physician demographics and the risk of medical malpractice

6
I t Physician Demographics and the Risk of Medical Malpractice MARK I. TARAGIN, M.D., M.P.H., New Brunswick, New Jersey, ADAM P. WILCZEK, Lawrenceville, New Jersey, M. ELIZABETH KARNS, M.P.H., RICHARD TROUT, Ph.D., JEFFREY L. CARSON, M.D., New Brunswick, New Jersey PURPOSE: This study was undertaken to clarify which, if any, physiclnn demographic cheracter- istics are associated with an increased rate of medical malpractice clRims. METHODS:We 9nglyzed the malpractice experi- ence of 9250 physician~ insured for at least 2 years from 1977 to 1987 in the state of New Jer- sey. After adjusting for years at risk, physician clnims per year was categorized into low, medi- um, and high. RESULTS: M a l e physicians were three times as likely to be in the high-claims group as female physicians, even after adjusting for other demo- graphic variables (relative risk, 3.1; 99% confi- dence interval, 22 to 4.4). Specialty was strongly associated with claims rate, with neurosurgery, orthopedics, and obstetrics/gynecology having 7 to 12 times the number of claims per year as psy- chiatry, the specialty with the fewest claimR. The rate of ciAim~ varied with age (p <0.001) and peaked at approximately age 40. No associa- tion was evident between claims rate and a phy- sician's site of training or type of degree. CONCLUSION: M a l e physicians are three times as likely to be in a high-claims category as fe- male physicians. We suspect that the most likely explanation for this finding is that women inter- act more effectively with patients. Understand- ing the reasons for the variation in claim rates between physicians may lead to the development of methods to reduce the overall rate of mal- practice claims. From the Division of General Internal Medicine (MIT, MEK, JLC), De- partment of Medicine, University of Medicine and Dentistry of New Jer- sey (UMDNJ), Robert Wood Johnson Medic31 School, New Brunswick, New Jersey; New Jersey !nterinsurance Exchange (APW), Lawrenceville, New Jersey; and the Department of Statistics (RT), Rutgers University, New Brunswick, New Jersey. This work was supported in part by a general research grant from the UMDNJ-RobertWoodJohnsonMedicalSchool, New Brunswick, NewJer- sey, and was presentedin part at the 13th AnnualMeetingfor the Society of General Internal Medicine,Arlington, Virginia, May 4, 1990. Requests for reprints should be addressedto Mark I. Taragin, M.D., M.P.H., UMDNJ-RobertWood Johnson Medical School, Divisionof Gen- eral Internal Medicine, 97 Paterson Street CN-19, New Brunswick, New Jersey, 08903-0019. Manuscript submitted October 3, 1991. and accepted in revisedform May 18, ]992. M alpractice is a serious problem for the United States health care system. From 1975 to 1984, the number of claims per physician has risen ap- proximately 10% a year, while the associated cost has increased at twice the rate of the Consumer Price Index [1]. The risk of malpractice has led to defensive medicine, with an estimated annual cost of 12 to 40 billion dollars [2,3]. The reasons why physicians are sued are not al- ways clear. Physicians are usually not sued when malpractice occurs and are often sued in circum- stances when they are not at fault [4-8]. This sug- gests that filing a claim against a physician is moti- vated by more than physician error and may be related to physician characteristics. Research aimed at identifying the characteristics of physicians who are frequently sued has yielded conflicting results [9-13]. Physician specialties are clearly associated with different rates of malprac- tice [14]. However, other physician characteristics including age; previous educational successes, con- tinuing medical education efforts, board certifica- tion, and site of medical school training; medical degree versus osteopathic degree; type of practice; and measures of competency have all yielded con- tradictory conclusions [9,10,12,15,16]. Unfortu- nately, many of these studies examined only one variable at a time and did not adjust for specialty. In this study, we examined the influence of physi- cian demographic variables on the risk of having a claim filed by analyzing 10 years of data on 9,250 physicians from the State of New Jersey. Our re- sults suggest that men are three times as likely as women to be in a high-claims category. METHODS Data Source We obtained the data from the New Jersey Inter- Insurance Exchange, a physician-owned insurance company, that insures approximately 60% of the physicians in the state of New Jersey. Since 1977, data have been entered into a standardized com- puter database composed of physician demographic information and detailed descriptive information on each claim. Study Design and Population We performed a retrospective cohort study in- volving 11,720 physicians who were insured from November 1992 The American Journal of Medicine Volume 93 537

