Cost of Health Services Regulation

Working Paper Series

Medical Tort System

Health Facilities Regulation

Working Paper No. MTS-1

Prepared by

Christopher J. Conover

Center for Health Policy, Law and Management

 

Duke University

 

Under contract to the

Agency for Healthcare Research and Quality
With funding from ASPE/DALTCP

May 2004

Draft: Do Not Circulate without Author Permission


Medical Tort System

Background

Rationale.  The medical tort system serves both as a compensation mechanism as well as a potential deterrent to medical injury. Clinicians are expected to be more likely to change their behavior if the odds of getting caught increase and/or the financial consequences of negligence increase.

Statutory Authority. States historically have been responsible for establishing and enforcing rules governing tort liability in medical care and other settings. A number of states have adopted major reforms such as caps on non-economic damages or caps on attorneys’ fees. The focus in this section is on measuring the overall cost of the current “system” rather than measure the potential gains from reforming it.

Key Elements.  In the current system, both facilities and health professionals may be involved in decisions to purchase professional liability insurance. Since it is up to individual patients to decide whether to sue in a given case, not surprisingly a relatively high fraction (98%) of actual victims of negligence opt not to sue. Conversely, less than one in five malpractice claims appear to involve actual negligence (Studdert et al. 2004).

Theoretical Impact

Costs. Critics of the malpractice system argue that it is neither fair nor efficient, too often seeming to be a random lottery in which a lucky few are overcompensated and the majority of victims are undercompensated. There is substantial controversy over whether the current system results in any appreciable degree of defensive medicine, defined as medical services not expected to benefit the patient but which are undertaken to minimize the risk of subsequent lawsuit.  The problem with defensive medicine and its measurement is that “some so-called defensive medicine may be motivated less by liability concerns than by the income it generates for physicians or by the positive (albeit small) benefits to patients” (CBO 2004: 6; see also Danzon 2000 for further discussion). Yet at the same time, there are widespread reports by medical professionals of its presence [Reynolds ????].

Benefits. In theory, negligent behavior will be deterred if the negligent party has to pay the consequence of acting negligently. So in principle, the malpractice system may improve patient safety as well as compensate those who are injured.

Empirical Evidence

·        Government Regulatory Costs. The Institute of Justice at RAND Corporation has estimated that court costs in the entire tort system amount to 1 percent of medical tort system costs (Tillinghast-Towers Perrin 2003).

·        Compliance Costs: Medical Professional Liability Insurance.  Total professional liability insurance in 2002, inclusive of amounts for hospitals and health professionals, and accounting for self-insured losses and loss adjustment expenses for all health institutions and medical providers, amounted to $24.598 billion according to actuarial consulting firm Tillinghast-Towers Perrin (2003). Danzon (2000) reports that patient compensation in the medical tort system amounts to roughly 40 percent of medical liability premiums. In the general tort system, Tillinghast-Towers-Perrin estimates that 22 percent of awards are for economic loss and 24 percent are for non-economic loss 

·        Compliance Costs: Defensive Medicine.  Over the past two decades, a number of studies have sought to empirically measure the nature and extent of defensive medicine:

·        A 1983 report by the AMA Committee on Professional Liability estimated the total costs of professional liability to be $15 to $40 billion for physicians alone. A follow-up study in 1987 (Reynolds, Rizzo and Gonzalez 1987) used two alternative methods to estimate the total cost in 1984 of professional liability for physicians at $12.1 to $13.7 billion or about 15 percent of physician (this would extrapolate to $51 billion in 2002, of which roughly three quarters would represent defensive medicine). These estimates are based on surveying physicians about changes in their practice patterns and by empirically analyzing the effect of professional liability premiums on physician fees and utilization. The chief criticism of such studies is that a) physicians have a financial incentive to overreport their costs; b) higher resource use associated with higher premiums cannot necessarily be attributed to self-conscious responses to tort risks; and c) at least some of the additional spending benefits patients. For all these reasons, results such as these typical are viewed as providing no more than an upper bound on the plausible extent of defensive medicine (Rubin and Mendelson 1994).

·        Other empirical work is less tainted, but nevertheless has limitations. For example, Localio et al. (1993) found that use of caesarean section is positively and significantly related to three measures of malpractice risk, including local premium rates, subjective estimates of the risk of suits (based on surveys of physicians) and the prior claim history of the hospital/medical staff where the attending physician practiced. But as with any cross-sectional study, these findings may be biased due to omitted variables (Danzon 2000), a limitation that combined with small sample sizes typifies the earliest studies in this area.

