By Fred J. Hellinger, Ph.D* and William E. Encinosa, Ph.D.*
* U.S. Department of Health and Human Services, Agency for Healthcare Research and Quality, Center for Organization and Delivery Studies, 540 Gaither Road, Room 5319, Rockville Maryland 20850. Phone: (301) 427-1408; Fax: (301) 427-1430; E-mail: FHelling@ahrq.gov
Select to download print version (PDF File, 234 KB). Plugin Software Help.
Contents
Abstract
Introduction
Background
Methodology
Results
Discussion
References
Abstract
Researchers at the Agency for Healthcare Research and Quality (AHRQ) have
examined the impact of different kinds of State laws in a number of previous
studies. This study examines the impact of State legislation that caps damage
awards in malpractice cases on decisions of physicians about where to practice medicine.
Twenty-four States now have laws that limit damage payments in malpractice
cases. Most of these laws limit the amounts paid for noneconomic damages
(e.g., pain and suffering) but a few limit both economic (e.g., medical expenses
and lost wages) and noneconomic damages. There is currently a national debate
on the desirability of extending caps on malpractice damage awards to all
States, and President Bush recently introduced a proposal to cap payments
for noneconomic damages in medical malpractice cases at $250,000.
Supporters of legislation to cap damages in malpractice cases maintain that
it reduces malpractice premiums and helps insure an adequate supply of physicians.
They also assert that escalating, multi-million-dollar jury awards are driving
malpractice premium increases and that capping damage awards for pain and
suffering helps restrain the rate of increase. Without such a law, it is
asserted that the loss of affordable medical malpractice insurance for physicians
could eventually lead to the loss of affordable, accessible health care.
Opponents of this legislation maintain that insurance companies are trying
to compensate for poor business decisions and fading investment income.
Although there is some evidence in the literature demonstrating that physicians
in States with tort reform laws capping malpractice awards enjoy lower malpractice
premiums, there is no evidence about the impact of malpractice cap legislation
on decisions by physicians regarding geographic location. This study is the
first to supply such evidence.
A simple comparison of the supply of physicians per capita between States
that did and did not adopt a cap revealed that States with caps experienced
a more rapid increase in their supply of physicians. In 1970, before any
States had a law capping damage payments in malpractice cases, States that
eventually adopted a cap and States that did not eventually adopt a cap had
virtually identical levels of physicians per 100,000 citizens per county (69
vs. 67). Thirty years later in 2000, States that adopted a cap averaged
135 physicians per 100,000 citizens per county while States without a cap averaged 120.
Adjusting for a variety of factors in a multivariate regression model, we
found that States with caps on noneconomic damages experienced about 12 percent
more physicians per capita than States without such a cap. Moreover, we found
that States with relatively high caps were less likely to experience an increase
in physician supply than States with lower caps.
Return to Contents
Introduction
In recent months, physicians in New Jersey, West Virginia, and Florida have
conducted work stoppages in response to the rapid increases in malpractice
insurance premiums and in support of legislation limiting payments for noneconomic
damages in malpractice cases.1,2 Malpractice premium rates for internists, general surgeons, and obstetrician/gynecologists increased 25 percent, 25 percent, and 20 percent, respectively, in 2002;3 and last year, legislation limiting noneconomic damage awards in malpractice cases was signed into law in Nevada and Mississippi.
This year bills limiting noneconomic damage awards in malpractice cases have
been signed into law in Ohio and in Texas.4,5,6
There are now 24 States that have a law that caps noneconomic damages or
a law that limits total damages: Alaska, California, Colorado, Hawaii, Idaho,
Indiana, Kansas, Louisiana, Maryland, Massachusetts, Michigan, Mississippi,
Missouri, Montana, Nevada, New Mexico, North Dakota, Ohio, South Dakota, Texas,
Utah, Virginia, West Virginia, and Wisconsin.
We include States that limit total damages only (Indiana, Louisiana, and Virginia), as well as Colorado, which has a law that imposes separate limits on economic and noneconomic damages,
and New Mexico, which has a law that limits total damages less punitive damages and medical expenses.