Upload: mark-i-taragin

Post on 19-Oct-2016

214 views

Category:

Documents


2 download

TRANSCRIPT

I t Physician Demographics and the Risk of Medical Malpractice MARK I. TARAGIN, M.D., M.P.H., New Brunswick, New Jersey, ADAM P. WILCZEK,

Lawrenceville, New Jersey, M. ELIZABETH KARNS, M.P.H., RICHARD TROUT, Ph.D.,

JEFFREY L. CARSON, M.D., New Brunswick, New Jersey

PURPOSE: This study was undertaken to clarify which, if any, physiclnn demographic cheracter- istics are associated with an increased rate of medical malpractice clRims.

METHODS: We 9nglyzed the malpractice experi- ence of 9250 physician~ insured for at least 2 years from 1977 to 1987 in the state of New Jer- sey. After adjusting for years at risk, physician clnims per year was categorized into low, medi- um, and high.

RESULTS: Male physicians were three times as likely to be in the high-claims group as female physicians, even af ter adjusting for other demo- graphic variables (relative risk, 3.1; 99% confi- dence interval, 22 to 4.4). Specialty was strongly associated with claims rate, with neurosurgery, orthopedics, and obstetrics/gynecology having 7 to 12 times the number of claims per year as psy- chiatry, the specialty with the fewest claimR. The rate of ciAim~ varied with age (p <0.001) and peaked at approximately age 40. No associa- tion was evident between claims rate and a phy- sician's site of training or type of degree.

CONCLUSION: Male physicians are three times as likely to be in a high-claims category as fe- male physicians. We suspect that the most likely explanation for this finding is that women inter- act more effectively with patients. Understand- ing the reasons for the variation in claim rates between physicians may lead to the development of methods to reduce the overall rate of mal- practice claims.

From the Division of General Internal Medicine (MIT, MEK, JLC), De- partment of Medicine, University of Medicine and Dentistry of New Jer- sey (UMDNJ), Robert Wood Johnson Medic31 School, New Brunswick, New Jersey; New Jersey !nterinsurance Exchange (APW), Lawrenceville, New Jersey; and the Department of Statistics (RT), Rutgers University, New Brunswick, New Jersey.

This work was supported in part by a general research grant from the UMDNJ-Robert Wood Johnson Medical School, New Brunswick, New Jer- sey, and was presented in part at the 13th Annual Meeting for the Society of General Internal Medicine, Arlington, Virginia, May 4, 1990.

Requests for reprints should be addressed to Mark I. Taragin, M.D., M.P.H., UMDNJ-Robert Wood Johnson Medical School, Division of Gen- eral Internal Medicine, 97 Paterson Street CN-19, New Brunswick, New Jersey, 08903-0019.

Manuscript submitted October 3, 1991. and accepted in revised form May 18, ]992.

M alpractice is a serious problem for the United States health care system. From 1975 to 1984,

the number of claims per physician has risen ap- proximately 10% a year, while the associated cost has increased at twice the rate of the Consumer Price Index [1]. The risk of malpractice has led to defensive medicine, with an estimated annual cost of 12 to 40 billion dollars [2,3].

The reasons why physicians are sued are not al- ways clear. Physicians are usually not sued when malpractice occurs and are often sued in circum- stances when they are not at fault [4-8]. This sug- gests that filing a claim against a physician is moti- vated by more than physician error and may be related to physician characteristics.

Research aimed at identifying the characteristics of physicians who are frequently sued has yielded conflicting results [9-13]. Physician specialties are clearly associated with different rates of malprac- tice [14]. However, other physician characteristics including age; previous educational successes, con- tinuing medical education efforts, board certifica- tion, and site of medical school training; medical degree versus osteopathic degree; type of practice; and measures of competency have all yielded con- tradictory conclusions [9,10,12,15,16]. Unfortu- nately, many of these studies examined only one variable at a time and did not adjust for specialty.

