defense of breast cancer malpractice claims · clinical and risk management factors that could help...

© 2001 Blackwell Science Inc., 1075-122X/01/$15.00/0The Breast Journal, Volume 7, Number 2, 2001 76–90

Defense of Breast Cancer

Malpractice Claims

Samuel Zylstra, MD, MPH, FACOG,* Carl J. D’Orsi, MD,* Barbara A. Ricci,

†

Elizabeth E. Halloran,

†

Laurence J. Resseguie, PhD,

†

Linda Greenwald, RN, MS,

†

and Maureen C. Mondor, ARM

†

*

University of Massachusetts Medical School, Worcester, Massachusetts,

and

†

ProMutual, Boston, Massachusetts

fense model predicts that the probability of successful de-fense is lessened with inadequate record keeping, a patientthat has metastasis and is alive, and a delay in diagnosis of 12months or more. The overall indemnity model predicts ahigher indemnity with the spread of disease at the time ofevaluation, a patient who has metastasis and is alive, and adate of occurence closer to the present. Indemnity is less inpatients who have had a lymph node dissection, who havedied, or who are alive without metastasis. The WHP modelpredicts an increased overall indemnity with the spread of dis-ease at the time of evaluation and the presence of a masswithout pain. Indemnity decreases with a history of preg-nancy, absence of presenting symptoms, or presentation withpain with or without a mass, and the performance of a lymph

node dissection.

j

Key Words:

breast cancer, data tracking, litigation, mal-practice, medicolegal, risk management, risk prevention

T

he risk management of breast cancer is in manyways a reflection of the clinical approach of the

medical community to the early diagnosis and treatmentof breast cancer. The current U.S. health care system al-lows major disparities in quality. In the absence of uni-form guidelines for excellence, the system often becomesa frustrating maze for the people it is intended to serve.To conduct a clinical breast examination (CBE) in onesetting, a mammographic or sonographic evaluation inanother, and a fine needle aspiration (FNA), core, orsurgical biopsy somewhere else is not ideal (1).

Early breast cancer is often very difficult to diagnose,and the correct diagnosis is usually the result of multiple

complementary examinations. Each examination reveals

j

Abstract:

The goal of this study was to determinewhether factors associated with the successful defense andcost of malpractice cases involving the failure to diagnosebreast cancer could be identified in medical and legal records.Secondary goals were to develop a multidisciplinary clinicalalgorithm utilizing National Comprehensive Cancer Network(NCCN) practice guidelines with practitioner risk managementstrategies. Physician deviations from these guidelines weretracked to identify high-risk areas in the diagnosis of breastcancer. A multidisciplinary clinical algorithm was introducedand practitioner risk management issues were addressed. Inthis study specific medical, legal, and cost factors were retro-spectively abstracted and analyzed to identify associations be-tween medical and legal factors and medicolegal outcome.ProMutual handled 156 malpractice cases involving breastcancer between January 22, 1986, and November 20, 1997. Ofthe total, 124 cases involving 212 defendants were closed. Theclosed cases were analyzed, using multivariable stepwise lo-gistic and linear regression, to identify associations betweenclinical factors and case outcome. Women’s health practitio-ners (WHPs), including obstetrician-gynecologists (OB-GYNs),family medicine, and internal medicine clinicians, were thelargest group of defendants (97). Others included radiologists(43), surgeons (33), and pathologists (3). OB-GYNs accountedfor 31% of these defendants, with a cost of more than $16million. The greatest number of specialists represented in theopen cases were radiologists, with 38% of the total. The de-

Address correspondence and reprint requests to: Samuel Zylstra, MD,MPH, FACOG, Department of Obstetrics and Gynecology, University ofMassachusetts Medical School, Whitinsville Medical Center, 18 Granite St.,Whitinsville, MA 01588, U.S.A.

Defense of Breast Cancer Malpractice Claims

•

77

different characteristics of breast tissue, benign or ma-lignant. A CBE evaluates symmetry, contour, texture,nodularities, masses, tenderness, and nipple discharge.Mammographic examination provides information aboutvariations in density. Ultrasound provides informationabout variations in tissue sound transmission. Cytologicevaluation reveals the microscopic appearance of thecells and other tissue components. There is no best or“most specialized” test. Each has its place and its limita-tions. The relative value of a test varies among patientsand situations.

Early diagnosis of breast cancer is especially difficultin young patients with dense breast tissue. Dense fibro-glandular tissue can obscure the diagnosis for both theevaluating clinician and the radiologist. A general policyof risk prevention should focuss on understandingwhich patients and physicians fall into a high-risk pro-file for diagnostic errors. Kern (2) identified a “triad oferrors” involving young patients with self-discoveredbreast masses and negative mammograms, who ac-counted for the majority of breast cancer cases. Studiesof such cases found that these patients, when diagnosedat stage II or higher, are more likely to file claims (3–7).

Breast cancer studies have shown that when litigationis pursued for alleged failure to diagnose, multiple spe-cialists are named in the suit. These specialists include,in descending order of frequency, women’s health prac-titioners (WHPs), surgeons, radiologists, and patholo-gists (8–11). WHPs include obstetrician-gynecologists(OB-GYNs) and family and internal medicine clinicians.A recent trend has been an increase in the number of law-suits against health maintenance organizations (HMOs)and insurance companies (12). Malpractice claims againstpathologists for misdiagnosis of breast cancer cases havebeen sharply rising, especially in the area of fine needleaspiration (FNA) with inadequate sampling (13,14). Law-suits against radiologists nearly doubled between 1988and 1992. This is due to the increasing frequency of in-terventional procedures radiologists are performing andtheir failure to obtain informed consent (15–17). Misdi-agnosed breast cancer is the leading malpractice claimwithin general surgery (2). WHPs, especially gynecolo-gists, are the most cited specialists in lawsuits, perhapsbecause patients frequently consider gynecologists to bethe primary physicians for evaluating breast disorders(6,8,11,18).