·        An extensive review of the literature and new empirical work conducted by Office of Technology Assessment found that based on ‘limited evidence,” less than 8 percent of all diagnostic procedures might be attributable to conscious worries about medical (OTA 1993). The study did not reach a definitive conclusion about the scope of defensive medicine concluding that “it is impossible to accurately measure the overall level and national cost of defensive medicine” (OTA 1994: 1) and “leaving open the possibility that defensive medicine costs compose a large fraction of health spending or that they are negligible” (Rubin and Mendelson 1994: 13).

·        Another early literature synthesis concluded that in light of all the uncertainties and limitations of studies extant, a reasonable estimate of the costs of defensive medicine was 40 percent of professional liability insurance (PLI) costs other than premiums. For both physicians and hospitals, PLI costs other than premiums constituted 2.7 times the cost of professional liability insurance premiums (Rubin and Mendelson 1994); hence defensive medicine equals 1.08 x  premiums. 

·        The most definitive studies to date have used a difference-in-difference estimator to control for omitted variables bias, controlling for patient demographic characteristics, state legal and political characteristics and state and time fixed effects. The first of these found that in states with any of four restrictions (caps on non economic or total damages, prohibitions on punitive damages, no automatic addition of prejudgment interest, and offsets for collateral-source benefits) decreased long run hospital expenditures for patients with acute myocardial infarction (AMI) by 5.8 percent; for patients with ischemic heart disease, the reduction resulting from reforms was 8.9 percent (Kessler and McClellan 1996). (We ignore findings regarding a different set of tort restrictions showing spending increases of 2-3 percent as these were only observed in the short run but not the long run). Because there were no significant changes in patient outcomes, it is presumed that these spending differences represent a measure of defensive medicine—i.e., expenditures that could be safely eliminated without apparent harm to patients. If these estimates were applied to total health spending, the savings would amount to $69 billion to $124 billion in 2001 (JEC 2003).

·        To respond to concerns that states with aggressive managed care were more likely to adopt reforms (and hence their findings were attributable to managed care rather than defensive medicine), the authors did a follow-up study that explicitly controlled for managed care penetration, finding smaller but still significant effects. Specifically, they found that direct reforms decreased long run hospital expenditures for AMI patients with by approximately 4.2 percent; and ischemic heart disease patients by 4.4 percent (Kessler and McClellan February 2000a). The authors estimate from their latest findings that “at least for elderly heart disease patients, an untried reform that reduced the legal-defense burden on physicians and hospitals by one-quarter—which is within the range of policy possibilities—could be expected to reduce medical treatment intensity by approximately 6.2 percent, but not to increase the incidence of adverse health effects” (Kessler and McClellan February 2000b).

·        CBO has applied the Kessler and McClellan method to a broader set of medical conditions, but found no parallel evidence that restrictions on tort liability reduce medical spending.  When the same method was applied to a different data set, there were no statistically significant differences in per capita health care spending between states with and without limits on malpractice torts.  CBO regards the question of whether such limits reduce medical spending an open one and is continuing to examine it using alternative research approaches (CBO 2004).

There are two issues with the most recent estimates. First, unlike earlier estimates, they do not purport to measure the full extent of defensive medicine. Instead, they focus on a much narrower question, which is whether tort reforms might be expected to yield reductions in defensive medicine.  A recent analysis of state data from 1985-2001 found that earned premiums per physician were 12.7 percent lower in states that capped non economic or economic damages compared to states without such reforms (Thorpe 2004). Assuming that the proclivity to practice defensive medicine is roughly correlated with the final risk associated with professional liability suits, one could arguably multiply the Kessler and McClellan figures by a factor of 8 in order to approximate the overall amount of defensive medicine that would be eliminated were the current level of financial risk removed entirely.

This relates to the second issue. Kessler and McClellan selected two procedures for which they could measure both spending impacts and also make definitive statements about patient outcomes; their data set by its nature happened to be restricted to Medicare patients.  A conservative approach would limit the estimate of defensive medicine to only hospital costs associated with those two procedures for Medicare-eligible patients. But this would tacitly ignore the widespread evidence of potential defensive medicine in other procedures such as C-sections and diagnostic testing. The most extreme approach would be to use the CBO estimates of no measurable defensive medicine as a lower bound on grounds that they examined a broader set of conditions.  But if the real motivation for defensive medicine is to avoid being sued at all (as opposed to lowering the payout in cases where a suit is won), then even reducing premiums by a modest amount should not be expected to produce any evidence of a reduction in defensive medicine even if the practice of defensive medicine is widespread. While there are legitimate concerns over whether older methods may overstate the extent of defensive medicine, the weight of the evidence overall could not support a claim that there was zero defensive medicine, even as a lower bound.  Such a claim would negate the findings of the best empirical study done to date even if the results of that study are limited to the patient population actually studied rather than extrapolated to either Medicare spending overall or health expenditures by the general population.

