| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1
Division of Radiology, The Cleveland Clinic Foundation, 9500 Euclid Ave.,
Cleveland, OH 44195-5103.
2
Departments of Biostatistics and Epidemiology, The Cleveland Clinic
Foundation, Cleveland, OH 44195-5103.
3
Department of Cardiothoracic Surgery, The Cleveland Clinic Foundation,
Cleveland, OH 44195-5103.
4
Department of Hematology and Medical Oncology, The Cleveland Clinic
Foundation, Cleveland, OH 44195-5103.
Received May 21, 1999;
accepted after revision September 3, 1999.
Presented at the annual meeting of the Radiological Society of North
America, Chicago, November 1996.
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionAppendicesReferences |
|---|
MATERIALS AND METHODS. A decision-analysis model was used to compare the cost-effectiveness of nine strategies. Initial imaging included unenhanced CT using an adenoma or nonadenoma threshold of 0 or 10 H or in- and opposed-phase MR imaging. When initial imaging did not confirm an adenoma, CT-guided biopsy or subsequent imaging was performed. Medicare reimbursement was used as a surrogate of cost. Net costs were calculated as the difference in costs between two limbs of the decision tree. Net benefits were calculated as the difference between strategies and were calculated for life expectancy in years. MR imaging, CT, and biopsy accuracy, average life expectancy, and surgical mortality rates were based on the literature.
RESULTS. The base case analysis determined that the most cost-effective strategy was CT with an adenoma or nonadenoma threshold of 10 H followed by MR imaging, if necessary. CT with a threshold of 0 H followed by biopsy, if necessary, was the least costly. The incremental cost-effectiveness ratio between these two strategies was $16,370 per year of life gained.
CONCLUSION. Unenhanced CT using a 10 H threshold followed by MR imaging, if needed, was the most cost-effective strategy for evaluating an adrenal mass in a patient with newly diagnosed non-small cell lung cancer.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionAppendicesReferences |
|---|
Discriminating benign from malignant adrenal masses in patients with non-small cell carcinoma of the lung is essential because identification of distant metastatic disease precludes operative resection. Imaging-guided percutaneous biopsy can confirm metastatic disease, but is not without risk [3,4]. Several reports have described noninvasive methods of characterizing adrenal masses: unenhanced CT [5,6,7,8] and in- and opposed-phase MR imaging [9,10,11,12]. Both CT and MR imaging are highly specific, but they differ in their sensitivity. One previous study proposed an imaging algorithm incorporating both CT and MR imaging [13], but the study was not outcomes-based. A second study assessed the costs of MR imaging then biopsy, when needed, and of biopsy without initial imaging [14]. A third cost-effectiveness analysis [15] investigated CT, MR imaging, and 131I-6ß-iodomethylnorcholesterol scanning of the adrenal mass. To our knowledge, the cost-effectiveness of different imaging strategies, including percutaneous biopsy, in patients with an adrenal mass and newly diagnosed non-small cell carcinoma of the lung has not been studied. In this study, we constructed a decision-tree model to describe the treatment of patients with both newly diagnosed lung carcinoma and an adrenal mass identified on initial CT after IV contrast material was administered. The analysis was performed from a societal perspective. This model was analyzed to determine the cost-effectiveness of various CT and MR imaging strategies.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionAppendicesReferences |
|---|
Decision-Analysis Model
The model used is based on eight assumptions. First, unenhanced adrenal CT
and adrenal MR imaging were performed and interpreted in a standard fashion.
That is, attenuation in Hounsfield units or signal intensity loss from in- to
opposed-phase gradient-echo images, respectively, distinguished adenoma from
nonadenoma. Second, the diagnosis of non-small cell lung cancer was previously
established with a contrast-enhanced CT. Third, all patients with benign
adrenal lesions had resectable primary tumors. Fourth, patients were
asymptomatic, had pulmonary reserve and general health to tolerate
thoracotomy, and did not have clinically evident nodal disease involving
extrathoracic regions, vessels, or nerves. Fifth, patients who were deemed to
have unresectable adrenal metastasis did not undergo additional therapy.
Sixth, potential biopsy-related complications, such as pneumothorax,
hemorrhage, or needle-tract seeding, were not included in the analysis.
Seventh, cost and life expectancy data were accounted for the future
devaluation of both present benefits and dollars because more than 90% of
patients would be dead in 2 years
[16]. Eighth, the cost of
treating subsequently diagnosed metastatic disease was not included in the
analysis.