According to surveys conducted by the Physician In-surers Association of America (PIAA) in 1990 and 1995,breast cancer ranks number one in claims frequencyamong misdiagnosed conditions (2,8,18). The most

common reason given by the majority of the expert re-viewers for the delay in diagnosis in the PIAA study was,“Physical findings failed to impress the physician.”

The purpose of ProMutual’s study was to identifyclinical and risk management factors that could helpphysicians make an earlier diagnosis of breast cancer.Medical and legal records were used to identify factorsthat resulted in increased malpractice risk for the defen-dant physicians. Review by data abstractors of specificdeviations from routine National Comprehensive Can-cer Network (NCCN) practice guidelines by defendantsresulted in the creation of a multidisciplinary clinical al-gorithm among involved specialties (19). Multispecialtyrisk management issues are addressed.

MATERIALS AND METHODS

ProMutual provides professional liability coverage formore than 70% of all practitioners in Massachusetts. Atotal of 156 breast cancer cases were handled by ProMu-tual between January 22, 1986, and November 20, 1997.The 124 cases that closed during this time period were thematerial for this study. Use of the model to predict resultsfor years outside the range of data have two serious dan-gers: (a) Because the year will be far from the mean of thedata, small errors in the parameter estimates producelarge errors in the predicted indemnity; (b) The shape ofthe curve may change radically for later years and bebadly represented by the model. Demographic, medical,and legal information available for each case was ab-stracted. This review included a detailed history, physicalexamination, mammography, and pathology data. Plain-tiff and defense presentations were reviewed. To deter-mine the length of the delay in diagnosis we consideredboth the delays documented by objective information andthose speculated by the plaintiff and defense experts. To-tal defense costs are the total of indemnity payments plusexpenses. Study variables are found in Table 1.

Logistic regression and Fisher’s exact test were usedto test the independence of the successful defense of thesuit from clinical and legal factors. Variables thought tohave clinical or legal relevance or for which preliminaryunivariate analysis suggested a relationship between po-tential explanatory variables and the dependant variablewere retained for multivariable analysis. The model ofsuccessful defense of a suit was developed by use of theLOGISTIC procedure of the SAS statistical softwarepackage (SAS Institute, Cary, NC). The response vari-able is log odds

5

ln{

p

/(1 –

p

)}, where

p

is the probabil-ity of success. Stepwise selection was used to choose theexplanatory variables for the final model.

78

•

zylstra et al.

Least-squares regression was used to identify the rela-tionship between various clinical and legal factors andcase indemnity. Variables thought to have clinical or le-gal relevance or for which preliminary univariate analy-sis suggested a relationship between potential explana-tory variables and the dependant variable were retainedfor multivariable analysis. Models for continuous re-sponse variables were developed using the REG proce-dure of the SAS statistical software package. The caseindemnity was transformed to the natural log of indem-nity in order to minimize violation of the assumptions ofleast squares regression. The

R

2

method with the

C

(

P

)statistics of Mallows was used to make a preliminary se-lection of explanatory variables (20). Dates are definedas the number of years after January 1, 1986.

A multidisciplinary clinical algorithm was createdbased on ProMutual’s historical breast cancer claimsdata in accordance with the NCCN practice guidelines.Realizing that current trends in the evaluation of breastcancer have evolved, the algorithm was then modified toreflect current paradigms. Defendants’ deviations fromthese accepted guidelines were noted and compared tocase closure with and without indemnity.

RESULTS

The 156 malpractice cases alleging negligence in thediagnosis and/or treatment of breast cancer included124 closed cases with 212 defendants and 32 open caseswith 65 defendants. Many cases involved more than onedefendant from multiple medical specialties and corpo-

Table 1. Study Variables

Type Definition Coding

Indemnity Payment to plaintiff based on the outcome DollarsMedical

Agegroup Age of plaintiffs grouped in three categories

,

40

5

1/40–59

5

2/

.

59

5

3Records Medical records of plaintiff Adequate

5

1/inadequate

5

0Smoke Plaintiff is a smoker Yes

5

1/no

5

0Pregnant Plaintiff was pregnant at time of diagnosis Yes

5

1/no

5

0Pregnancy related Plaintiff recently was pregnant/breast feeding Yes

5

1/no

5

0Menopause Plaintiff had gone through menopause at time of diagnosis Yes

5

1/no

5

0Parous State Plaintiff has children Yes

5

1/no

5

0Screening mammogram A screening mammogram was done prior to the date of loss Yes

