S o c i a l a n d E c o n o m i c I s s u e s in R a d i o l o g y
Estimating the Cost-Effectiveness of Stereotaxic Biopsy for Nonpalpable Breast Abnormalities' A Decision Analysis Model Bruce E. Hillner, MD 1, Harry D. Bear, MD, PhD 2, Laurie L. Fajardo, MD 3
Rationale and Objectives. We examined the clinical and economic trade-offs of shifting from surgical excisional biopsy to stereotaxic core breast biopsy for evaluating nonpalpable, mammographically detected breast lesions. Methods. A decision analysis model c o m p a r e d strategies beginning with excisional or stereotaxic core biopsy for hypothetical cohorts of 1,000 women. All w o m e n with negative initial biopsies had a 6-month follow-up m a m m o g r a m . Sensitivities and specificities w e r e based on the literature and expert estimates. Pretest probabilities of invasive cancer and in situ cancer were each 10% based on m a m m o g r a p h i c features. Adjusted costs were based on an audit of patients evaluated at the Medical College of Virginia and physician relative value units., Results. Per 1,000 w o m e n , with an expected rate of 100 invasive and 100 in situ cancers, the stereotaxic core biopsy strategy w o u l d initially miss 6.7 invasive and 12.4 in situ cases. Most of these would be detected at 6-month follow-ups. Of the w o m e n having a stereotaxic core biopsy, 75.7% avoided a surgical procedure. Using stereotaxic core biopsy saved $804 per woman. Continuing to initially use surgical biopsy, total management costs were an additional $42,100 per each case of early detected invasive or in situ cancer. A speculative sensitivity analysis, in which the prognosis of invasive cancer was worse if diagnosis was delayed by 6 months, indicated that surgical biopsy had an incremental cost of $156,700 per additional life year gained.
Conclusion. Using conservative estimates for the false-negative rate of stereotaxic core breast biopsy, widespread use of stereotaxic biopsy is projected to have substantial cost savings with a slight c o m p r o m i s e in the rate of early detection. Whether the decrementai cost-effectiveness is acceptable is dependent on the natural history of cancers w h o s e diagnosis is delayed. Key Words. Decision modeling; breast biopsy; breast neoplasm; cost-effectiveness analysis.
lthough screening m a m m o g r a p h y has b e e n Shown to reduce breast cancer mortality, the positive predictive value of m a m m o g r a p h i c abnormalities is only 10-30% [1]. Therefore, screening results in most abnormalities being diagnosed as benign, but the incurred evaluation has substantial financial and psychological costs. About one third of the financial cost of a breast cancer screening program is estimated to be from the costs of o p e n surgical biopsy for benign disease [2, 3]. Stereotaxic core breast biopsy has b e e n advocated as a less invasive, lower cost, and effective a p p r o a c h that m a n y believe is ready to replace open surgical exclsional biopsy to diagnose nonpalpable lesions detected by m a m m o g r a p h y [4]. Since the first reports of stereotaxic biopsy in 1989, there has b e e n an accelerating adoption of this technology. Unfortunately, stereotaxic biopsy has disseminated into routine use without con-
A
From the 1Massey Cancer Center and Department of Medicine, Medical College of Virginia, Virginia Commonwealth University, Richmond, VA; 2 Massey Cancer Center and Department of Surgery, Medical College of Virginia, Virginia Commonwealth University, Richmond, VA; and aDepartment of Radiology, University of Virginia School of Medicine, Charlottesville, VA. This research was supportedpartly by a Faculty Research Award to B. E. Hillner, MD, from the American Cancer Society and a General Electric Radiology Research Academic Fellowship to L. L. Fajardo, MD. A preliminary version of this work was presented at the American Society of Clinical Oncology meeting held in Los Angeles, CA, in May 1995.
Address reprint requests to B. E. Hillner, MD, Division of General Medicine, Medical College of Virginia, Box 980170, Richmond, VA 23298. Received August 31, 1995, and accepted for publication after revision November 21, 1995. Acad Radiol 1996;3:351-360
© 1996, Association of University Radiologists
351
H I L L N E R ET AL.
trolled scientific studies [5, 6]. Results from case series of ~tereotaxic biopsy are flawed because most patients did not have a confirmatory excisional biopsy, the reference standard, or they had only minimal follow-up [7-9]. Therefore, the frequency at which stereotaxic bidpsy misses cancer cannot be definitively determined, but it is unlikely to be zero. Decision analysis is useful in projecting the benefits and costs of a technology before randomized trials results are known or if they are unlikely to be performed [i0]. In this article, we used decision analysis to estimate the decremental cost-effectiveness of using stereotaxic core biopsy rather than excisional biopsy. Such decremental cost-effectiveness (i.e., financial savings gained by a potential compromise in cancer detection [11]) is being implied as acceptable by advocates of stereotaxic core biopsy. Given the global pressures to contain health care costs, finding alternatives that reduce costs to such a degree that physicians or patients are willing to sacrifice a small risk is a new situation in health care, which preViously was concerned primarily with efficacy [12]. MATERIALS AND METHODS
Consider a population of patients who have a single nonpalpable abnormality on screening mammography sufficiently worrisome to warrant definitive histopathologic diagnosis. Should these patients have an excisional or a stereotaxic core breast biopsy? Using available estim;/tes of test characteristics and costs, what are the tradeoffs between unnecessary surgical biopsies, delayed or missed cancer diagnoses, and financial savings?