Proponents of tort reform maintain that the size and frequency of large jury
awards and settlements in medical malpractice cases is behind the rapid increase
in malpractice insurance premiums and that legislation limiting damage awards
is necessary to stem these increases. They also maintain that high malpractice
rates are driving physicians out of business or to States where there is legislation
capping malpractice awards.7,8,9
The market for medical malpractice insurance is volatile, and there have
been numerous "crises" in this market over the past three decades.10 In response to a crisis in the early 1970s, California passed the Medical Injury Compensation Reform Act of 1975 (MICRA) limiting noneconomic damages
in medical malpractice cases. MICRA is often cited as a model for State legislation.
Research has shown that between 1975 and 2000, malpractice premiums grew
more slowly in California than they did in the rest of the Nation (167 percent
vs. 505 percent).11
A recent publication of the American Medical Association (AMA) discusses
the determinants of professional liability insurance (PLI) rates:12
"The increase in the frequency and amount of very large awards may be one
of the significant drivers of the rapid escalation in PLI costs. If this is
true, then one would expect, over time, that PLI rates in states that have
effective damage caps would diverge from the PLI rates in states that have
effective tort reform."
There is a sizable body of economic literature demonstrating that the legal
environment in a State affects the frequency of malpractice claims and the
size of the awards.13
For examples, Zuckerman, Bovbjerg, and Sloan demonstrated that physicians
in States with caps on damages in malpractice cases experience lower premiums
than physicians in States without such laws.14
Danzon found that damage awards in States with caps on damages were 23 percent
lower than in States without caps.15
In another article, Kessler and McClellan examined the impact of tort reforms
on the practice of defensive medicine and found that tort reforms such as
reasonable limits on noneconomic damages, which have been in effect in California
for 25 years, can reduce health care costs by 5 percent to 9 percent without
substantial effects on mortality or medical complications.16 Proponents of tort reform legislation emphasize that only 28 percent of physician
payments for malpractice insurance are allotted to patients and that the remaining
72 percent are consumed by administrative and related costs.17
Opponents of tort reform legislation that caps damage awards in malpractice
cases maintain that poor quality and poor investments by insurance companies
are to blame for the recent spike in malpractice rates. They argue that caps
will harm those patients who suffer the most damage and who need help the
most, and that payments for medical malpractice claims are not the underlying
cause of rapidly increasing malpractice premiums. A recent article states:
"According to the Consumer Federation of America, the average pay-out by
medical malpractice insurance companies is about $30,000 per claim and has
been virtually unchanged for the last decade."18
Although there is little agreement about the underlying causes of increases
in malpractice premium rates, there is little dispute that rapidly increasing
malpractice premium rates have mobilized physicians and engendered considerable
support for legislation limiting malpractice damage awards.19 Increasing rates for malpractice premiums and calls for tort reform coincide
with increasing concerns about access to care. A recent BlueCross/BlueShield
publication adds:
"What is not in dispute is that the medical liability problem has gained
prominence at a time when public concerns about access to care and the cost
of that care have re-emerged with new strength."20
Supporters of legislation capping malpractice damage awards maintain that
this legislation is necessary to assure adequate access to health care. One
newspaper article points out:21
"The American Medical Association says patients' access to care already is
seriously threatened in a dozen states and a crisis is looming in seven others
because of rising premiums for malpractice insurance."
A 2003 report by the U.S. Department of Health and Human Services has stated:
"Increasingly, Americans are at risk of not being able to find a doctor
when they most need one. Doctors have given up their practices, limited their
practices to patients who do not have health conditions that are more likely
to lead to lawsuits, or have moved to states with a fairer legal system where
insurance can be obtained at a lower price."22
And, last year another article reported:
"Nationally, medical liability insurance rates have skyrocketed with several
states facing a meltdown of their health care system as a result. In the
states with the fastest-growing rates, doctors have begun 'running bare',
without insurance coverage, or have left the state altogether."23
Return to Contents
Background
Two types of liability are germane to this study: contract and tort.24,25,26 Contracts are voluntary agreements entered into for significant benefit between
parties, and contract liability involves implementing the provision of contracts.
Contracts specify in detail the services that will be afforded, and the liabilities
created by contracts are limited to the cost of the services specified in
the contract (e.g., there are no punitive damages for breach of contract or
liability for unanticipated outcomes following the breach of contract). This
certitude and the limited liability required under contracts have been an
effective mechanism by which to assist fruitful relationships among distinct
contributors in our economic system, and courts have been hesitant to void
the provisions of contracts between consenting parties.