5

1/no

5

0CBE Clinical breast exam was documented Yes

5

1/no

5

0SBE Self breast exam was instructed and documented Yes

5

1/no

5

0Present Plaintiff presented with symptoms Yes

5

1/no

5

0P

0

(reference) No presenting symptoms Yes

5

1/no

5

0P

1

Pain with or without mass Yes

5

1/no

5

0P

2

Mass with no pain Yes

5

1/no

5

0Mass The presence of a mass on clinical breast exam Yes

5

1/no

5

0Biopsy A biopsy was performed Yes

5

1/no

5

0Spread Cancer spread from primary site at time of evaluation Yes

5

1/no

5

0Infiltrative/intraductal The presence of infiltrative/intraductal carcinoma Yes

5

1/no

5

0Lumpectomy/simple A lumpectomy or simple mastectomy was performed Yes

5

1/no

5

0MastectomyModified radical mastectomy A modified mastectomy was performed Yes

5

1/no

5

0Node dissection A node dissection was performed Yes

5

1/no

5

0Bone marrow transplant A bone marrow transplant was performed Yes

5

1/no

5

0Tamoxifen Plaintiff was treated with tamoxifen Yes

5

1/no

5

0Medicolegal

Date of loss Number of years from 1/1/60 to the date the case occurred YearsDate of notification Number of years from 1/1/60 to the date the case was reported to ProMutual YearsLossNot Number of years from the date of loss to the date of notification YearsDate of settlement Number of years from 1/1/60 to the date the case was settled YearsDelay

D

0

(reference) 0–5 months delay in diagnosis Yes

5

1/no

5

0D

1

6–11 months delay in diagnosis Yes

5

1/no

5

0D

2

Greater than or equal to twelve month delay in diagnosis Yes

5

1/no

5

0Successful defense Settlement without indemnity payment Win/loseCase expense Expenses incurred by ProMutual in the evaluation and defense of each case DollarsPatient outcome Patient status at closure of case

O

1

Alive without metastatic spread Yes

5

1/no

5

0O

2

Alive with metastatic spread Yes

5

1/no

5

0O

3

(reference) Death Yes

5

1/no

5

0

Defense of Breast Cancer Malpractice Claims

•

79

rate entities. WHPs were defendants in 46% of theclosed cases, followed by radiologists (20%), surgeons(16%), and pathologists (1%). The remaining 17% con-sisted of 5 hospitals, 1 physician assistant, and 22 cor-porations. Radiologists were defendants in 38% of theopen cases, followed by WHPs (25%), surgeons (14%),and other (23%). A lesion was initially discovered bythe patient as a breast self-examination abnormality in77% of cases, followed by screening mammogram(11%), CBE (8%), and other (4%). A screening mam-mogram was not done in 56% of cases and was done in41%. In 3% of cases it was impossible to determinewhether a mammogram was done. The radiologist hadnot communicated the mammography report to the re-ferring physician in 47% of cases. In 40% of cases it wasimpossible to determine if this communication hadtaken place. Communication between a radiologist andthe referring physician was documented in only 13% ofcases. Double reading of mammograms occured in 15%of cases. Additional characteristics of study claims arelisted in Table 2.

Of the 124 closed cases, 86 (69%) resulted in an in-demnity payment. A total of 82 cases were settled out ofcourt, with a mean indemnity of $422,644 and a meanexpense of $46,890. The highest mean indemnity($783,450) and expense ($99,126) resulted from fourjury verdicts for the plaintiff (Table 3). Twelve of the 124(10%) cases were pregnancy related and 9 of these re-sulted in the payment of indemnity. The mean indemnitypayment in pregnancy-related cases was $725,030 ver-sus $406,045 for non-pregnancy-related cases (Table 4).

The surgeons incurred the highest percentage of casesthat closed with indemnity (76%), followed by patholo-gists (67%), WHPs (55%), and radiologists (26%). Theaverage indemnity paid out by ProMutual was highestfor the WHPs. This was followed closely by surgeonsand corporations, for example, HMOs (Table 5).

Based on these data, three models were developed toexplain indemnity payment and the probability of suc-cessful defense of a case. Potential explanatory variablesconsidered for multivariable analysis in the developmentof the model predicting successful defense of a case areshown in Table 6. Settlement date, use of tamoxifen,and spread of disease were not included in the finalmodel because they were found not to have statisticalsignificance in multivariable analysis.

where

x

is the log odds of a successful defense.The model for a successful defense demonstrates that

inadequate record keeping is associated with a lower like-lihood of winning the case. The lowest chance of a suc-cessful defense is with a plaintiff who is alive

with

me-tastasis (

O

2

). A successful defense is more likely if theplaintiff has died (

O

3

). The highest chance of winning oc-curs when a patient is living and without metastasis (

O

1

).With respect to delay in diagnosis the lowest chance

of a successful defense occurs when the delay is 12months or more (

D

2

). A successful defense is more likelywhen the delay in diagnosis is 6 months or more, butless than 12 months (

D

1

). The highest chance of win-ning is when the delay in diagnosis is less than 6 months(

D

0

). Overall this demonstrates that the longer the delaythe lower the probability of a successful defense.