In this study, we examined the influence of physi- cian demographic variables on the risk of having a claim filed by analyzing 10 years of data on 9,250 physicians from the State of New Jersey. Our re- sults suggest that men are three times as likely as women to be in a high-claims category.

METHODS Data Source

We obtained the data from the New Jersey Inter- Insurance Exchange, a physician-owned insurance company, that insures approximately 60% of the physicians in the state of New Jersey. Since 1977, data have been entered into a standardized com- puter database composed of physician demographic information and detailed descriptive information on each claim.

Study Design and Population We performed a retrospective cohort study in-

volving 11,720 physicians who were insured from

November 1992 The American Journal of Medicine Volume 93 537

DEMOGRAPHICS AND THE RISK OF MALPRACTICE / TARAGIN ET AL

TABLE I

Characteristics of Physician Population

Category Factor Total (%)* Low (%)1- Moderate (%)t High (%)~

Age (y) <40 3,889 (42)2,412 (62) 754 (19) 723 (19) 41-50 2,224 (24) 1,008 (45) 635 (29) 581 (26) 51-60 1,679 (18) 868 (52) 437 (26) 374 (22) >60 1,458(16)1,040(71) 227(16) 191 (13)

Gender Male 8,322 (90)4,611 (55) 1,916 (23} 1,795 (22) Female 927 (10) 716 (77) 137 (15) 74 (8)

Title Medical degree 8,681 (94)4,967 (57) 1,951 (22) 1,763 (20) Osteopathic degree 567 (6) 360 (63) 101 (18) 106 (19)

Board certification Yes 5,489 (59)2,789 (51) 1,398 (25) 1,302 (24) No 3,761 (41)2,539 (68) 655 (17) 567 (15)

Medical school U.S. 6,023 (65)3,378 (56) 1,341 (22) 1,304 (22) Foreign 3,227 (35)1,950 (60) 712 (22) 565 (18)

* Percent of strata in factor. tpercent of group in strata.

1977 to 1987. Since a brief physician experience may inaccurately reflect risk, we excluded 2,470 (21%) physicians with less than 2 years of observa- tion. The 9,250 physicians in our study had a mean age of 45.8, whereas the original cohort had a mean age of 43.8. The remaining demographic character- istics did not differ.

Study Variables The study outcome was physician claims per

year. A claim is defined as a formal notification that the patient believes a physician's action was mal- practice. A malpractice case is distinguished from a malpractice claim in that a case involves one patient but may include claims against more than one physician.

Claims per year was calculated by dividing the number of claims by the length of insurance cover- age. After adjusting for years at risk, we created three categories of claims frequency based upon the distribution of claims. The low-claims category in- cluded the 60% of physicians with the lowest rate of claims (less than 0.16 claims per year), the medium- claims category included the 20% of physicians with an intermediate claim rate (0.16 to 0.32 claims per year), and the high-claims group included the 20% of physicians with the highest claim rate (greater than 0.32 claims per year).

We also assessed the following physician vari- ables: age, gender, title (medical degree or osteo- pathic degree), board certification status, location of medical school attended (United States or for- eign), medical specialty, part-time versus full-time

practice, and self-reported number of patients seen per week. Medical specialty was collapsed into 25 groups. When an interaction between specialty and other variables was assessed, we further collapsed medical specialty into three categories (medical, surgical, and other).

Statistical Analysis Analyses were stratified by specialty and summa-

rized using the Mantel-Haenszel procedure [17]. Relative risk and 99% confidence intervals were cal- culated for each independent variable using the low-claims category as the reference. We then used logistic regression to simultaneously control for all the physician variables and calculated the relative risk with 99% confidence intervals [18]. The depen- dent variable was claims per year (one model used the medium-claims category, and the second model used the high-claims category), and the indepen- dent variables were specialty, age, gender, title, site of training, and board certification status. Logistic regression produces an adjusted odds ratio that is a good approximation of relative risk when the preva- lence of the outcome is less than 10%, as it is in our study. Due to our large sample size, we required a p value of 0.01 for statistical significance.