vol. 3, No. 4, April 1996
ity analysis,-the cost implications of different rates of reexcisions or mastectomy after initial excisional biopsy were assessed. (7) The time horizon was limited to 6 months. (8) There was no change in the stage or prognosis of invasive cancer if a false-negative stereotaxic biopsy resulted in a 6-month delay in diagnosis. This is a critical assumption and biases the analysis toward stereotaxic biopsy. Although there is extensive literature on the change in progiq0sis in women with palpable breast lesions who delay seeking treatment, we could find no reports about the consequences of a 6-month or longer delay in the diagnosis of nonpalpable lesions detectable solely by mammography. In the sensitivity analysis, we explored the consequences of a change in the survival of women whose diagnoses was delayed. Three disease categories were modeled: invasive breast cancer, in situ cancer, and no cancer. These categories and their temporal occurrence defined the model's end points: (1) early detection of disease (i.e., invasive breast cancer or in situ cancer); (2) diagnosis delayed for 6 months on the basis of detection at mammographic follow-up; (3) missed diagnosis, in which diagnosis is delayed beyond 6 months; (4) the financial cost to each of these respective end points; and (5) the number of unnecessary or negative excisional biopsies. Traditional decision trees were used given the short study period. For ease of reporting, the model followed the course for groups of 1,000 patients through each strategy. The model was developed using the decision software SMLTREE (version 3.0; James Hollenberg, Roslyn, NY). Decision Tree
Assumptions
The following simplifying assumptions were made in structuring the model: (1) All patients having negative initial biopsy results had mammographic follow-up at 6 months. (2) There were no serious complications associated with the procedure or general anesthesia. (3) There Was no specific patient age. (4) Concurrent in situ and mvaslve cancers were excluded but were assessed in a Sensitivity analysis. (5) The probabilities of indeterminate and insufficient specimens were constant and were not conditional on the probability of cancer. (6) All patients with invasive cancer or in situ cancer had breast-conserving surgery (i.e., lumpectomy for in situ cancer and lumpectomy plus axillary node dissection for invasive cancer). By minimizing the number of surgical procedures, the baseline assumptions for the cost analysis were biased in favor of excisional biopsy. In the sensitiv-
352
The structure of the decision tree, shown in Figure 1, is similar for the two approaches. The initial branch reflects the adequacy of the biopsy specimen for pathologic examination. A specimen would be considered inadequate if it did not contain breast parenchyma or if the postprocedure specimen X-ray or man~nogram showed that the suspected lesion or microcalcification was missed. If the specimen was inadequate, the ..original pro:. cedure would be repeated. All excisional biopsies were preceded by a mammograln-guided needle localization. Besides finding disease or no cancer, a biopsy specimen could be indeterminate, meaning that the specimen was large enough but that a definitive benign or malignant pathologic diagnosis could not be made. In the baseline analysis, we assumed that an indeterminate result could occur only after a stereotaxic biopsy and that the subsequent action would be to repeat a
Vol. 3, No. 4, April 1996
BREAST
BIOPSY
DECISION
ANALYSIS
T Inadequate
IProcedure I o~n~
Miss
0%~.i'a,
~'~m~°
le Mammo~
Indeterminate
~
.
Original Repeat
FIGURE 1. Decision tree. The common decision tree was similar for the two different biopsy strategies. The solid circles represent chance nodes or events. The triangle is a boolean or if-then node. The biopsy may be adequate or inadequate in size for pathologic examination. Of adequate biopsy, the true pathologic status is either not cancer (Not CA), in situ cancer (ductal carcinoma in situ [DCIS]), or invasive cancer (CA). The potential biopsy results are shown on the next row as either indeterminate, missed, or early detection. No false-positive results were considered. The subsequent actions are shown on the bottom row. If the biopsy showed no cancer, patients went to mammogram follow-up at 6 months. The action after detection of in situ or invasive cancer depended on the type of original biopsy (shown as a triangle or boolean node). BCS = breast conserving surgery, AND = axillary node dissection, ExB = excisional biopsy.
stereotaxic biopsy. If, on repeat biopsy, an inadequate or indeterminate specimen was found, then an excisional biopsy would be done. The results of the biopsy could suggest invasive cancer, in situ cancer, or no cancer. If invasive cancer was found, the patient had definitive cancer surgery. The subsequent action d e p e n d e d on the type of initial biopsy. If the initial procedure was an excisional biopsy, assuming that the excision was complete, then the surgery was an axillary node dissection. If the initial action was a stereotaxic biopsy, then the Surgery was a lumpectomy and axillary node dissection. If the excisional biopsy indicated in situ cancer, then the subsequent action was no further surgery. If the initial action was stereotaxic biopsy, then the patient had a lumpectomy. If invasive cancer or in situ cancer was missed, the patient had a m a m m o g r a m at 6 months that could detect the cancer, which d e p e n d e d on the sensitivity of mammography. If the initial biopsy did not suggest invasive cancer or in situ cancer, patients had a 6month mammogram that could be normal or a falsepositive. If the follow-up m a m m o g r a m was abnormal, patients went to excisional biopsy.