The potential explanatory variables considered formultivariable analysis in the development of the modelfor overall indemnity paid by ProMutual in a lost case isshown in Table 7. The explanatory variables were iden-tified by least-squares regression performed on the 86cases. The variables bone marrow transplant, smoke,

x 21.4101 1.3264 records 0.6410O1 .8510O2 0.5387D1 1.5054D2,––

–+ +=

Table 2. Characteristics of Study Claims

CharacteristicAll cases (

n

5

156)Closed cases

(

n

5

124)

Mean age (years) (range) 43 (25–72) 43 (25–72)Race

White 139 (89.2%) 115 (93%)Black 2 (1.3%) 1 (1%)Hispanic 1 (0.6%) —Other 1 (0.6%) —Unknown 13 (8.3%) 8 (6%)

Marital statusMarried 107 (68.6%) 9 (7%)Single 11 (7.1%) 84 (68%)Divorced 25 (16.0%) 21 (17%)Widowed 6 (3.8%) 6 (5%)Separated 4 (2.6%) 4 (3%)Unknown 3 (1.9%) —

InsuranceCommercial 50 (32.1%) 41 (33%)HMO 64 (41.0%) 48 (39%)Medicare 12 (7.7%) 11 (9%)Public assistance 11 (7.1%) 7 (6%)Other 5 (3.2%) 8 (7%)Uninsured 8 (5.1%) 3 (2%)Unknown 6 (3.8%) 6 (4%)

Menopausal statusPremenopausal 108 (69.2%) 85 (68%)Perimenopausal 13 (8.3%) 11 (9%)Postmenopausal 33 (21.2%) 27 (22%)Unknown 2 (1.3%) 1 (1%)

Fibrocystic changes

a

History of FCC 43 (27.6%) 46 (37%)Dense fibroglandular tissue on mammogram 42 (26.9%) 42 (34%)FCC on exam 31 (19.9%) 31 (25%)Unknown/negative 40 (25.6%) 26 (21%)

a

74 (59.6%) patients had at least one of the above.

80

•

zylstra et al.

pregnancy, pregnancy related, menopause, screeningmammogram, lumpectomy/simple mastectomy, modi-fied radical mastectomy, and infiltrative intraductal car-cinoma were considered for inclusion in the model butwere not retained because of a lack of statistical signifi-cance in multivariable analysis.

Indemnity model:

where

X

1

5

1 if spread of disease at time of evaluation, 0 oth-erwiseX2 5 1 if lymph node dissection was done, 0 otherwiseX3 5 number of years after January 1,1986X4 5 (X3)2

X5 5 1 if patient is alive without metastatic disease atclosure of case, 0 otherwiseX6 5 1 if patient is alive with metastatic disease at clo-sure of case, 0 otherwise

The overall indemnity model demonstrates that in-demnity is increased with the spread of disease at thetime of evaluation. The indemnity paid out is highest inthose cases in which the patient is alive with metastaticdisease. Indemnity is lessened in cases in which patientshave died. Indemnity is least when patients are alivewithout metastatic disease. The performance of a lymph

Y e11.35752 1.0645X1 .4199X2.007391X3 .0201X4 .0130X5 .5497X6+–+

––+=

node dissection is associated with lower indemnity inthis model. Date of loss and date of loss squared jointlydescribe increasing indemnity during the study period.This reflects both general inflation and a change in legalpractice (see Fig. 1). The combination of date of loss anddate of loss squared determine the shape of this curve.

The WHP model was developed using the samemethod as the overall indemnity model, with restrictionby defendant specialty. The potential explanatory vari-ables considered for multivariable analysis in the develop-ment of the model for indemnity incurred by WHPs areshown in Table 8. The variables pregnancy related, CBEdocumented, SBE instruction, mass, biopsy, lumpectomy/simple mastectomy, modified radical mastectomy, andtamoxifen were not retained in the final model because ofa lack of statistical significance in multivariable analysis.

WHP indemnity model:

whereX1 5 1 if patient has at least one child, 0 otherwiseX2 5 1 if spread of disease at time of evaluation, 0 oth-erwiseX3 5 1 if lymph node dissection was done, 0 otherwiseX4 5 1 if patient presented with pain with or without amass, 0 otherwise

Y e11.085864 .5857X1 1.9242X2.9354X3 .0965X4 .7376X5 ,+ +

–+–=

Table 3. Indemnity and Legal Expenses on Breast Cancer Study Casesa

Number of cases Indemnity Expenses Total costs

Plaintiff jury verdict 4 $783,450 ($88,800–$1,500,000)

$99,126($40,521–$168,363)

$882,576($129,321–$1,625,930)

Defense jury verdict 18 $0 $64,442($27,456–$114,033)

$64,442($27,456–$114,033)

Settlement with indemnity 82 $422,644 ($18,000–$1,900,000)

$46,890 ($350–$225,999)

$469,534 ($18,350–$2,007,384)

Settlement without indemnity 20 $0 $17,259 ($200–$93,296)

$17,259($200–$93,296)

aData are presented as mean (range).

Table 4. Pregnancy Related

Cases withoutindemnity

Cases with indemnity

Variable Number Indemnitya Total

Pregnancy related 3 9 $725,030($150,000–$1,200,000)

$6,525,270

Non-pregnancy related 35 77 $406,043($18,000–$1,900,000)

$31,265,3300

aMean (range).

Defense of Breast Cancer Malpractice Claims • 81

X5 5 1 if patient presented mass with no pain, 0 other-wise.

The WHP model demonstrates that indemnity is in-creased with the spread of disease at the time of evalua-tion (consistent with the overall indemnity model). In-demnity is highest in those patients who present with amass with no pain (P2). Indemnity is less for patients whopresent with pain with or without a mass (P1). Indem-nity is the least for patients who have no presenting symp-toms (P0). The performance of a lymph node dissectionis associated with lower indemnity. Patients who are pa-rous at the time of diagnosis have decreased indemnity.