Three additional analyses were performed to as- sess other potential confounders. First, we repeated the analysis after excluding cases in which more than one physician was named. Second, we exclud- ed physicians practicing part time. Third, we added the number of patients seen per week in the logistic regression model.

RESULTS

The characteristics of the study population are described in Table I. The mean age of the physi- cians was 46, 90% were male, 94% had medical de- grees, 59% were board certified, and 65% were U.S. medical school graduates. The physicians were in- sured for a median of 8 years, with 27% insured between 2 and 5 years and 73% for greater than 5 years.

There were 10,151 malpractice cases filed during the study period. Since more than one physician can be named in a case, this resulted in 15,153 claims. There were four or more physicians named in 5% of the cases, two to three physicians named in 24% of the cases, and only one physician named in 71% of the cases. Of the 9,250 physicians studied, 46% had no claims filed against them, 38% had one to three claims, and 16% had more than three claims.

The claims history for selected specialty groups is presented in Table II. In general, the surgical sub- specialties had the highest rate of claims per year. Psychiatry was the specialty with the lowest claim

538 November 1992 The American Journal of Medicine Volume 93

DEMOGRAPHICS AND THE RISK OF MALPRACTICE / TARAGIN ET AL

TABLE II

Claims History for Selected Specialties

No. of Claims/ Relative Risk Phy- Total Physician/ (99% Confidence

Specialty Group sicians Claims Year Interval)

Neurosurgery 64 352 0.65 12.5 (9.8-]6.1) Orthopedics 425 1,947 0.53 10.3 (8.3-12.7) Ob-Gyn 594 1,824 0.40 7.7 (6.2-9.5) General Surgery 499 1,363 0.30 5.7 (4.6-7.2) Urology 213 520 0.28 5.4 (4.2-6.8) HEENT 152 311 0.24 4.7 (3.6-6.0) Radiology 501 945 0.23 4.4 (3.5-5.5) Anesthesiology 575 914 0.21 4.0 (3.2-5.1) Emergency Medicine 21 ] 236 O. ] 7 3.3 (2.5-4.3) Ophthalmology 428 543 0.14 2.8 (2.2-3.5) Internal Medicine 1,805 1,880 0.13 2.5 (2.0-3.1) Dermatology 202 196 0.10 2.0 (1.5-2.7) Pediatrics 697 503 0.08 1.6 (1.3-2.1) Pathology 208 104 0.07 1.3 (0.9-1.8) Psychiatry 151 45 0.05 Reference All physicians 9,250 14,538 0.19 - -

Ob-Gyn = Obstetrics and Gynecology; HEENT = Head, Eyes, Ears, Nose, and Throat.

rate and was used as the reference to calculate rela- tive risks for the remaining specialties. Neurosur- gery, orthopedics, and obstetrics/gynecology had 7 to 12 times the number of claims per year as psychi- atry. Internal medicine, the specialty with the most physicians, had 2.5 times the number of claims per year as psychiatry.

Figure 1 shows the relationship between the rate of claims per year and age. Age was associated with the claims rate per year (p <0.001). A physician's risk of having a claim filed against him/her was greatest between the ages of 36 and 55, peaking around 40.

Table I further characterizes the distribution of the demographic variables by claims category. Of the male physicians, 22% were in the high-claims category. In contrast, only 8% of female physicians were in the high-claims category. Controlling for physician specialty only, men were 2.2 (99% confi- dence interval, 1.8 to 2.8) and 1.7 (99% confidence interval, 1.4 to 2.1) times more likely than women to be in the high- and moderate-claims groups, respec- tively (Table III). When the analysis adjusted for all the demographic factors using logistic regres- sion, the relative risk for men compared with wom- en increased to 3.1 (99% confidence interval, 2.2 to 4.4) and 2.0 (99% confidence interval, 1.6 to 2.6) times for the high- and moderate-claims groups, respectively.

The association between gender and claims cate- gory stratified by specialty is displayed in Table IV. In general, the pattern is consistent among spe- cialties in showing an elevated risk in men. Special- ties not displayed had insufficient women to calcu- late relative risks.