Torts are civil wrongs where the injured person asks for monetary damages
from an individual in a situation where there is no contractual relationship.
Tort law sets in place public procedures about how people and businesses are
anticipated to act toward one another. Most people who are engaged in a "learned
profession" may be sued in tort for malpractice (e.g., negligence claims by
patients against their physicians for malpractice are tort claims). Compensation
in malpractice cases may consist of expenses for all harm endured by the patient
counting medical care costs, lost wages, pain, and suffering, as well as punitive
payments in situations where there was malicious intent.
Return to Contents
Methodology
The theoretical structure underlying the empirical analysis in this study
is that one of the factors taken into consideration by physicians in selecting
a site to practice is the market for medical malpractice insurance.27 In particular, it is hypothesized that physicians are more likely to settle in a State with a law that limits their exposure to malpractice damage awards.
One recent newspaper article maintains:
"On a much broader level, it the litigation crisis] brought new attention
to a national problem that doctors say is obliging many of them to flee certain
states or give up certain specialties—or the entire profession—because
of skyrocketing insurance premiums linked to soaring jury awards."28
And another adds:
"Yet while the doctors will be the ones to feel the pain first, it is the
patients who will do the real suffering, perhaps, in the form of higher fees,
and in declining health care as more doctors hang up their surgical gowns."29
Our model presupposes that factors affecting the demand for physician services
also affect the geographic distribution of physicians. For example, recent
research has shown that economic development measured by per capita income
is positively correlated with physician supply across a variety of countries.30 In our study, we presume that States with higher personal incomes are more desirable locations in which to practice because they have a higher demand
for health services, and this, in turn, will result in higher physician incomes
and a greater supply of physicians. For this reason, we include personal
income in our model.
Similarly, we presume that States with higher unemployment rates are likely
to have a lower demand for health services and this will result in lower physician
incomes. As a result, we include a State's unemployment rate in our model.
Because of the longer distances involved in seeing patients and the relative
scarcity of health care resources, it is assumed that physicians will be more
likely to settle in more densely populated areas. In discussing States where
physicians have a problem in obtaining affordable malpractice insurance, a
recent newspaper article maintains:
"Larger malpractice claims mean higher insurance premiums and more money for
trial lawyers. They also mean fewer doctors, particularly in the states most
affected. Within those states, the hardest hit communities are rural, where
a doctor's income is not enough to offset higher premiums. Those doctors will
leave the small towns for the cities, leave the state for a more friendly
environment or simply quit practicing."31
For this reason, we include a variable that measures the number of citizens
(measured in thousands) per square mile for each State. Older persons have
a greater demand for health care services than younger citizens due to the
increased frequency of illness. Moreover, persons over the age of 65 are
almost always covered by Medicare. Thus, it is hypothesized herein that physicians
will be more likely to settle in areas with relatively high proportions of
elderly citizens. Consequently, this study includes a variable that measures
the proportion of each State's population that is 65 years or older.
The proportion of persons working on farms is assumed to be negatively related
to the demand for health services. Farm workers are more likely to lack insurance
and receive low wages and thus are expected to have little disposable income
to spend on health care services. Consequently, a variable measuring the
percentage of the State domestic product (i.e., a measure of the value of
goods and services produced within a State) attributable to farm activities
is included in the model.
This study estimates the impact of State laws limiting damage awards in malpractice
cases on physician availability first using statewide aggregate data and then
using county data. Physician availability is measured by the number of active,
non-Federal physicians practicing in each State per 100,000 population using
data provided by the AMA. The primary independent variable of interest is
set equal to 1 if the State has a law that limits the level of damage awards
and zero otherwise. That is, this variable is set equal to 1 for the 19 States
listed in Table 1A (excluding Alaska).
The aforementioned variables are utilized in the analyses based on State
data. The State-level analyses are conducted on State characteristics at
four points in time: 1985, 1990, 1995, and 2000. To test the robustness of
these State-level analyses, we perform an additional analysis at the county
level for the final 5 years (1996-2000) using two additional control variables
available for these years of county data.