Review of the multidisciplinary algorithm deviationsper defendant as they related to frequency and indem-nity in the closed cases are shown in Table 9. Areas ofdeviation per specialty relative to frequency and indem-nity payments are shown in Table 10. The most preva-lent area of deviation in terms of frequency and indem-nity payments for WHPs is the failure to refer forsurgery or ultrasound. Seventy-five percent of thesecases were young women who presented with a breastabnormality, had a negative mammogram, and forwhom the physician suggested no further examination.Multidisciplinary clinical algorithms are shown in Fig-ures 2–6. Areas of inadequate documentation included

Table 5. Indemnity by Defendant Type on Breast Cancer Study Casesa

Cases withoutindemnity

Cases with indemnity

Defendant type Number Indemnityb Total

Women’s health practitioners 44 53 $404,425 ($10,000–$1,045,000)

$21,434,530

Radiologists 32 11 $283,409($30,000–$700,000)

$3,117,500

Surgeons 8 25 $404,263 ($62,500–$1,500,000)

$10,106,570

Pathologists 1 2 $208,500 ($42,000–$375,000)

$417,000

Physician assistant 1 0 $0 $0Radiology technician 0 1 $250,000 $250,000Corporations 22 7 $348,571

($100,000–$900,000)$2,440,000

Hospitals 5 0 $0 $0Total 113 99 $37,765,600

aSome cases had more than one specialty in the suit.bMean (range).

Table 6. Variables Selected for Multivariable Analysis in Development of the Model for a Successful Defense of a Case

Univariate Final model

Independent variableOdds ratio P

Oddsratio

Confidenceinterval

Records 0.43 ,0.001b 3.767 1.25–14.196Outcome — 0.043b — —O1 (alive without

metastatic spread) 1.34 — 1.898 .720–5.312O2 (alive with

metastatic spread) 0.29 — 0.427 .082–1.768O3 (death) 1.00 1.00Delay — 0.054a — —D1 (6–11 months) 0.49 — 0.584 .187–1.752D2 (.12 months) 0.22 — 0.222 .077–.5940Date of settlement — 0.068a — —(Date of settlement)2 — 0.069a — —Tamoxifen 2.61 0.036b — —Spread 3.13 0.085b — —

aLogistic regression.bFisher’s exact test.

Table 7. Variables Selected for Multivariable Analysisin Development of the Model for Indemnity on All Defendants

Independent variable

Univariate regression coefficient P

Final model regression coefficient

Spread 1.50 0.001 1.0645Node dissection 20.68 0.002 20.4199O1 (alive without metastatic spread) 20.09 — 20.0130O2 (alive with metastatic spread) 0.60 — 0.5497Date of loss 21.15 0.048 20.0074(Date of loss)2 0.02 — 0.0210Bone marrow transplant 20.53 0.355Smoke 20.30 0.187 —Patient pregnant 0.75 0.069 —Pregnancy related 0.80 0.018 —Menopause 20.30 0.184 —Screening mammogram 20.36 0.088 —Lumpectomy/simple mastectomy 20.54 0.017 —Modified radical mastectomy 0.41 0.058 —Infiltrative/intraductal carcinoma 0.45 0.149 —

82 • zylstra et al.

failed notation of CBE (13), recommended follow-up(12), presenting symptom (7), altered record (4), tele-phone calls (3), informed (3), and illegibility (2). Multi-specialty risk management concerns are the greatest interms of number of defendants and indemnity paid inthe areas of follow-up, referral, and communication(Table 11).

DISCUSSION

According to a recent PIAA study, insurers now paymore money for breast cancer cases than for any otherdisease or condition except those involving brain-dam-aged infants (2,18). This PIAA study was headlined ina recent law journal: “The Worst List—Breast Cancer

Now Leading Source of Medical Malpractice Claims”(21).

The authors of the PIAA study were impressed withthe young age of most of the plaintiffs and concludedthat “once a patient presents with a problem, regardlessof age, a physician should treat the complaint seriouslyand pursue all methods available in making a diagno-sis.” The most common reason given for delay in diag-nosis was “Physical findings failed to impress the physi-cian”; the second most common reason was “Negativemammogram report.” Kern (2) proposed that a success-ful risk prevention strategy depended on a better under-standing of which patients and physicians fall into ahigh-risk profile for frequent diagnostic errors. He de-scribed a “triad of errors” for delayed diagnosis ofbreast cancer which profiles women at highest risk to bethose under age 45 years with a self-discovered breastmass and a negative mammogram. Certainly a negativemammogram report should not deter a clinician fromfurther evaluation in the presence of a palpable masssuspicious for carcinoma.

In young patients, masses are frequently assumed tobe fibrocystic change. Both mammography and CBE areless sensitive in young patients. Kern (22) recommendsthat immediate tissue biopsy by fine needle aspirationbiopsy (FNAB) or core-cutting needle biopsy, followedby excision of solid lesions would be cost effective andprevent the majority of diagnostic delays that lead to lit-igation in these high-risk women. Mitnick (23) prefersthe primary use of mammography and ultrasound priorto FNAB, core-cutting biopsy, or surgical biopsy. Thesensitivity of FNAB is about 90%, so 10% of carcino-mas go undetected with this approach (24,25). Adjunc-tive ultrasound use in young women with dense fibro-glandular tissue with suggestive dominant masses wouldlikely reduce the false-negative rate of this technique andwould further characterize cancerous lesions that can belocalized for core or surgical excision at little addedcost. Review of the specific portion of the multidisci-plinary clinical algorithm showed that clinicians werenot faulted for failure to perform a mammogram, butwere faulted for failure to perform an examination, ul-trasound, and/or biopsy in pregnant patients.