Table III also shows that neither site of training (U.S. versus foreign) nor medical degree versus os-

C L A I M S / P H Y S I C I A N YEAR 0 .25

0.2

0 . 1 5

0.1

0 . 0 5

November 1992

I

- 35 3 6 - 4 0 41-45 4 6 - 5 0 51-55 5 6 - 6 0 61-65 65 -

A G E GROUPS

Figure 1. Claims rate by age.

TABLE III

Risk of a Medical Malpractice Claim for Select Physician Variables

Logistic Regression Specialty-Adjusted Relative

Claims Risk Relative Risk *t Risk Group (99% CI)* (99% CI) *§

Gender (male versus female)

Medium 1.7 (1.4-2.1) 2.0 (1.6-2.6) High 2.2 (1.8-2.8) 3.1 (2.2-4.4)

Degree (D.O. versus M.D.)

Medium 1.0 (0.8-1.3) 0.8 (0.6-1.2) High 1.4 (1.1-] .7) 1.3 (0.9-1.8)

Site of training (foreign versus U.S.)

Medium 1.0 (0.9-1.2) 1.0 (0.8-1.1) High 1.2 (1.1-1.3) 0.9 (0.7-1.0)

Board certification status (certified versus not certified)

Medium 1.5 (1'14-]7)/ 1.4 (1.1-2.0) High II

CI - confidence interval; D.O. = osteopathic degree; M.D. = medical degree. *Compared with low category and adjusted for specialty only. tValues listed satisfied Breslow's test of homogeneity. tlf confidence interval includes 1.0, then results are not significant at p < O.Ol. §Logistic regression model used claims per year, either medium or high, compared with the low category as the dependent variable. Age, gender, title, site of training, board certification status, specialty, and specialty-board certification interact on a e the ndependent variables, IIBoard certification interacted with specialty in the high-claims group. Logistic regression relative risk 1.7 (1.1-2.6) for surgery specialties, 1.2 (0.8-1.9) for medical specialties, compared with other specialties. See text.

teopathic degree was statistically significant. No as- sociation was evident between board certification status in the medium-claims analysis after adjust-

The American Journal of Medicine Volume 93 539

DEMOGRAPHICS AND THE RISK OF MALPRACTICE / TARAGIN ET AL

TABLE IV

Risk of a Medical Malpractice Claim for Select Specialties by Gender*+

Category Total Low Medium High Relative Risk (95% Confidence Interval) M (%) M (%) M (%) M (%) Medium High to

Specialty F (%) F (%) F (%) F (%) to Low Low

internal 1,660 (92.0) 1,069 (64.4) 396 (23.9) 195 (11.7) 2.2 (1.1-3.5) 4.0 (1.7-9.4) Medicine 145 (8.0) 123 (84.8) 17 (11.7) 5 (3.4)

Pediatrics 523 (75.0) 387 (74.0) 109 (20.8) 27 (5.2) 2.2 (1.4-3.6) 3.4 (1.]-] 1.1) 174 (25.0) 154 (88.5) 17 (9.8) 3 (1.7)

Obstetrics and 504 (84.8) 140 (27.8) 104 (20.6) 260 (51.6) 1.0 (0.7-1.4) 1.7 (1.2-2.3) Gynecology 90 (]5.2) 38 (42.2) 28 (31.1) 24 (26.7)

Anesthesiology 472 (82.2) 212 (44.9) 123 (26.1) 137 (29.0) 1.2 (0.9-1.7) 2.3 (].4-3.8) 102 (17.8) 62 (60.8) 27 (26.5) 13 (12.7)

Radiology 454 (90.6) 174 (38.3) 155 (34.1) 125 (27,5) 1.5 (0.9-2.3) 3.5 (1.4-8.7) 47 (9.4) 29 (61.7) 14 (29.8) 4 (8.5)

General 488 (97.8) ]48 (30.3) 155 (31.8) 185 (37.9) 1.4 (0.6-3.4) 1.2 (0.5-3.0) Surgery 11 (2,2) 5 (45.4) 3 (27.3) 3 (27.3)

Emergency 190 (90.0) ] 16 (61.1) 37 (19.5) 37 (19.5) 1,3 (0.5-3.7) 0.9 (0.4-1.9) Medicine 21 (10.0) 13 (61.9) 3 (14.3) 5 (23.8)

