big judgment. 

Well, what this hospital did was whenever the legal team told them that they were fine, that they did not commit a tort, they refused to settle.  And whenever the legal team told them you committed a tort, they tried to settle immediately.  And what the study further did was follow the process and eventually, I cannot remember the exact numbers, there are eventually about 40 cases that went to trial.  About 20 of which the hospital was deemed by the consultants to have actually been at fault, and about 20 in which the consultants said the hospital was not at fault at all. 

In terms of the lottery, out of those 20 cases where the hospital was not at fault, the hospital won every single jury trial, and in the roughly 20 cases where the hospital was at fault, the hospital won 19 of them.  On study, but, boy, this lottery effect really … we really should figure out what exactly is going on here because maybe this is misperception, and it feeds patients’ views, it feeds the lawyers’ views, and I think it creates a real problem with all the excessive lawsuits.

Randy Bovbjerg:  Big topic.  Let me start with some common sense about lotteries.  How many of you have ever bought a lottery ticket and thrown it away before the drawing?  Jesus, not many.

Jon E. Calfee:  You bought one as opposed to being given a ticket?

Randy Bovbjerg:  Bought a ticket.  Yes.  It was not just at the kid’s school and the prize was the goldfish.  It was the power ball.  You bought a ticket, and you just threw it away.  Anybody?  Okay, as we just heard, an awful lot of cases closed without payment early on.  They threw away their lottery ticket.  Now, it is not just like a lottery ticket because it is not just the sum cost of buying it.  You have to prosecute the case, but nonetheless.  On the other hand, if the other end of the process you observe behavior, plaintiffs lawyers go into a jury room, excuse me, into a courtroom of the jury and at the end, the data are quite right, but all the cases resolved by a jury, only 20 percent are won by the claimants.  So what you would expect, what the econometricians would tell you is that given various reasonable assumptions, it should be 50-50. 

Okay, and that is what it is on auto.  Why is it only 20 percent in med mal?  Well, I think this and then the lottery effect, what there is, is a jackpot effect.  Got a bunch of sunk cost, we have the chance of a retirement amount.  There is that at the end and it is terribly unpredictable in an individual case.  Those lawyers are still negotiating when the jury is out.  They do not know what is going to happen.  Beforehand, they may have reached a secret high-low agreement that regardless of what the jury says, the claimant would get at least this much and not more than that much.  Why, because of this jackpot unpredictability.  So that is very clearly there.  You can see that from behavior. 

There is more than one study about this.  Do not forget most of the stuff now is run not just on the advice of consultants, but by physician- or hospital-run liability insurers whether those are mutuals or whether they are trust or whether they are offshore wholly-owned company, etcetera, etcetera, there are a lot of arrangements.  You have got the experts deciding this.  An early study of the behavior of the physician mutual in New Jersey took as the gold standard their own peer review of the cases.  Many, many more cases than for the one hospital study that you talked about.  They found pretty reasonable consistency between legal outcomes and their own physician panel’s judgments, okay?  When the cases did go … and do not forget, this is 90-95 percent of cases are being resolved not by these unpredictable jury process but by the experts on both sides.  That does not mean that the damages are not off, it does not mean the mistakes do not happen, it does not mean that that erasures in the record, as opposed to actual care does not make a difference, it does not matter that the doctor being a drunkard, but was not drunk that day. 

All these things matter and there is theatricality, but by and large, it is not as bad as the doc say.  It is not as good as the lawyers say.  The big study by the anesthesiologists, who were the one shining example of where the legal system, in an indirect way through national premium volume, actually affected and made better care, the anesthesiologist reviewed every single closed claims for many, many years and they tried to figure out did something go wrong, 

Was it something that we could prevent?  How can we write new standards?  What can we do differently to get better?  And they learned, and they decided that most of the cases that they reviewed, there was indeed negligence that did indeed … where there was payment.  It did indeed involve, but did it affect care?  But to the end, something like 40 percent of the pay claimants, it was not.  And do not take that number.  Go to the website and read where it is written down because I’m relying on my memory here.  That was a very, very big study. 

And there are lots of things that indicate that this is not a lottery, but it sure is a lot more haphazard than we would like, and there is no reason that it should be so haphazard and a lot of it has to do with not being a system.  It is not really an insurance system, it is not really a deterrent system, it is really a dispute resolution system.  That is all it does.