First, in our county-level analyses, we use a variable set equal to 1 if
a county has a hospital with a physician residency training program, and we
hypothesize that this variable has a positive coefficient because medical
residents are more likely to settle in areas where they have trained. We
do not use this variable in the State-level analyses because every State has
at least one hospital with a residency program.
Second, in the county-level analysis, we are able to control for the county's
health maintenance organization (HMO) enrollment. We use a variable set equal
to 1 if the county has high HMO penetration (an HMO enrollment above 30 percent)
at the midpoint of the 5-year period: 1998. We hypothesize that physician
availability will be lower for counties with high HMO penetration since HMOs
tend to restrict patient access to doctors through closed networks. We do
not use this variable in the State-level analyses because of the high correlation
between population per square mile and HMO penetration.
Physician availability is measured by the number of active, non-Federal physicians
practicing in each county per 100,000 population. In addition, in the county
analysis, we derive a measure of rural influence from a variable constructed
by the U.S. Department of Agriculture that is available in the Area Resource
File (ARF). We hypothesize that this variable, which we refer to as "ruralness,"
is negatively related to the supply of physicians.
We also use a variable measuring the number of births per capita in each
county. This variable measures the youthfulness of the population, and we
hypothesize that it will have a negative coefficient in our equations.
A variable measuring the unemployment rate in each county also is included.
However, we do not utilize a variable that measures the proportion of income
attributable to farm activities because this information is not readily available for counties.
Finally, we also include a variable that is set equal to 1 if the county
has an average annual temperature of 70 degrees or higher. We hypothesize
that doctors may tend to set up practice in temperate climates of the country.
Moreover, the elderly tend to retire to these areas, and they require a greater
level of physician services.
We estimate our model using State data and then county data because these
approaches have offsetting strengths and weaknesses. The empirical analyses
utilizing State data provide information about the effectiveness of State
laws limiting damage awards on the supply of physicians in each State. And,
because we are interested in ascertaining the impact of State laws on physician
supply in a State, the use of the State as a unit of observation is reasonable.
However, models using State data provide a relatively blunt instrument to
assess the impact of a law that limits payments for damages in medical malpractice
cases because this approach obscures the impact of variables within specific
markets within a State.
Analyses based on county data include information about counties with different
characteristics within each State. Thus, analyses based on county data can
tell us whether a county with a hospital that has a residency program has
a larger supply of physicians than a county without such a hospital.
Moreover, the use of county data may be more appropriate than State data
to the extent that the impact of specific variables is felt within each county
rather than within each State. For example, the unemployment rate of each
county (as opposed to the unemployment rate in the State) may be a better
measure of the impact of unemployment on physician supply in a given county
than the unemployment rate in the State. However, in cases where the market
for physician care extends beyond a county's border, the use of the county
as the unit of observation may distort estimates of the impact of the law.
Adjusting for the simultaneous impact of multiple
factors (i.e., independent variables including the existence of a State law
limiting malpractice damage awards) on the dependent variable is accomplished
using multivariate linear regression analysis.
Coefficients for the independent variables in our
multivariate linear regression analysis are estimated using least-squares
estimators (i.e., the estimated coefficients are obtained so that they result
in the lowest sums of squares of the differences between the actual and estimated
value of the dependent variable). This model is estimated under the usual
assumptions that the relationship between the dependent and the independent
variables is linear and that the error term is normally distributed.32
The robust standard errors in the county analysis
are heteroskedasticity-consistent and are corrected for clustering at the
county level. Influential outliers were removed from the county data: about
30 counties were dropped since they were coded with either less than 10 doctors
per 100,000 residents or over 1,000 doctors per 100,000 residents. This was
less than 1 percent of the county sample.
Data
Information about State medical liability laws was obtained from the National
Conference of State Legislatures (NCSL),33
the American Tort Reform Association (ATRA),34
and from publications of a large law firm.35
The NCSL provides a listing by State of all State medical liability laws
that includes the type of reform implemented (e.g., limit on economic and
noneconomic damage awards) and the specific legislation that enacted this
reform. In 1994, the ATRA created a publication that displayed the status
of each State law on medical liability. This publication has been updated
several times since that time, and it is currently available on the ATRA Web site.