DATA
The data required, the estimates used, and the reference sources used are shown in Table 1: (1) the probabilities of invasive and in situ cancer based on the mammographic findings; (2) test characteristics of each of the respective approaches; and (3) the financial costs. Probabilities
The predictive values for different mammographic patterns of detecting breast cancer have been rel3orted previously [13, 14]. Categories of suspicion have been suggested that range from "probably benign" to "high suspicion." For the current analysis, we assumed that suspicious lesions had an overalt 0-iobal~ility of cancer of 20%: a probability of invasive cancer of 10% and a probability of in situ cancer of 10%. Such a group would be a heterogenous group of stellate and nonstellate masses or clusters of microcalcifications [15]. Test C h a r a c t e r i s t i c s
The test characteristics for each of the diagnostic procedures were based on our synthesis of the litera353
H I L L N E R ET AL.
voi. 3, No. 4, April 1996
TABLE 1: Data Elements i
Variable
Base Case
Range
.10 .10
.05-.20 .05-.20
Probabilities (%) Invasive cancer In situ cancer Stereotaxic breast biopsy
Source [13-151
Authors' synthesis based on Table 2
Inadequate .05 Indeterminate .02 Sensitivity for invasive cancer .90 Sensitivity for in situ cancer .85 Specificity for invasive cancer or in situ .98 cancer Surgical biopsy Inadequate .01 Indeterminate .00 Sensitivity for invasive cancer .99 Sensitivity for in situ cancer .99 Specificity for invasive cancer or in situ 1.00 cancer Mammogram at 6-month follow-up Sensitivity for invasive cancer .85 Sensitivity for in situ cancer .85 Specificity for invasive cancer or in situ .95 cancer Cost a Stereotaxic biopsy $800 Surgical biopsy with needle localization $2,000 Lumpectomy $3,000 Lumpectomy and axillary node $6,500 dissection Mastectomy with axillary node $7,500 dissection Axillary node dissection $5,000 Bilateral mammogram $60
.01 -.10 .01-.10 ~ .85-.98 .80-.93 ,96-1.00 [38, 39] .00-.05 ,00-.01 .98-1.00 .98-1.00 1.00-1.00 [40-42] .80-,90 .75-,90 .90-.98
$600-1,500 $1,500-4,000 $2,000-6,000 $5,000-13,000 $6,000-15,000 $3,500-10,000 $40-120
i
aCosts are estimates of true costs based on assessment of patients at the Medical College of Virginia on all professional and facility charges adjusted by institution-specific cost-to-charge ratios and professional relative value units.
ture. The critical variable was the false-negative rate for the stereotaxic biopsy. The actual false-negative rate was difficult to estimate because of the inconsistency of the findings, patient characteristics, and limited follow-up from available series summarized in Table 2. The estimate used was based on our expert synthesis, not on any formal statistical or meta-analysis methodology. The test characteristics for stereotaxic biopsy at first glance may appear low, but they were due to the inclusion of insufficient and indeterminate results in the model. For both strategies, no false-positive results were considered because the rare reports of stereotaxic biopsy showing invasive cancer not confirmed at surgery probably reflect complete removal of tumors only a few millimeters in diameter. This was shown for stereotaxic biopsy by adding .02 indeterminate and .98 354
specificity. The true-positive rate or the sensitivity of stereotaxic biopsy was estimated to be higher for invasive cancer (.90) than for in situ cancer (.85).
TABLE 2: Stereotaxic Breast Biopsy Series: Agreement with Surgical Biopsy i
Source
No. of Biopsies Reported
Agreement with SurgicaL Biopsy
,,.. % Surgical Confirmation of" Core Biopsy
[4] [23] [43] [44] [23] [45] [46] [47]
1363 104 254 100 53 250 102 102
99 90 83 94 91 71 96 87
23 100 48 100 76 100 100 100
These results assume that an adequate specimen was obtained.
Vol. 3, No. 4, April 1996
For the initial stereotaxic biopsy strategy, the falsenegative rate was .08 for invasive cancer and .13 for in situ cancer. In the sensitivity analysis, lower false-negative rates, consistent with the best published series (which might have been due to publication bias), and higher rates were assessed. We could not find any published estimates of the accuracy of follow-up m a m m o g r a p h y versus screening m a m m o g r a p h y . In our opinion, the sensitivity should be similar to screening m a m m o g r a p h y , but the specificity may be better with fewer false-positives. Costs
We adopted a societal viewpoint and considered only direct health care costs [16, 17]. Financial costs were based on a detailed analysis of charges for patients from 1994 to 1995 at the Medical College of Virginia, a hospital-based facility. These charges included all facility and physician components. All patients were biopsied as outpatients. All stereotaxic biopsies and 50% of the excisional biopsies were done under local anesthesia. Costs for subsequent surgery were considered. Because charges overestimate costs and Medicare payments may inconsistently reflect costs, we used a combination of hospital and professional charges to estimate actual costs. Hospital-based charges were broken down into 15 cost centers and adjusted by institution-specific cost to charge estimates. These specific cost-to-charge estimates have been used in other econornic analyses [10, 18, 19]. Costs for professional services (e.g., surgical, anesthesia, pathology, radiologist) for each procedure code were estimated flom the relative value units for each of these services multiplied by $8 per unit [20]. The choice of $8 per unit was based on our best estimate of professional effort in light of local patterns of payment by preferred provide>type insurance plans. The model considered financial costs througl:~ 6 months of follow-up to two different end points: cost-to-definitive diagnosis or cost to-definitive diagnosis and treatment. In the calculation of cost-effectivoness ratios and sensitivity analysis, costs were restricted to total management; that is, defined diagnosis and treatment because this reflected the total care costs. The excisional biopsy was assumed to be a definitive, one-step treatment with clear margins. Therefore, reexcision was not required, and only patients with invasive cancer required an additional axillary node dissection. In sensitivity analysis, we assessed the financial consequences of additional local reexcisions or subsequent rnastectomy after initial excisional biopsy or the
BREAST BIOPSY DECISION ANALYSIS
additional cost of mastectomy instead of lumpectomy for in situ cancer after stereotaxic biopsy. RESULTS