John E. Calfee:  I want to get to the audience so we are going to allocate one of these things.  And I will point as an economist that however rare the jury trials are, and however rare the large judgments are, those trials and those judgments are the governing factors in the negotiations that determine most of the cases.  Okay.  We are having questions with the audience and please identify yourself because this will be taped and we want everyone to know who you are.  Jim?

Jim Frogue:  Thank for this panel.  I just want to start by…

John E. Calfee:  And if you could identify yourself for the rest of us.

Jim Frogue:  Jim Frogue, the Center for Health Transformation.  I start with correcting something that Ted said.  The key stop between L.A. and Vegas and Barstow is in the In-N-Out Burger right off the 15.  So I’m sure you know that.  I was just wondering, this is a little bit tied outside the scope of the panel but Texas enacted some pretty serious tort reform in 2003.  And the evidence so far is that the number of doctors, specifically in higher specialties like neurosurgery and Ob-gyn, has gone out markedly, particularly in under-served areas.  And med mal premiums have actually declined 30 up to 40 percent in real terms, which is highly unusual for any state.  So Texas is a really interesting case study and Missouri did similar reforms in spring of last year, so it is certainly probably too early to get in the evidence there but the early evidence out of Texas is very striking.  So, I’m wondering if you have any reactions to that.  Anyone on the panel.

Randy Bovbjerg:  You got six figures so we can go study it?

John E. Calfee:  There is more research to be done than one unanimous conclusion of every panel we have ever had.  Oh, sure.  Ted.

Ted Frank:  Texas is also a good example of what I spoke about earlier because there, the legislator did pass reforms.  The court struck it down and they had to go to the voters and get a constitutional amendment to make the reforms happen.  So that is an example of the problem in measuring when a reform takes place.  But I think the huge decreases in insurance cost in Texas certainly show up the myth fact that caps do nothing on insurance rights.

David Dranove:  I would just ask that you ask the state of Texas legislature to make their data available through the healthcare utilization projects so that we could actually study this systematically rather than rely on reports, kind of anecdotal from data that cannot be scrutinized independently.

John E. Calfee:  Anybody else  - way in the back… all the way in the back.

Norman:  Norman [indiscernible], I’m a thoracic and vascular surgeon here in Washington, D.C.  I just think there may be a couple of comments that I can tell you that my medical malpractice premium this year is $110,000.  Obstetricians in the District of Columbia probably pay anywhere from $120,000 to $130,000 per year.  We have no tort reform in the District of Columbia.  There has been a campaign for many years to try to get this accomplished.  It has been unsuccessful because it has been basically destroyed by the plaintiffs’ attorneys.  The costs in many of the malpractice suits are not trivial because there is a defense team and the defense lawyers have to make money, too.  So the average amount of money that your insurance company will expend trying to help you is anywhere between $20,000 and $30,000.  That is even if you do not need to go to trial. 

One of the other things that the gentleman in the left mentioned about Medicare covering the cost of the malpractice premiums - that is correct because in the RBRVS, there is a component.  Again, that component is trivial and does not even come close, even if you advance it over three to four years that is going to come close to paying off our premiums.  And for the past 10 years, Medicare has decreased physician’s income and thereby the RBRVS lie anywhere from two to five percent.  I think physician’s income over the past 10 years have been reduced by about 50 percent.  All of this liability  is also having its effect on, you mentioned briefly, the choices that medical students make and who goes into what residencies, and I think this will certainly play out down the line as far as manpower is concerned.

John E. Calfee:  Thank you.  Are there any…?

David Dranove:  Sure.  I mentioned about a study Bert Chiaris.  He was a doctoral student of mine.  He actually studied residency choices of new physicians and found no connection between malpractice premiums and tort reform and residency choices, [the report] did not say it is not going to happen, but at least that it is not as clear cut. 

In terms of the malpractice adjustment, it is meant to only account for the Medicare portion, and it is based on surveys, it is based on the specific procedures the doctor performs rather than their specialty.  It is supposed to account for the full malpractice premium or the Medicare proportion of the full malpractice premium.  It is however lagged three to four years.  So in 2006, you would be receiving less than half of what we … well, less than half of the increase that you have already experienced.  It would have already gone into effect so you have a long way to go before you finally get the full increase.