McCullough, Campbell & Lane is a large general practice law firm located
in Chicago with a specialty in insurance law, and this firm publishes a compendium
of all legislation relating to medical malpractice for each state. This compendium
is available on the McCullough, Campbell & Lane Web site (http://www.mcandl.com/states.html).
These data sources were used to ascertain the date of the legislation enacting
state laws that limit damage awards in medical malpractice cases (Table 1A). Five States enacted legislation capping awards before 1985, and the
dummy variable for the cap variable in our 1985 data set was set equal to
1 for each of these five States. Each of these laws was enacted in 1975 or
1976 in response to the medical malpractice crisis in the early 1970s.
Ten States enacted laws implementing damage caps in malpractice cases in
1985 or 1986 in response to the medical malpractice crisis in the early 1980s.
The 1986 Alaska law was exceptional among these laws because it excluded cases
involving physical impairment or severe disfigurement, and it is uncertain
how many malpractice cases were subject to this exclusion. In any event,
we excluded Alaska from our analyses because of this ambiguity and because
the empirical relationship between factors affecting physician decisions whether
or not to locate in Alaska is likely to be quite different from this relationship
for other States. The dummy variable for the cap variable in our 1990 data
set was set equal to 1 for each of the nine States (excluding Alaska) that
adopted caps in 1985 or 1986.
Two States implemented legislation capping damages in 1988, one in 1990,
and two in 1995. Thus, we set the dummy variable indicating the existence
of a law limiting damage awards to 1 for the 19 States with such a law (excluding
Alaska) in our 1995 data set and we set this variable equal to 1 for the same
19 States in our 2000 data set (access Table 1A for a list of the States).
Data on State characteristics for the years 1980, 1990, 1995, and 2000 are
used in our model, and these data were obtained from various issues of the
Statistical Abstract of the United States. The following paragraphs
define each variable and indicate the underlying data source.
The variable population per square mile of land area was derived from data
on each State's population and its number of square miles as provided by the
U.S. Census Bureau (U.S. Department of Commerce).36 The U.S. Census Bureau issues State population estimates that are updated
annually and are based on the preceding decennial census as well as other
more limited surveys. Data on proportion of the population 65 years or older
for each State were obtained from the U.S. Census Bureau.
Data on State unemployment rates were obtained from the U.S. Department of
Labor's Current Population Survey (CPS).37
The CPS is a monthly, random, national survey of the noninstitutionalized
population in the United States. About 50,000 households are sampled each month.
Data on mean State per capita personal income were obtained from the various
issues of the Survey of Current Business, a publication of the Bureau
of Economic Analysis, U.S. Department of Commerce.
Data on the proportion of the State domestic product attributable to farm
income also were obtained from reports issued by the U.S. Department of Commerce.38 Farm income comprises cash receipts from the marketing of crops and livestock as well as government payments made directly to farmers for farm-related activities.
Information about the number of hospital beds in each State was obtained
from data published by the American Hospital Association (AHA).39 The AHA provides information about the number of hospital beds in non-Federal, short-term community hospitals in each State that are acceptable for registration with AHA.
The data in our county analyses were obtained from the 2002 Area Resource
File. The ARF is maintained by Quality Resource Systems, Inc., under contract
with the Bureau of Health Professions, Health Resources and Services Administration,
U.S. Department of Health and Human Services. The ARF is a county database
that includes statistics on health facilities, health professions, economic
activity, and health training programs. Just as in the Statistical Abstract
of the United States, the ARF uses existing data sources. Indeed, in many
instances, the Statistical Abstract of the Unites States and the ARF
use the same underlying source of data.
The dependent variable in both our State-level and county-level analyses
is the number of active, non-Federal physicians per 100,000 civilians residing
in each State. Both the Statistical Abstract of the Unites States
and the ARF obtain the number of active, non-Federal physicians from the AMA.40 AMA publications contain information about the professional and individual characteristics of all practicing physicians.
Data on the population in each county are based on publications of the U.S. Bureau of the Census. Data on births in each county were obtained from the National Center for Health Statistics, Centers for Disease Control and Prevention (CDC), and data on the unemployment rate in each county were provided by the U.S. Department of Labor.
Return to Contents
Proceed to Next Section