Table 3 shows the baseline results per 1,000 w o m e n broken down by invasive cancer, in situ cancer, and costs. Of these 1,000 women, 100 have invasive cancer and 100 have in situ cancer. The initial biopsy using stereotaxic biopsy would be predicted to miss 6.7 invasive cancers. However, most of these would be detected at the 6-month mammogram. Therefore, only about one invasive cancer is not detected by 6 months per 1,000 women. In a similar manner, the model projects for in situ cancer. The outcome's differences between stereotaxic and excisional biopsy are 9.8 cases of in situ cancer per 1,000 w o m e n delayed until 6-month follow-up and 2.6 cases per 1,000 not detected within 6 months. The average cost per woman is lower for stereotaxic biopsy. The excisional biopsy strategy cost per w o m a n is $1,065 greater to the end point of making a diagnosis and $804 more to completion of definitive surgical treatment compared with stereotaxic biopsy. In all subsequent results, we report only the costs for total management. If the initial excisional biopsy strategy is preferred, then the financial trade-offs c o m p a r e d with using initial stereotaxic biopsy can be estimated as s h o w n in Table 4. The incremental cost per effect depends on whether TABLE 3: Results per 1,000 Women Strategy
End Point
Difference Between Stereotaxic Excisienal Stereotaxic Biopsy Biopsy Biopsy and Excisional Biopsy
Invasive cancer 100 100 , Diagnosis at initial biopsy 92.3 99.0 -6.7 Diagnosis delayed until 6.4 0.8 5.6 6-month follow-up Diagnosis delayed 1.3 0.2 1.1 beyond 6-month follow-up In situ cancer 100 - " :100 Diagnosis at initial biopsy 86.6 99.0 -12.4 Diagnosis delayed until 10.6 0.8 9.8 6-month follow-up Diagnosis delayed 2.8 0.2 2.6 beyond 6-month follow-up Excisional biopsies or 243 1,000 -757 surgeries performed Costs $ to Diagnosis 1,066,000 2,131,000 -1,065,000 $ to Definitive treatment 1,975,000 2,779,000 -804,000
355
H I L L N E R ET AL.
vol. 3, No. 4, April 1996
the end point is solely invasive cancer or invasive and in situ cancer and whether the time line of detection is immediate or occurs within 6 months. The excisional biopsy strategy costs $120,000 per additional invasive cancer, or $43,000 per additional case of in situ or invasive cancer. In other words, an additional 19 of 1,000 w o m e n will have earlier management of their invasive or in situ cancer at an additional cost of $804,000 tot all 1,000 women. The incremental cost ratios associated with excisional biopsy markedly increase when the time horizon extends to 6 months. This increase is attributable to the finding that in 15 of the 19 w o m e n whose cancer was initially missed, the cancer was detected at the 6-month followup. Therefore, the incremental cost per diagnosis and treatment after 6 months ranges from $730,000 per invasive cancer and $217,300 per invasive and in situ cancer. Sensitivity A n a l y s i s
Multiple "what it" or sensitivity analyses were explored. The first was the impact of different probabilities of cancer (Fig. 2). In this analysis, both invasive and in situ cancer were considered and the ratio of invasive cancer to in situ cancer was assumed to be one to one. If the total cancer
TABLE 4: Financial Trade-Offs of Excisional Biopsy Compared with Stereotaxic Biopsy
End Point
To diagnosis and treatment at initial evaluation Per additional early invasive cancer detected (additional cost $804,000 per additional early detection = 6.7) Per additional early invasive and in situ cancer detected (additional cost $804,000 per additional early detection = 19.1) To Diagnosis and Treatment within 6 months Per additional early invasive cancer detected (additional cost $804,000 per additional detection within 6 months = 1.1) Per additional early invasive and in situ cancer detected (additional cost $804,000 per additional detection within 6 months = 3.7) Alternate surgical treatment assumptions If 50% of women have mastectomy for in situ cancer if 50% of women have mastectomy for invasive cancer, and if 25% of women need local reexcision after excisional biopsy To diagnosis and treatment at initial evaluation To diagnosis and treatment within 6 months aCosts incurred over 6 months. 356
Incremental CostEffectiveness per Woman a ($) 120,000
42,100
730,000
217,300
51,100 263,800
risk is 10% on the basis of initial mammography, an additional 1% of the w o m e n will have an earlier diagnosis of their cancer by excisional biopsy compared with stereotaxic biopsy at a cost of about $100,000 per additional early detected and treated invasive and in situ cancer. At higher risks, such as 30%, the number of early cancers diagnosed increases to 3% and the additional cost decreases to abou~ $24,000 per early detected and treated invasive and in situ cancer. The false-negative rate associated with stereotaxic biopsy is the other variable that is uncertain and may influence the outcome. Figure 3 shows the effect of alternate estimates for the sensitivity of stereotaxic biopsy. For this sensitivity analysis, the test characteristics for stereotaxic biopsy for invasive cancer and in situ cancer were varied in parallel such that the true-positive rates for invasive cancer were always .05 higher than for in sire cancer. For example, when the sensitivity of stereotaxic biopsy was assessed at .95 for invasive cancer, the sensitivity of stereotaxic biopsy would be .90 for in situ cancer. The best possible results would be a sensitivity for invasive cancer of .98 because indeterminate results were assumed to be constant at .02. As shown in the graph, as the estimated sensitivity of stereotaxic biopsy increases (fewer false-negatives), the advantage of excisional biopsy decreases. Using an arbitrary target goal of a 1% cancer "miss rate," the sensitivity of stereotaxic biopsy would have to be .944 or greater. If stereotaxic biopsy missed no cases of invasive cancer, then the cost savings compared with excisional biopsy was $240,000 per cancer detected and treated. It is well-known that using a stereotaxic biopsy may yield an in situ cancer but that w h e n subsequent excisional biopsy or definitive surgery is done, invasive cancer is found [21-23]. These patients would require an additional surgical procedure done after definitive excision--an axillary node dissection=-and their expectations about the surgery and prognosis would be altered. If 10% of the w o m e n in the stereotaxic biopsy strategy initially thought to have only in situ cancer subsequently were found to have invasive cancer, then ° 1% of all w o m e n would have had incorrect, premature reassurance and two separate surgical procedures instead of one. The financial consequences of the additional surgery, w h e n averaged over all women, was modest at only $15 per w o m a n if this occurred with 10% of invasive cancers. The initial assumption that all w o m e n w o u l d have a lumpectomy was an intentional bias that minimized the cost of the excisional biopsy. In Table 4, an alternate
Vol. 3, No. 4, April 1996 FIGURE. 2. Sensitivity analysis of the probability of cancer. The pr0babi!ity of cancer is shown on the xaxis by combining the probability of invasive cancer and in situ cancer (ductal carcinoma in situ [DCIS]). The baseline estimate is 20%; 10% invasive and 10% in situ cancer is shown by the arrow. The left vertical axis, white open circles, is the additional number of cases of cancers that initial excisional biopsy detects compared with stereotaxic biopsy. The right vertical axis, the closed squares, is the additional dollars spent in the initial excisional biopsy strategy per each early cancer case detected and treated.