Pathology 169 (81.6) 136 (80.5) 29 (17.2) 5 (3.0) 3.3 (0.8-13.0) ].3 (0.2-10.6) 38 (18.4) 35 (92.1) 2 (5.3) 1 (2.6)

Dermatology 171 (84.7) 1 ]4 (66.7) 40 (23.4) 17 (9.9) 3.9 (1.0-15.3) 3.8 (0.5-27.2) 31 (]5.3) 28 (90.3) 2 (6.5) 1 (3.2)

*Ranked by number of physicians per specialty. tSpecialties with insufficient females to calculate relative risks: HEENT (Head, Eyes, Ears, Nose, and Throat), Neurosurgery, Orthopedics, Psychiatry, and Urology.

ing for specialty alone. However, an interaction be- tween board certification and specialty was evident in the high-claims analysis (Mantel-Haenszel test for heterogeneity, p <0.01). Since the logistic re- gression model was not stable with 25 specialty in- teraction terms, we collapsed the specialty groups into medicine, surgery, and other. Only the interac- tion between board certification status and surgical specialties was significant (relative risk 1.7 [99% confidence interval, 1.1 to 2.6]).

We then performed three subanalyses, and the gender effect remained stable. First, we restricted the analysis to the 7,209 malpractice cases in which only 1 physician was named. Men were 2.2 (99% confidence interval, 1.3 to 3.8) times more likely to be in the high-risk group. Second, we restricted the analysis to full-time physicians (89% of the study population), and the relative risk for men was 2.9 (99% confidence interval, 2.0 to 4.2).

Finally, we controlled for the self-reported vol- ume of patients seen per week for the 81% of physi- cians who provided this information. The median number of patients seen by male physicians (60) was significantly greater than that seen by female physicians (50) (p <0.001). When this variable was added to the logistic regression model, the relative risk for men remained 3.0 (99% confidence interval, 2.0 to 4.5). Also, the interaction between board cer-

tification and specialty, and board certification it- self were no longer significant.

COMMENTS Our study was undertaken to clarify which, if any,

physician demographic characteristics are associ- ated with an increased rate of medical malpractice claims. We found that male physicians were three times as likely to be in the high-claims group as female physicians. Other physician characteristics such as specialty and age were also associated with a physician's rate of claims. There was no association between the claims rate and the country the physi- cian trained in or what type of degree the physician received from medical school. These findings per- sisted after simultaneously controlling for special- ty, age, gender, certification status, country of training, type of degree, self-reported volume of pa- tients, and number of years at risk.

These results directly quantify, for the first time, the influence of gender on malpractice experience. Sloan et al [19] found that the predicted probability (derived from a regression model) of being in a high- payment group was greater for male than female physicians. However, this was significant in only one of three specialty groups.

There are several possible explanations for the lower rate of claims against women. It is unlikely

540 November 1992 The American Journal of Medicine Volume 93

DEMOGRAPHICS AND THE RISK OF MALPRACTICE / TARAGIN ET AL

that female physicians have better medical knowl- edge. Day et al [20] found that directors of internal medicine programs rated men higher than women in medical knowledge and procedural skills. Fur- thermore, the pass rates for the internal medicine certifying examination over 4 years was higher for men, 85% to 86%, than for women, 79% to 81%.

The effect of gender on the rate of claims might be explained by male physicians caring for more and/or sicker patients than female physicians. In the National Ambulatory Care Survey, female phy- sicians saw approximately 15% fewer patients than their male counterparts [21]. Our study also found that women reported caring for fewer patients. However, when this effect was controlled for, the results did not change. Although it is possible that the number of patients that a physician reports to see per week is not accurate, it is unlikely that physician estimates would differ systematically by gender. There is no information available on wheth- er women care for patients who are less ill, although physician specialty should partially adjust for this factor. Furthermore, it is unlikely that the large differences we observed reflect a variation within specialty with regard to the number or risk of proce- dures performed or the severity of illness in the patients cared for.

A number of other explanations exist for the gen- der differences we observed. Women may be more risk averse, less aggressive, or more meticulous than men. It is also possible that more female physicians practice in an academic setting, although this has not been documented to affect malpractice rates. Also, patients may have different expectations when seeing a female physician or be less likely to sue a woman.