Randy Bovbjerg:  I think the comment brings forward something that I often say earlier, we all sit up here and we talk about footnotes and regression models and we use cartoons and it is easy to forget that there is an incredible amount of very real pain out there for someone who thinks they have been injured by a medical care, just injured in general and, of course, for the physician who makes a mistake or is unjustly accused of making a mistake. 

The really, really big picture is, does this system of liability law and courts and liability insurance that we have, is it really doing what we want, and is it really reducing the cost of accidents and the cost and pain of coping?  And I think the jury is very, very much out on that but that goes way beyond these papers.

David Dranove:  So one more comment.  You talked about all the upfront cost of all the teams of lawyers.  I’m sure the panel says… I do not know if the audience understands just how expensive our liability system is.  We know in healthcare in general for health insurance, administrative cost in total account for somewhere between 10, 12, 15 percent of total healthcare cost, so that for every dollar that goes to the doctors, about 20 cents goes to somebody else. 

In the legal liability system, for every dollar that ends up going to the victims of tort or those who have suffered a bad healthcare outcome, more than a dollar ends up going to lawyers and expert witnesses.  It is an incredibly expensive way of compensating victims.

John E. Calfee:  Yes.  To reinforce that, the classic justifications for the modern liability system since the last 30 or 40 years are to provide compensation or some sort of an insurance and to improve the quality of care.  And I know of almost no one who defends the system as a compensation system.  And I think it is probably fair to say that right now the main debate is over whether or not the liability system on the whole improves the access to care and the quality of a care, or whether it actually makes it worse in some sense.  And I think that what the literature tells us is it is extremely difficult to resolve those questions.  It is very, very hard to get the data you need.

Do we have more?  We have another one way in the back corner.

Randy Pate:  Randy Pate with the Heritage Foundation and just as a follow-up to your comments about… well, the jury is still being out on the tort system as far as improving the quality of medical care, and that is something that an international comparison I think is useful for.  And there was a health affairs article that came out over the summer that said that Canadian physicians, for example, American doctors are sued 350 percent more often and also 50 percent more than British physicians and Australian physicians, and these are all similar common law traditions we are looking at here.  I mean, their medical systems are not… they are not experienced [audio skips] simple measure of why our tort system is not working to improve medical care.  If we could only measure error rates … Randy?

Randy Bovbjerg:  Well, actually, the only attempt I know to measure error rates in Australia compared with the United States was people in Australia who adopted the so-called Harvard approach to reviewing medical records and they found a higher error rate in Australia which I imagined some trial error somewhere has used, but what the people in Australia thought was that there were some researcher bias.  They thought they were really was not that much difference, but that they were trying really hard to find them because they were going to use them for improvement, and the researchers who did the other work were trying, in some sense, to consider the feasibility of alternate compensation system which will be more expensive if you found more.  So it is very, very hard to measure error rates across countries. 

There is just no agreement here and that is another thing that is not appreciated.  We all talk about negligence, but we only know it when we see it.  It is not an observable fact like a certain DRG or a low-weight baby or a deaf.  Those are observable facts.  Negligence is a conclusion of some process.  Preventability, mistake, and the process can be peer review of the process; it can be state disciplinary hearing, the process could be a court, process could be arbitration and there is not any agreement.  That is partly why this is all so expensive.  So one of the reforms I would like to see tried is to systematize and make lists of outcomes that may not be always absolutely preventable, generally preventable just to make this work better. 

I’m also in favor systematizing damages, so you do not have so much haphazardness.  Mrs. Jones in courtroom A and Mrs. Smith in courtroom B, they have no chance in hell that they will get the same result in the same case.  The system has no way of achieving that.  And if the trial lawyers looked at social security disability payments and saw that kind of unpredictability, they would have it declared unconstitutional as arbitrary and capricious in a New York minute.  That is one minute, but probably same zip code but… it is mind-boggling. 

And as a lawyer, when I went and first studied this and I read the jury instructions on damages, so-called non-monetary general damages, I was just embarrassed at the imposition there and the degree of variation that we come out with.  I think it is possible to do better, whether you are talking to the tort system or whether you are talking of an alternative to the tort system that we really ought to be asking, are these things doing what we want?  And instead, we are arguing about does tort affect insurance rates, is there a little bit of difference on access?  That is just published and they are getting better at it, but the real big issues are different.