BREAST
BIOPSY
DECISION
ANALYSIS
$250000
0.03
$200000 0.02"
$150000
Additional
Early Cancer Detection
Additional $ Per Early Detection and Treatment
$100000 0.01 $50000
0.00
0.0
i
I
0.1
0.2
$0 0.3
Probability of C a n c e r (50% Invasive, 50% DCIS)
0.03
$250000
$200000
Additional
0.02' $150000
Early Cancer Detection
Additional $ per Early Detection and Treatment
$100000 FIGURE 3. Sensitivity analysis of the true-positive rate of stereotaxic biopsy (SBB). The sensitivity of SBB for invasive cancer is shown on the x-axis. The sensitivity of SBB for in situ cancer is assumed to be .05 less than for invasive cancer. The baseline estimate, shown by the arrow, is a sensitivity of SBB for invasive cancer (.90) and for in situ cancer (.85). The left vertical axis, white open circles, is the additional number of c a s e s of cancers that excisional biopsy detects compared with SBB. The right vertical axis, the closed squares, is the additional dollars spent bythe initial excisional biopsy per each early cancer case detected and treated.
0.01' $50000
~ Baseline o.o0 0.86
combination of surgical management was explored for its financial effects w h e n more w o m e n chose-to have a mastectomy. Using this combination of additional surgery, the difference in the additional cost to total management per w o m a n increases $172, or $51,100 per additional case of early detection of invasive cancer and in situ cancer.
Natural History Sensitivity Analysis A speculative sensitivity analysis assessed the possible detrimental effect on prognosis of w o m e n whose inva-
,
,
0.90
0.94
$o 0.98
Sensitivity SBB
sive cancer diagnosis was delayed or missed in the stereotaxic biopsy strategy. Assumptior~.~ about changes in the natural history of w o m e n with invasive and in situ cancer were made, as outlined in Table 5, to assess the incremental cost-effectiveness of excisional biopsy compared with stereotaxic biopsy. If a delay in the detection of invasive cancer by 6 months and for in situ cancer by more than 6 months was associated with a change in breast cancer survival of these women, the analysis indicated that the average incremental survival advantage to each woman in the initial excisional biopsy strategy.was 357
HILLNER
Vol. 3, No. 4, April 1996
ET AL.
TABLE 5: Sensitivity Analysis if Prognosis Is Changed Because of Delayed Diagnosis End Point
Result
Incremental costs per w o m a n if initial excisional biopsy Incremental survival per w o m a n if initial excisional biopsy Cost-effectiveness ratio (excisional biopsy to stereotaxic biopsy) dollars per additional life year gained Threshold analysis of quality of life Amount of survival per year a w o m a n must be wllhng to forego to avoid having an unnecessary excisional biopsy to prefer stereotaxic biopsy
$786 0.005 years (2 days) $156,700
0.4 days
i
Assumptions of altered natural history if delayed cancer diagnoses are as follows: i l ) 60-year-old cohorts of women with probability of invasive cancer i0% and in situ cancer 10%. (2) If invasive cancer, initial lesion is 5-9 mm, which has a risk of developing metastatic disease of about 2.5% per year (5year survival of 95%) [19]. (3) If invasive cancer diagnosis is delayed for 6 months or more, then the invasive cancer increases to 10-15 mm. This change in size is associated with a 50% relative increase in risk of death from breast cancer (5-year survival of 90%). This increase in risk is associated with a 5-month reduction universal discounted life expectancy. (4) if in situ cancer diagnosis is delayed for 6 months, no change in survival. (5) If in situ cancer diagnosis is delayed beyond 6 months, 5% of women will develop invasive cancer within 2 years [48]. (6) Costs for late surgical management of delayed treatment are included. (7) Results are adjusted to the present value using a 5% discount. (8) A 60-year-old woman without breast cancer has an average discounted survival of 22.5 years.