Indirect Benefits: Morbidity Gains from Deterrence of Negligence.  In the current system, 46,000 to 98,000 die annually due to medical negligence (IOM 1998). Danzon's recent literature synthesis states "the only credible study of deterrence of medical negligence is from Weiler et al. (1993)." In New York, the medical malpractice system reportedly deters 28.8 percent of all malpractice, but this estimate was not statistically significant, possibly due to small sample size (Weiler et al. 1993). The ratio of negligent injuries to negligent deaths was 2.9:1 in New York (Weiler et al. 1993), but is 10.4:1 in a study of Colorado and Utah (Thomas et al. 2000).

·        Indirect Benefits: Mortality Gains from Deterrence of Negligence.  The percent of hospital patients experiencing an adverse event ranges from 2.9% in Colorado and Utah (Thomas et al. 2000) to 3.7% in New York (Weiler et al. 1993). The percent of adverse events due to negligence ranges from 27.5% in Colorado to 32.6% in Utah. The percent of negligent adverse events resulting in death ranges from 8.8% (Colorado and Utah) to 25.4% (New York).


Net Assessment

We have calculated the regulatory costs in the following fashion (minimum and maximum parameter estimates are shown in parentheses).

·        Government Regulatory Costs. We multiply total medical tort system costs times 1 percent and assume that 10 percent of these are incurred by federal courts, with the balance borne by states.

·        Compliance Costs: Administration. We inflate total reported malpractice premiums in 2001 by 8.1 percent to obtain a 2002 estimate (we assume growth is half this fast as a lower bound and twice this fast as an upper bound).

·        Compliance Costs: Defensive Medicine.  In light of the foregoing discussion, we opted to apply the Kessler and McClellan hospitalization savings of 4.2 percent, applying this to the estimated share of inpatient heart disease spending that is financed by Medicare (based on 1997 disease cost estimates by Cohen and Krauss (2003) applied to 2002 national health expenditures) to produce our lower bound figures. Based on the Thorpe finding of 12.7% premium savings from capped awards, we derived a multiplier (1/.127) to extrapolate the Kessler and McClellan results to a counterfactual world in which there was no tort liability. For our upper bound, we used the identical method applied to heart disease expenditures for the general population instead of just the Medicare population. Since heart disease only accounts for about 10 percent of personal health spending, we then multiplied this estimate by 5 on the assumption that in the worst possible case, the Kessler/McClellan-estimated savings would apply to roughly half of all health spending.

·        Indirect Benefits: Morbidity Gains from Deterrence of Negligence. We use the New York estimate that the tort system deters 29 percent of all negligence as an upper bound on grounds a) it was not statistically significant and b) New York probably has a more stringent-than-average malpractice system relative to other states on average. The expected figure gives equal weight to the null hypothesis (zero effect) and the average estimated effect across NY, CO and UT. It is assumed that the potential percent reduction in negligent injury is proportional to the actual likelihood of a negligent event. That is, the potential for reducing negligent injuries presumably is higher in a state such as New York because the baseline injury rate is higher. For CO and UT we therefore adjusted the 29%  reduction in negligent injury rates by multiplying by the ratio of negligent death rate per 1,000 admissions in these states relative to NY (this indicator was calculated by multiplying hospital admissions times the percent experiencing adverse events times the percent of adverse events due to negligence times the percent of negligent adverse events resulting in death). This indicator was selected as it was the most visible indicator of negligence and therefore is most likely to affect the number of suits filed). Figures from New York and Colorado/Utah studies showing distribution of negligent injuries by severity were used to calculate an average severity score, with a QOL reduction of 55.9% assigned to “permanent, >50% disability” and subjective estimates used to score remaining categories (e.g. “minimal, recovery 1 month” was assigned a reduction in QOL of 2.3%). Based on the number of patients in each category a weighted average was obtained (4.6%) and then multiplied by value of statistical life of $4.4 million to obtain the average loss per negligent disability.

·        Indirect Benefits: Morbidity Gains from Deterrence of Negligence.

·        Social Welfare Losses: Efficiency Losses from Tax Collection.  To account for the efficiency losses associated with raising taxes to pay for government regulatory costs, we multiply the latter times the marginal cost of income tax collections (see Table B-1 for how these costs are calculated).

The overall expected cost of the medical tort system in 2002 is $113.7 billion (40.7, 195.4) while the expected benefits are $33.0 billion (8.4, 295.8). 


References

Final Ref List in Pro Cite CHSRMTS1

Table 1. Cost of Medical Tort System (millions of 2002 dollars)