We suspect that one important factor contribut- ing to our findings is that women interact more effectively with patients. Day et al [20] observed that internal medicine program directors rated women higher in humanistic attributes. L inne t al [21,22] found that patients who had seen a female resident had significantly higher art-of-care and to- tell satisfaction scores than patients who had seen a male resident. Furthermore, women spend more time per patient visit than men do. It is probable that patients are less likely to sue a physician with whom they have a satisfactory personal relation- ship. Clarifying the reason(s) for the effect of physi- cian gender on the likelihood of a malpractice claim may identify interventions that could be used to reduce the frequency of malpractice claims.

We also found that the rate of claims varied with age. Similar to other work [23], the rate of claims peaked at approximately age 40. The reasons for this finding are difficult to determine. Age is a corn-

plex variable that might r~flect many factors. Dur- ing the early years of practice, a lower claim rate may result from seeing fewer patients or from being current with recent developments in medical care. However, physicians are also less experienced and, perhaps, more prone to make mistakes. In later years, physicians typically see a larger number of patients. This could result in shorter patient visits, adversely affecting the patient-physician relation- ship and perhaps even the quality of care. A busy practice may also leave the physicians with less time to read or to attend continuing medical educa- tion programs. On the other hand, physicians are more experienced and are still reasonably close to their training. As physicians age further, they may benefit from their many years of experience, reduce the number of patients seen, and have a long-stand- ing relationship with those they care for. Alterna- tively, it may be more difficult to remain current.

Other studies support the notion that there is little difference between foreign graduates and U.S. graduates in malpractice claim rates or quality of care. Two studies found the frequency of malprac- tice claims to be independent of the site of training [9,24]. Other studies have also found that foreign medical graduates provide quality of care equal to that of U.S. graduates [25,26].

We also found no difference between physicians with an osteopathic degree and those with a medical degree. We are not aware of any other studies that have examined the association between the type of degree and the rate of medical malpractice claims.

The effect of board certification is more difficult to interpret. Board-certified physicians had a slightly elevated risk when the medium-claims group was cdthpared with the low-claims group. An interaction was present in the specialty-adjusted analysis when comparing the high-claims group with the low-claims group. In the main analyses, the surgical specialties had a higher risk than the other categories of specialty. However, this was no longer significant after adjusting for number of patients seen per week. This suggests that board-certified physicians (especially surgeons) may see more pa- tients, which would place them at greater risk for malpractice claims.

We suspect that our results can be generalized, even though the study was performed with a subset of physicians from one state. The demographic vari- ables of the physicians included in this study are similar to those of the overall population of physi- cians in New Jersey and vary only slightly from national figures [27]. Similar to other northeast states, there are fewer general practitioners, slightly more board-certified physicians, and a higher per- centage of foreign medical graduates than the na-

November 1992 The American Journal of Medicine Volume 93 541

DEMOGRAPHICS AND THE RISK OF MALPRACTICE / TARAGIN ET AL

tional average [23,24,27]. Also, as in other studies, greater than 70% of claims involved only one physi- cian [24].

In summary, our results suggest that male physi- cians are three times as likely to be in a high-claims category as female physicians. Furthermore, we have confirmed that claim rates vary dramatically among specialties and that physician age is a signifi- cant factor. Understanding the basis of these effects might delineate differences in how health care is provided, reveal what patients expect from their physicians, and elucidate the important compo- nents of a patient-physician relationship. Clarifica- tion of these issues might allow for the development of effective strategies to reduce the malpractice claim rates for all physicians.

ACKNOWLEDGMENT We thank Louise Russell, Ph.D., for review of the manuscript; Amy E, Duff and Mildred Evans for technical assistance; and Pat Cerro and Dawn Kaly for manu- script preparation.