an increased survival of 2 days at an incremental cost of $156,700 per additional life year gained compared with stereotaxic biopsy. This speculative analysis was extended to assess the threshold changes in the quality of life, scored as the amount of survival a w o m e n would be Willing to give up to each year of life, associated with a negative excisional biopsy scar. Given the small difference in survival between the strategies in this scenario, it was not surprising that the amount of survival per year that a w o m a n would have to be willing to forego to avoid an unnecessary excisional biopsy also was small. If a w o m a n were willing to forego 0.4 or more days per year of survival to avoid having an unnecessary surgical excision and its associated scar, then the stereotaxic biopsy strategy would be the preferred approach. DISCUSSION In this study, the trade-offs between the previous standard of care, excisional biopsy, and a new approach, stereotaxic biopsy, were compared in terms of surgical biopsies avoided, early cancers detected, and financial savings. This is an important question given the rapid growth in the use of stereotaxic biopsy and its advantages of being less "invasive," avoiding anesthesia risk, 358
causing minimal changes in breast cosmesis and mammographic appearance, and lowering the initial charge for the procedure. The appropriate baseline sensitivity of stereotaxic biopsy is problematic because few studies of stereotaxic biopsy have had complete excisional biopsy confirmation of the stereotaxic biopsy findings. Many case series have been reported, but the number of patients having a confirmatory exdsional biopsy is generally less than one third of all patients. The nonsurgical stereotaxic biopsy patient follow-up is incomplete and rarely extends beyond 12 months. Therefore, the first principle of correct evaluation of a new diagnostic test cannot be satisfied and estimates must be used. In addition, even these estimates may be suspect because it remains to be shown whether effectiveness, the test performance based on widespread community practice, will be equal to the literature reports from high-volume centers. These results can be considered from the patient outcome and financial perspectives. From the patients' perspective, the most encouraging finding is that the model estimated that 76% of the w o m e n would avoid any surgical breast procedure. The trade-off comparison markedly varied depending on the group considered. Should this be the w o m e n with the yet-to-be-identified cancers or all the w o m e n undergoing biopsy? Detection was delayed until 6-month follow-up for invasive cancer in about 1 in 18 w o m e n with invasive cancer, or 1 in 180 of the initial cohorts undergoing stereotaxic biopsy. When both in situ and invasive cancer were considered, the ratios decreased to one in 6.5 w o m e n with disease, or 1 in 65 of the total biopsied cohort. These two different ways of looking at the same results will likely lead to markedly different interpretations. A decision analysis model cannot state which approach is better, but it can facilitate dearer communication. In addition, the model projected a scenario wherein, although infrequent, because some cancers will be missed on initial stereotaxic biopsy and will not be detected at follow-up mammography, about 1 in 100 w o m e n with invasive cancer, or 1 in 1,000 of stereotaxic biopsied women, will have the detection of their cancer delayed beyond 6 months. The most interesting or problematic feature of this work is the lack of benchmarks to compare the trade-offs in exchanging excisional biopsy with stereotaxic biopsy. The trade-off of missing about 1% of invasive cancer to avoid 75 excisional breast biopsies cannot be judged. A purely financial perspective is heavily weighted in favor of stereotaxic biopsy. In this situation, the patient's preferences are critical but have never been reported, to our knowledge.
Vol. 3, No. 4, April 1996
This model is-comprehensive c o m p a r e d with other assessments of stereotaxic biopsy in that it considers the total costs of m a n a g e m e n t and trade-offs in effectiveness. Lindfors and Rosenquist [24], as part of a m a m m o g r a p h i c screening assessment, assessed the effect of using a stereotaxic biopsy over an excisional biopsy. In this analysis it was assumed that stereotaxic biopsy would be just as effective, only less expensive, than excisional biopsy. Liberman et al. [25] performed a detailed assessment of cost to diagnosis using stereotaxic biopsy c o m p a r e d with excisional biopsy but did not consider the cost to treatment or the possibility of false-negative stereotaxic biopsies. The preprocedure probability of cancer that was assumed greatly affected the model's results. Although most radiologists can identify low-risk cases [26] for which m a m m o g r a p h i c follow-up is appropriate, for higher risk abnormalities radiologists can differ, sometimes substantially, in their estimated risk of cancer and in their recommendations for m a n a g e m e n t [27]. It is likely that variations a m o n g radiologists' cancer risk estimates account for m a n y more delays in the detection of invasive cancer than is associated with falsenegative stereotaxic biopsies. I m p r o v e d consistency in probability estimates w o u l d improve this model and reduce geographic variation in the detection rates. At a low estimated risk of cancer, the model may not be useful. At probabilities of less than 2%, the appropriate comparison should be between a stereotaxic biopsy and 6-month mammographic follow-up. The low yield of excisionat biopsy is quantified with the model: At a 2% probability of invasive cancer, excisional biopsy would cost more than $500,000 per each additional invasive cancer detected and treated. Future expansions of the model are needed to address the relative advantages of stereotaxic biopsy and .mammogram follow-up for low-risk patients [8]. Although economic analyses of new treatment or clinical strategies have markedly increased over the past decade [12, 28], there are few examples in which the focus is on the decremental cost savings at expense of efficacy. In cardiology, a prominent example is an evaluation that showed that the use of a coronary care unit rather than an intermediate care unit for patients with a low risk of acute myocardial infarction saved $139,000 (expressed in 1980 U.S. dollars) per year of life saved. Eddy's [29] evaluation of the frequency of screening for cervical cancer with Papanicolaou smears was one that changed national recommendations because it showed that using a 3-year frequency of test-