The successful defense model shows that the least de-fensible cases were those in which the plaintiff was alivewith metastatic disease, there had been a delay in diag-nosis of 1 year or more, and the physician’s record keep-ing was poor. Current diagnostic paradigms assert thatdelay between onset of symptoms and initiation of treat-ment for breast cancer results in presentation at a more

Figure 1. Overall indemnity model.

Table 8. Variables Selected for Multivariable Analysis in Development of the Model for Indemnity on Women’s Health Practitioners

Independent variable

Univariate regression coefficient P

Final model regression coefficient

Spread 2.12 0.028 1.9242Node dissection 20.86 0.003 20.9354Present — 0.004 —P1 (pain with or without mass) 21.07 — 20.0965P2 (mass without pain) 20.68 — 20.7376Parous state 20.52 0.187 20.5857Patient pregnant 1.02 0.001 —Pregnancy related 0.98 0.001 —CBE documented 20.45 0.116 —SBE instruction 0.41 0.104 —Mass 0.64 0.062 —Biopsy 0.77 0.001 —Lumpectomy/simple mastectomy 20.62 0.064 —Modified radical mastectomy 0.54 0.077 —Tamoxifen 0.36 0.175 —


advanced state with poorer survival (22,24–28). Thisstudy demonstrates that inadequate record keeping willdestroy any chance of a successful defense.

The overall indemnity model indicates that indemnitypayments were highest in cases in which the followingthree conditions were met: the plaintiff was alive butwith metastatic disease, metastasis was present at thetime of evaluation, and the date of occurence was closerto the present (see Fig. 1). Cases in which the plaintiffhas already succumbed to the disease are often easier todefend than those in which the patient is still alive. Thedefense strategy in many of these cases revolves arounddetermining whether earlier diagnosis would have changedclinical outcome. Expert reviewers in oncology often be-lieve the patient’s cancer to be so aggressive that earlierdiagnosis would not have made a difference in the ulti-mate clinical course. Despite that, 70% of cases closedwith an indemnity payment.

The WHP model, like the overall indemnity model,demonstrates that indemnity is increased with the spreadof disease at the time of evaluation. The highest indem-nity was found in cases in which the patient presentedwith a mass without pain. Our study showed that 77%of the lesions were initially discovered by the patient onbreast self-examination. It is imperative that physiciansbe diligent in their imaging and pathology examinationsand not act quickly to dismiss an abnormality as fibro-cystic breast disease.

Indemnity was less for patients who presented withpain and for those who had no presenting symptoms.Courts lower indemnity awards somewhat if an aggres-sive diagnostic procedure (lymph node dissection) hasbeen performed. In our study, 80% of the patients whohad metastatic disease and were alive did not have alymph node dissection performed. Parous patients had alower indemnity. In summary, the indemnity in breast

Table 9. Algorithm Deviations per Defendant

Number of defendantswithout indemnity

Defendants with indemnity

Deviation Number Indemnitya Total

Failure to refer surgery/ultrasoundb 11 23 $408,125 ($10,000–$750,000)

$9,386,863

Failure to do biopsy 2 14 $453,684 ($75,000–$1,500,000)

$6,351,570

Failure to order mammogram 8 12 $380,417 ($40,000–$1,000,000)

$4,565,000

Follow-up 8 11 $373,652($62,500–$900,000)

$4,110,167

CBE 1 9 $485,556 ($150,000–$1,045,000)

$4,370,000

Misread mammogram/failure to do additional views 15 10 $296,750($30,000–$700,000)

$2,967,500

Surgical evaluation 1 6 $402,917 ($90,000–$1,500,000)

$2,417,500

Failure to send FNA to cytology 1 0 $0 $0Totals 47 85 $34,168,600

aMean (range).bCBE 5 breast abnormality mammogram 5 negative: failure to refer to surgeon or ultrasonography (account for most of these cases).

Table 10. Primary Areas of Deviations per Defendant Group

DeviationNumber of cases

Indemnity payments

Women’s health practitionersFailure to refer for surgery or ultrasound 24 $6.3 millionFailure to perform a CBE 10 $4.4 millionFailure to order a diagnostic mammogram 14 $3.7 million

RadiologistsMisreading of a mammogram 10 $2.3 millionFailure to obtain spot compression or magnification views 15 $ .7 million

SurgeonsFailure to perform image-guided biopsy in nonpalpable mass 8 $3.6 millionFailure to biopsy a palpable mass 8 $2.8 millionFailure to fully evaluate prior history examination, mammogram, or biopsy result 7 $2.4 million


malpractice cases increases when system failures resultin advanced disease with resulting poor prognosis. In-demnity is exacerbated when the population served isyoung, nulliparous, or pregnant.

CONCLUSION

The failure to diagnose breast cancer is one of themost costly types of cases for malpractice insurers. Thenumbers of such claims are currently increasing. In or-der to reduce the risk of involvement in a suit alleging“failure to diagnose cancer,” physicians should incor-porate certain clinical guidelines and risk managementstrategies in their practice. Critical to appropriate qual-ity of care rendered is a comprehensive tracking systemthat incorporates documented clinical data and com-munication of that data with health care professionals.