John E. Calfee:  Right here.  Over here.  The mic will be there in a moment.

Joel Finkelstein:  Hello.  My name is Joel Finkelstein.  I’m freelance reporter.  You were saying that no one really cares whether the doctors are paying more premiums.  It is really an issue of whether patients have access.  And I’m wondering whether this research or previous research can give us an idea of whether there is a threshold effect in terms of mass dropouts, and whether we are heading towards a point in which there could be sort of catastrophic exit from the market for high-risk physicians.

David Dranove:  I do not know if there is a threshold effect in the sense that all of a sudden, everything just goes to hell in a hand basket, got to finish that expression.  However, I think the important starting point is to realize that 20 years ago, the research is already showing that if you are a bright young college student looking for a career to take advantage of your skills, and you had a choice between becoming a lawyer, working for three years and then going to a business school or going to medical school and then spending six years in residency and then started to earn an income  - from the net present value perspective, you are actually better off choosing one of the former two career paths and actually going into medicine. 

As managed care took root in the 1990s and especially with HMOs putting pressure in doctor’s incomes, we actually started to see some effects on specialty choices and medical school applications.  Then with the managed care back, class physician income started to go back up again and we start to see an increase in applications to medical schools.  But now, with the great uncertainty about Medicare, things are starting to go the other way, so it is absolutely the case that medical school applications and specialty choices do respond to these income opportunities.  And if there is rough equality across the different professions and we start to create a substantial inequality, then you could have maybe that slippery slope that you are talking about.

Randy Bovbjerg:  The part better evidence of it: a precipitous insurer exit from a place that they think is just unpredictable than of doctor exit.  But let me address what we just heard about doctor and lawyer choices. 

First of all, an awful lot more is involved in the choice of a career than money.  What else are we doing what we are doing, okay?  It is not money.  And the second thing is my former colleague, Bob Lee, in fact 25 years ago actually did a study of the net present value of being accepted to medical school.  And at that time in about 1980 dollars, it was a million dollars, something like that.  It was a lot, just the fact of acceptance because almost everybody who gets in, finishes and almost everyone who finishes got a good job. 

Now, it has become much harder and as we heard from the surgeon in D.C., it really is not a matter of just hanging out their shingle and charging what you want.  But do not forget people go to law school, a lot of them do not end up as lawyers.  It was quite a sobering thing for me to go buy a bottle of sherry in my first year in law school, show my law school I.D. and the guy bagging the sherry said, “Oh, I have a law degree.”  Doctors compare themselves to high-priced lawyers and certainly compared with a successful contingent fee attorney, doctors are pikers.  Certainly compared with a successful business executive, doctors are pikers.  But if you look at it from the certainty and the average probability of what you are going to get from going to medical school versus law school, you can work your butt off much harder at medical school, but I think you can do better.

David Dranove:  Well, I do not want to… the battle of citations.  I was referring to Monica Nothers’ dissertation from 1985, University in Chicago.  And Monica I think is well-known to the American Enterprise Institute.  I was on her committee and I think she did the work, right?

Ted Frank:  Anecdotally, I went to law school.

John E. Calfee:  Anecdotally, he graduated.  Anything else?  We will go back here.

Barbara Marone:  Hi.  I’m Barbara Marone with the College of Emergency Physicians and I appreciate getting the benefit of some of this additional research but as practitioners and people in advocacy, it seems a little bit arcane.  And one of the things, if you will, one of the things I keep wondering about is the kinds of research that is going to be required to understand what is over utilization, what is unjustifiable variation because all we hear from Medicare is, “We got to reduce the medical or the diagnostic imaging.  We got to reduce variation.” 

But it just seems to be a real dearth of what constitutes that, and as an anecdotal comment, we get a lot of third year emergency residents coming through our office for three or four weeks at a time.  And they all say the amount of imaging they do in the emergency department is basically foisted on the them and said, “You must do a scan, you must order this, you must order that for your protection and ours,” and I just wonder if the objective of the tort system is to have more quality care out there, how do you reconcile that with the driving up potential over utilization?

Ted Frank:  There is certainly… you are describing a defen