B R E A S T B I O P S Y DECISION A N A L Y S I S
ing would achieve about 96% of the benefit of annual screening with about a two-thirds reduction in inconvenience, risk, and cost. The only other report related to breast cancer diagnosis that we k n o w of was an assessment of the role of fine-needle aspiration b i o p s y compared with excisional biopsy for palpable breast masses [30]. Layfield et al. [30] found that c o m p a r e d with the two-step excisional biopsy, fine-needle aspiration saved more than $700 per patient at the cost of a 0.1% decrease in 10-year overall survival. They concluded that this was a reasonable alternative. The consequences of a 6-month delay in diagnosis in terms of patient survival due to an increased risk of micrometastatic disease before detection is unknown. The growth pattern and metastatic risk over time are probably related as much (or more) to the biology of the individual tumor as its size. We are unaware of any estimates of h o w large the risk of metastatic spread could be or h o w it currently affects physicians' biopsy decisions. However, physician and patient decision making may be dominated by this uncertainty. There is a sizable literature on the increased risk of death due to a delay in diagnosis in w o m e n with breast symptoms or masses [31-35], and such delays are the primary reason for malpractice suits in breast cancer [36, 37]. Sensitivity analysis of a reasonable combination of assumptions about the consequences of a 6-month delay indicated that the additional benefit of excisional biopsy to all w o m e n was only 2 days in survival at a relatively high cost. Additional evaluation of thresholds or cross-points where the preferred strategies change indicated that a minimal change in quality of life associated with an unnecessary excisional biopsy scar would make stereotaxic biopsy the preferred strategy. Given the growth in screening mammogram, furore research may be able to refine the assumptions used in these speculative analyses. Given the present economic climate, the desire to control medical expenditures and the distancing of patients from paying directly for their health care, physicians and policymakers will be increasingly pushed to focus on decremental cost-effectiveness questionsr-The:most definitive answer to this question will likely come from the ongoing multi-institutional trial sponsored by the Radiologic Diagnostic Oncology Group 5 at the National Institutes of Health, where all patients are undergoing stereotaxic or sonographic core needle or aspiration biopsy, or a combination of the two biopsy procedures, and where all patients will have a minimum 2-year follow-up. From this model, if the rare false negatives with stereotaxic biopsy are not associated with a significant adverse 359
HILLNER
ET AL.
change in the outcomes for w o m e n whose diagnoses are delayed, then the financial cost savings projected herein are likely to be realized with little, if any, excess mortality. These estimates should aid communication between radiologists with primary care providers and managed care directors when discussing stereotaxic biopsy. REFERENCES 1. Harris JR, Lippman ME, Veronesi U, Willett W. Breast cancer. N Engl J Med 1992;327:319-328. 2. Eddy DM. Screening for breast cancer. Ann Intern Mad 1989;111:389-399. 3. Cyrlak D. Induced costs of low-cost screening mammography. Radiology 1988; 168:661-663. 4. Parker SH, Burbank F, Jackman RJ, et al. Percutaneous large-core breast biopsy: a multi-institutional study. Radiology 1994;193:359-364. 5. Royal HD. Technology assessment: scientific challenges. AJR 1994; 163:503-507. 6. Kent DL, Haynor DR, Longstreth WT Jr., Larson EB. The clinical efficacy of magnetic resonance imaging in neuroimaging. Ann Intern Med 1994;120:856-871. 7. Jaeschke R, Guyatt GH, Sackett DL. Users' guides to the medical literature: II1. How to use an article about a diagnostic test. B. What are the results and will they help me in caring for my patients? The EvidenceBased Medicine Working Group. JAm Med Assoc 1994;271:703-707. 8. Kopans DB. Caution on core. Radiology 1994;193:325-328. 9. Reid MC, Glynn RJ, Lachs MS, Feinstein AR. Use of methodological standards in diagnostic test research: getting better but still not good. J Am Med Assoc 1995;274:645-651. 10. Hillner BE, Smith TJ, Desch CE. Efficacy and cost-effectiveness of autolegous bone marrow transplantation in metastatic breast cancer: estimates using decision analysis while awaiting clinical trial results. J Am Mad Assoc 1992;267:2055-206t. 11. Laupacis A, Feeny D, Detsky AS, Tugwell PX. How attractive does a new technology have to be to warrant adoption and utilization? Tentative guidelines for using clinical and economic evaluations. Can MedAssoc J 1992;t46:473--48t. 12. Smith T J, Hillner BE, Desch CE. Efficacy and cost-effectiveness of cancer treatment: rational allocation of resources based on decision analysis. J Natl Cancer Inst 1993;85:1460-1474. 13. Harris JR, Morrow M, Bonadonna G. Cancer of the breast. In: DeVita VT, Hellman S, Rosenberg SA, eds. Cancec principles and practice of oncology. Philadelphia: Lippincott, 1994:1264-1332. 14. Knutzen AM, Gisvold JJ. Likelihood of malignant disease for various categories of mammographically detected, nonpalpable breast lesions. Mayo Clin Proc 1993;68:454-460. 15. Meyer JE, Eberlein T J, Stomper PC, Sonnenfeld MR. Biopsy of occult breast lesions: analysis of 1261 abnormalities. J Am Mad Assoc 1990;263:2341-2343. 16. Detsky AS, Naglie IG. A clinician's guide to cost-effectiveness analysis. Ann Intern Mad1990;113:147-154. 17. Petitti DB. Meta-analysis, decision analysis, and cost-effectiveness analysis. New York: Oxford University Press, 1994. 18. Smith T J, Hillner BE, Neighbors DM, McSorley PA, Le Chevalier T. Acomparative evaluation of the cost-effectiveness of three regimens for nonsmall-cell lung cancer: navelbine, navelbine plus cisplatin, and vindesine plus cisplatin. J Clin Onco11995;13:2166-2173. 19. Hillner BE, Smith TJ. Efficacy and cost effectiveness of adjuvant chemotherapy in women with node-negative breast cancer: a decision-analysis model. N Engl J Med 1991 ;324:160-168. 20. HealthCare Consultants of America. 1995 Physicians fee and coding guide. Augusta, GA: Author, 1995. 21. Liberman L, Dershaw DD, Rosen PP, et al. Stereotaxic core biopsy of breast carcinoma: accuracy at predicting invasion. Radiology 1995; 194:379-381.