D’Orsi and Debor (29) recently highlighted the impor-tance of unambiguous radiological data communicatedalong established guidelines using standardized BreastImaging Reporting and Data Systems (BI-RADS) termi-nology. Other inclusive data kept as nonidentifiable aspossible consists of patient demographics and multispe-cialty tissue diagnosis. Biopsy and clinically relevantdata must be collected and correlated with mammo-graphic BI-RADS interpretation for quality assessmentas required under the Mammographic Quality Stan-dards Act of 1990 (30,31). Positive biopsy results mustbe entered into state cancer registries, as required by theAmerican College of Surgery, and can be validated withsuch a vehicle. Appropriate patient follow-up docu-mentation is automated to optimize accuracy and effi-ciency.

Figure 2. Multidisciplinary clinical breast care algorithm: clinical breast exam.


Communication with patients can be enhanced byproviders at the initial evaluation (risk assessment, CBE,and mammography). Effective communication is mostcrucial when results are abnormal and additional imag-ing is required. In these cases, the physician might con-sult the algorithm developed as part of this study and re-fer the patient for complementary ultrasound whichshould be performed immediately. With inconsistent di-agnostic findings, an appropriate tissue diagnosis, forexample, FNAB, core-cutting needle biopsy, or open bi-opsy for palpable lesions, and image-guided biopsies

(using ultrasound, stereotactic, or wire-guided methods)may be required. Pathology results, additional tests, andfollow-up should be discussed to patient satisfaction.

With a positive diagnosis, options for specialized careare best offered in the context of regionally associatedmultidisciplinary breast centers. The consumer of breastservices, for example, a woman with a breast abnormal-ity under clinical breast examination, expects to receivefocussed attention, with all diagnostic and treatment in-terventions governed by a well-planned, coordinatedstrategy. She needs to know that her care follows a cohe-

Figure 3. Multidisciplinary clinical breast care algorithm: mammography.


sive plan with predetermined expectations of outcome,well adapted to her personal beliefs, lifestyle, and socialcircumstances. She needs to be presented with choicesthat are available and consistent with optimal care sothat she can participate in the fundamental decisionsabout her care.

The referring health care provider is responsible forhis/her patients’ routine breast care. He/she is also re-

sponsible for the follow-up, monitoring, and trackingof women whose results are abnormal, including thosefor whom a biopsy is recommended according to theAgency for Health Care Policy and Research (32,33).This responsibility is shared with the mammography fa-cility. The facility that performed the initial mammo-gram is responsible for performing or arranging for fu-ture examinations or added views and/or procedures to

Figure 4. Multidisciplinary clinical breast care algorithm: ultrasonography.


be performed. Each facility must have a system to trackpositive mammographic findings and a process for cor-relating findings with biopsy results (32). It also has theduty to communicate urgent or significant findings to

physicians and sometimes to patients directly, as deter-mined by the courts (16,34–37). The Practice StandardsClaims Survey (38) states that 75% of all breast cancercases in which indemnity is paid include the allegation

Figure 5. Multidisciplinary clinical breast care algorithm: tissue diagnosis.


that the patient did not receive appropriate follow-up.This article echoes this (Table 11).

The courts have not fully defined the duty of radiolo-gists relative to tracking of patients after percutaneous

core biopsies; however, written documentation of fol-low-up is mandated (39,40). Once a patient is under aphysician’s care, no matter the specialty, that physicianis obligated to provide appropriate evaluation, treat-

Figure 6.


ment, and follow-up. All data gathered must be subjectto objective and balanced quality assurance review toensure its accuracy and privacy. These system changeswill lead to a quality of care that is already available topatients suffering from other illnesses and will likely re-duce patient dissatisfaction, unanticipated poor out-comes, and resultant litigation.

Freedom from a lawsuit alleging “failure to diagnosecancer” can never be guaranteed. However, the likeli-hood of involvement in such a case can be minimized ifthe physician uses formal guidelines to direct his/her diag-nostic examination of all patients, and particularly thoseconsidered at risk, that is, young nulliparous women whopresent with a painless mass. The chance that a suit canbe successfully defended cannot be predicted from thedata presented in our study. However, these data suggestthat the chance for a successful defense increases withcomprehensive documentation, including notation of clin-ical breast examination, recommended follow-up, and thepatient’s presenting clinical status.

History is not an absolute predictor of the future. Inbreast cancer cases, however, it can help guide presentclinical decision making. Physicians who have failed tomake timely diagnoses of breast cancer and patientswho have suffered injury or death because of those er-rors offer lessons for those who wish to avoid similar sit-uations in the future. Those are the lessons upon whichwe have built the models and developed the multidisci-plinary clinical care algorithm that are presented as thecore of this article.

REFERENCES

1. Ioannidou-Mouzaka L, Hortobagyi GN, Gros D, etal. Senology—the urgent need for a specialty. Breast 1998;4:280–84.

2. Kern KA. Breast cancer and malpractice: a surgeon’sperspective on risk prevention. Semin Breast Dis 1998;1:22–31.

3. Zylstra S, Bors-Koefoed R, Mondor M, et al. A statisti-cal model for predicting outcome in breast cancer malpracticelawsuits. Obstet Gynecol 1994;84:392–98.

4. Dennis CR, Gardner B, Lim B. Analysis of survival and

recurrence vs. patient and doctor delay in treatment of breastcancer. Cancer 1995;35:714–20.

5. Afzelius P, Zedeler K, Sommer H, et al. Patient’s and doc-tor’s delay in primary breast cancer. Acta Oncol 1994;33:345–51.