REFERENCES I. Danzon PM. New evidence on the frequency and severity of medical malprac- tice claims. Santa Monica, CA: Rand Corp., 1986; R-3410-ICJ. 2. Study of professional liability costs. In: AMA House of Delegates Proceedings, Dec 4-7, Board of Trustees Report N(I-83). Chicago: American Medical Associa- tion, 1983: 93-102. 3. Reynolds RA, Rizzo JA, Gonzalez ML. The cost of medical professional liabUity. JAMA 1987; 257: 2776-81. 4. Steel K, Gertman P, Crescenzi C, Anderson J. latrogenic illness on a general medical service at a university hospital. N Engl J Med 1981; 304: 638-42. 5. Pocincki LS, Dogger S J, Schwartz BP, The incidence of iatrogenic injuries. Appendix, Report of the Secretary's Commission on Medical Malpractice. Wash- ington, DC: U.S. Government Printing Office, DHEW Publication No. [OS] 1973: 73-89. 6. Iglehart JK. The professional liability crisis. N Engl J Med 1986: 315:1105-8. 7. Harness JK. A closer look at 1137 liability cases closed. Mich Med 1987; 86: 522-4. 8. Medical malpractice: insurance costs increased but varied among physicians

and hospitals. Washington, DC: U.S. General Accounting Office, 1986; HRD-86- 112. 9. Brook RH, Brutco RL, Williams KN. The relationship between medical mal- practice and quality of care. Santa Monica, CA: Rand Corp., 1975; P-5526. 10. Ferber S, Sheridan B. Six cherished malpractice beliefs put to rest. Medical Economics 1975; 52: 150-6. 11. Phelps CE. Experience rating in medical malpractice insurance. Santa Moni- ca, CA: Rand Corp., 1978; P-5877-1. 12. Adams EK, Zuckerman S. Variation in the growth and incidence of medical malpractice claims. J Health Polit Policy Law 1984; 9: 475-88. 13. Stoll A. The malpractice crisis and bad doctors. J Fla Med Assoc 1987; 74: 663-4. 14. Zuckerman S. Medical malpractice: claims, legal costs, and the practice of defensive medicine. Health Aft (Millwood) 1984; 3: 128-33. 15. Peterson OL, Andrews LP, Spain RS, et al. An analytic study of North Caro- lina general practice 1953-1954. Journal of Medical Education 1956; 31: 1-165. 16. Brook RH, Williams KN. Evaluation of the New Mexico peer review system, 1971 to 1973. Santa Monica, CA: Rand Corp., 1977; R-2110-HEW/RC. 17. Mantel N, Haenszel W. Statistical aspects of the analyses of data from retrospective studies of disease. J Natl Cancer Inst 1959; 22: 719-48. 18. Kleinbaum DG, Kupper LL. Applied regression and other multivariate meth- ods. Boston: Duxbury Press, 1978. 19. Sloan FA, Mergenhagen PM, Burfield WB, Bovbjerg RR, Hassan M. Medical malpractice experience of physicians, predictable or haphazard? JAMA 1989; 262: 3291-7. 20. Day SC, Norcini J J, Shea JA, Benson JA. Gender differences in the clinical competence of residents in internal medicine. J Gen Intern Med 1989; 4: 309-12. 21. Characteristics of visits to female and male physicians. The National Ambu- latory Medical Care Survey, 1977. June 1980: DHHS publication no. 80-1710. 22. Linn LS, Cope DW, Leake B. The effec t of gender and training of residents on satisfaction ratings by patients. Journal of Medical Education 1984; 59: 964-6. 23. Schwartz WB, Mendelson DN. Physicians who have lost their malpractice insurance: their demographic characteristics and the surplus in companies that insure them. JAMA 1989; 262: 1335-41. 24. U.S. General Accounting Office. Medical malpractice: characteristics of claims closed in 1984. Washington, DC: U.S. General Accounting Office, 1987; HRD-87-55. 25. Rhee S, Lyons TF, Payne BC, Moskowitz SE. USMG's versus FMG's: are there performance differences in the ambulatory care setting? Med Care 1986; 24: 248-58. 26. Saywell RM, Studnicki J, Bean JA, Ludke RL. A performance comparison: USMG-FMG attending physicians. Am J Public Health 1979; 69: 57-62. 27. Department of Data Release Services, Division of Survey and Data Re- sources. Physician characteristics and distribution. 1983 ed. Chicago: American Medical Association, 1984.

542 November 1992 The American Journal of Medicine Volume 93