360
Vol. 3, No. 4, April 1996 22. Gisvold J J, Goellner JR, Grant CS, et al. Breast biopsy: a comparative study of stereStaxically guided core and excisional techniques. A JR 1994; 162:815-820. 23. Dronkers DJ. Stereotaxic core biopsy of breast lesions. Radiology 1992; 183:631-634. 24. Lindfers KK, Rosenquist CJ. Needle core biopsy guided with mammography: a study of cost-effectiveness. Radiology 1994;190:217-222. 25. Liberman L, Fah MC, Dershaw DD, et al. Impact of stereotaxic core biopsy on cost of diagnosis. Radiology 1995; t95:633-637. 26. Sickles EA. Periodic mammographic follow-up of probably benign lesions: results in 3,184 consecutive cases. Radiology 1991 ;179:463-468. 27. Elmore JG, Wells CK, Lee CH, Howard DH, Feinstein AR. Variability in radiologists' interpretations of mammograms. N Engl J Med 1994; 331 : 1493-1499. 28. Kupersmith J, Holmes-Rovner M, Hogan A, Rovner D, Gardiner J. Costeffectiveness analysis in heart disease: II1. lschemia, congestive heart failure, and arrhythmias. Prog Cardiovasc Disease 1995;37:307-346. 29. Eddy DM. Screening for cervical cancer. Ann Intern Mad 1990; 113:214-226. 30. Layfield LJ, Chrischilles EA, Cohen MB, Bottles K. The palpable breast nodule: a cost-effectiveness analysis of alternate diagnostic approaches. Cancer 1993;72:1642-1651. 31. Wilkinson GS, Edgerton F, Wallace HJ, Reese P, Patterson J, Priore R. Delay, stage of disease and survival from breast cancer. J Chron Disease 1979;32:365-373. 32. Machiavelli M, Leone B, Romero A, et al. Relation between delay and survival in 596 patients with breast cancer. Oncology 1989;46:78-82. 33. Elwood JM, Moorehead WP. Delay in diagnosis and long-term survival in breast cancer. British Medical 1980; 1291-1294. 34. Charlson ME. Delay in the treatment of carcinoma of the breast. Surg Gynecol Obstet 1985;160:393-399. 35. Dohrmann PJ, Hughes ESR, McDermott F, Price A. Symptom duration, tumor staging and survival in patients with carcinoma of the breast. Surg Gynecol Obstet 1982; 154:707-710. 36. Mitnick JS, Vazquez MF, Plesser KP, Roses DE Breast cancer malpractice litigation in New York State. Radiology 1993;189:673-676. 37. Kern KA. Medicolegal analysis of the delayed diagnosis of cancer in 338 cases in the United States. Arch Surg 1994;129:397-394. 38. Morrow M. Management of nonpalpable breast lesions, tn: DeVita VT, Hellman S, Rosenberg SA, eds. PPO updates: cancer. Principles & practice of oncolog~ Philadelphia: Lippincott, 1990:1-11. 39. Morrow M. When can sterotactic core biopsy replace excisional biopsy: a clinical perspective. Breast Cancer Res Treat 1995;36:1-9. 40. Schmidt RA, Nishikawa RM. Digital screening mammography. ]n Rosenberg SA, ed. PPO updates: principles and practice of oncolog~ Philadelphia: Lippincott, 1994:1-16. 41. Specificity of screening in United Kingdom trial of early detection of breast cancer. Br Meal J 1992;304:346-349. 42. Sickles EA. Quality assurance: how to audit your own mammography practice. Radiol Clin North Am 1992;30:265-275. 43. Caines JS, McPhee MD, Kenok GP, Wright BA. Stereetaxic needle core biopsy of breast lesions using a regular mammographic table with an adaptable stereotaxic device. A JR 1994;163:317-321. 44. EIvecrog EL, Lechner MC, Nelson MT. Non-palpable breast lesions: correlation of stereotaxic large-core needle biopsy and surgical biopsy results. Radiology 1993;188:453-455. 45. Dewlatshahi K, Yaremko ML, Kluskens LF, Jokich PM. Nonpalpable breast lesions: findings of stereotaxic needle-core biopsy and fine-needle aspiration cytology. Radiology 1991 ;181:745-750. .,,,. 46. Parker SH, Lovin JD, Jobe WE, Burke BJ, Hopper KD, Yakes WE Nonpalpable breast lesions: stereotactic automated large-core biopsies. Radiology 1991 ;180:403-407. 47. Parker SH, Lovin JD, Jobe WE. Stereotactic breast biopsy with a biopsy gun. Radiology 1990; 176:741-747. 48. Fisher B, Costantino J, Redmond C, et al. Lumpectomy compared with lumpectomy and radiation therapy for the treatment of intraductal breast cancer. N Engl J Med 1993;328:1581-1586.