6. Mitnick JS, Vazquez MF, Kronovet SZ, et al. Malprac-tice litigation involving patients with carcinoma of the breast.J Am Coll Surg 1995;181:315–21.

7. Chew FS, O’Reilly MAR, Schuckenbrock CM, et al.Medicolegal aspects of breast cancer in U.S. federal medical fa-cilities: analysis of 80 claims. Mil Med 1996;161:329–33.

8. Physcian Insurers Association of America. Breast can-cer study. Lawrenceville, NJ: Physicians Insurers Associationof America, 1990.

9. Cady B. Common factors in claims of failure to diag-nose breast cancer. Boston, MA: Risk Management Founda-tion of the Harvard Medical Institutions, Inc., 1992;13:7–8.

10. Kern K. Causes of breast cancer malpractice litigation:a 20-year civil court review. Arch Surg 1992;127:542–47.

11. Osuch JR, Bonham VL. The timely diagnosis of breastcancer. Cancer 1994;74:271–78.

12. Manuel BM. Physician liability: new areas of concernunder managed care. Am Coll Surg Bull 1995;80:24–26.

13. Skoumai SM, Florell SR, Bydalek MK, et al. Malprac-tice protection: communication of diagnostic uncertainty. Di-agn Cytopathol 1996;14:385–89.

14. Rubenchik I, Sneige N, Edeiken B, et al. In search ofspecimen adequacy in fine-needle aspirates of nonpalpablebreast lesions. Am J Clin Pathol 1997;108:13–18.

15. Brenner RJ. Interventional procedures of the breast:medicolegal considerations. Radiology 1995;195:611–15.

16. Homer MJ, Berlin L. Radiography of the surgicalbreast biopsy specimen. AJR 1998;171:1197–99.

17. Racenstein MJ, Berlin L. False-negative core biopsy ofthe breast. AJR 1998;171:927–30.

18. Physician Insurers Association of America. Breast can-cer study 1995. Washington, DC: Physician Insurers Associa-tion of America, 1995.

19. NCCN practice guidelines. Screening for and evalua-tion of suspicious breast lesions. National ComprehensiveCancer Network. Oncology 1998;12(11A):89–138.

20. Freund RJ, Littell RC. SAS system for regression, 2nded. Cary, NC: SAS Institute, 1991:210.

21. Jones TL. The worst list: breast cancer now leading sourceof medical malpractice claims. Texas Med 1996;92(2):34–36.

22. Kern KA. Letter to the editor. J Am Coll Surg 1996;182:462–64.

23. Mitnick JS, Vazquez MF, Plesser KP, et al. Breast can-cer malpractice litigation in New York State. Radiology 1993;189:673–76.

24. Lannin DR, Silverman JF, Walker C, et al. Cost effec-tiveness of fine needle biopsy of the breast. Ann Surg 1986;203:474–80.

25. Donegan WL. Current concepts: evaluation of a palpa-ble mass. N Engl J Med 1992;327:937–42.

Table 11. Risk Management Deviations per Defendant

DeviationNumber of defendants

Indemnitypayments

Follow-up 58 $15.7 millionRefer issue 52 $12.8 millionCommunication 16 $2.7 millionContraindication of presenting symptoms 12 $2.8 millionMammogram tracking 8 $2.4 million


26. Charlson ME. Delay in the treatment of carcinoma ofthe breast. Surg Gynecol Obstet 1985;160:393–99.

27. Machiavelli M, Leone B, Romero A, et al. Relation be-tween delay and survival in 596 patients with breast cancer.Oncology 1989;49:78–82.

28. Dohrmann PJ, Hughes ESR, McDermott F, et al. Symp-tom duration, tumor staging and survival in patients with carci-noma of the breast. Surg Gynecol Obstet 1982;154:707–10.

29. D’Orsi CJ, Debor MD. Communication issues in breastimaging. Radiol Clin N Am 1995;33:1231–45.

30. American College of Radiology. ACR standard for per-formance of stereotactically guided breast interventional pro-cedures. Reston, VA: American College of Radiology, 1996.

31. Houn F. What a mammography facility should do toprepare for the MQSA inspection. Rockville, MD: Departmentof Health and Human Services, 1994.

32. Basset LW, Henrik RE, Bassford TL. High qualitymammography: information for referring providers. Agencyfor Health Care Policy and Research no. 95-0633. Rockville,MD: Department of Health and Human Services, 1994.

33. Basset LW, Henrik RE, Bassford TL. Quality determi-nants of mammography. Agency for Health Care Policy andResearch no. 95-0632. Rockville, MD: Department of Healthand Human Services, 1994.

34. Mammography quality standards: proposed rules(CFR 900). Fed Reg 1996;65:4911–12.

35. Berlin L. To telephone or not to telephone: how high isthe standard? AJR 1992;159:1335–39.

36. Berlin L. Communication of the urgent finding. AJR1996;166:513–15.

37. Berlin L. Communication of the significant but not ur-gent finding. AJR 1997;168:329–31.

38. Physicians Insurers Association of America and Ameri-can College of Radiology. Practice standards claims survey.Rockville, MD: Physician Insurers Association of America, 1997.

39. Goodman KA, Birdwell RL, Ikeda DM. Compliancewith recommended follow-up after percutaneous core biopsy.AJR 1997;170:89–92.

40. Berlin L. Tracking for breast cancer. AJR 1998;170:93–95.

defense of breast cancer malpractice claims · clinical and risk management factors that could help...

Documents