The decision tree (Fig. 1) begins with a choice between CT and MR imaging. The first test in the decision tree is unenhanced CT using 0 H as a threshold (strategies 1, 3), unenhanced CT using 10 H as a threshold (strategies 2, 4), or MR imaging (strategies 5-7) (Table 1). If the first test indicates an adenoma, then the patient undergoes thoracotomy, incurring operative risk, perhaps unnecessarily.
|
|
If the initial imaging strategy cannot reveal an adenoma, then there are two possibilities. The first involves a second noninvasive test (strategies 1, 2, 5, and 6). If the initial study was CT, then MR imaging is performed. If the initial study was MR imaging, then CT follows (both thresholds 0 H and 10 H are evaluated). If an adenoma is confirmed, then resection occurs. If adenoma is not confirmed by the second noninvasive test, then percutaneous CT-guided biopsy is performed. If metastasis is confirmed by biopsy, then no further treatment occurs. If no metastasis is found at biopsy, then the patient proceeds to thoracotomy. The second possibility (strategies 3, 4, and 7) bypasses a second noninvasive study and proceeds directly to CT-guided biopsy. Treatment, based on biopsy results, is the same as the first possibility. Finally, the default strategies of assuming the adrenal lesion is an adenoma or metastasis (without imaging or biopsy) and deciding whether to proceed directly to surgery without imaging are considered (strategies 8, 9).
The decision tree assumes that the adrenal mass was initially detected on a contrast-enhanced CT, but at some institutions chest CT is performed through the upper abdomen without contrast enhancement. The decision tree was modified and a supplementary analysis was performed to assess the cost-effectiveness of six different strategies in this circumstance. In this scenario, unenhanced CT would always be the initial test. There is no cost attributed to the initial CT in this analysis because it has occurred before the beginning of the decision tree. The strategies include using CT with 0 H or CT with 10 H as the initial test. If this test cannot diagnose adenoma, the patient undergoes MR imaging or biopsy. The last two strategies are the default strategies: assume that the mass is an adenoma and proceed to surgery or assume that the mass is a metastasis and terminate the workup and treatment.
Probabilities
Probabilities are listed in Table
2. The prevalence of adrenal metastasis from lung carcinoma at
presentation was derived from several large studies
[17,18,19]
and estimated to be 10%. The accuracy of the diagnostic tests and CT-guided
biopsy are derived from the literature. A summary of four
[5,6,7,8]
of the largest and most homogeneous studies of adrenal CT reported by Korobkin
et al. [8] provides the
sensitivity and specificity of CT. Pooled data from three large studies of
adrenal MR imaging [9,
11,
12] are the source of MR
imaging baseline values. Values for biopsy were derived from two large studies
[3,
20].
|
Costs
Medicare reimbursement was used as a surrogate for cost. Both professional
and technical reimbursements were included in the cost figures. Reimbursement
for CT, MR imaging, and CT-guided biopsy included the average reimbursement at
our hospital for billing available for 1998. The dollar value for CT-guided
biopsy includes reimbursement for cytopathologic evaluation of the specimen.
The thoracotomy reimbursement includes the average 1998 reimbursement for the
entire cost of the hospitalization.
Utilities
Mortality from thoracotomy is estimated to be 4%
[21,
22]. The life expectancy for
patients with resectable lung cancer following thoracotomy is estimated to be
4.45 years [21]. If these
patients are, inappropriately, not treated, their life expectancy falls to
2.64 years [21,
23]. Thus, the additional life
expectancy gained by surgery is 1.81 years. Untreated patients, receiving best
supportive care, with distant metastasis have average survival estimated at
0.35 years (range, 0.192-0.442 years)
[23,24,25].
Those who have adrenal metastasis and, inappropriately, undergo thoracotomy
are assumed to have no additional benefit over those who do not undergo
surgery.
Decision and Sensitivity Analysis
A base case analysis was performed, using baseline values from
Table 2, to calculate the cost
and average life expectancy of each of the strategies. A software program was
written to make these calculations.
A sensitivity analysis was then performed. In this type of analysis, uncertain baseline parameters are varied over predetermined ranges to identify how this uncertainty would affect the outcomes of the study [26]. The plausible ranges of baseline values are listed in Table 2. First, a one-way sensitivity analysis was performed for which each of the 19 baseline parameters was varied. Additionally, the assumption that CT and MR imaging performed in studies are conditionally independent was abandoned. The effect of various degrees of correlation between CT and MR imaging on the outcome of the analysis was studied. A two-way sensitivity analysis was then performed in which two baseline parameters were varied simultaneously to assess impact on results of the study. The two-way analysis varied the sensitivity of CT using 10 H and the cost of MR imaging.
In each analysis, the best test strategy was identified as follows: values were first assessed for simple dominance (i.e., is one strategy less costly and more effective than the others?). If the best test strategy could not be identified by simple dominance, then we computed the incremental cost-effectiveness ratio, defined as the increased cost per year gained ($test1 — $test2 / LE1 — LE2) where LE is life expectancy. If the incremental cost-effectiveness ratio exceeded $50,000, one benchmark used for medical interventions [27], we deemed the less expensive test the best strategy. If the ratio did not exceed $50,000, then we chose the more effective test.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionAppendicesReferences |
|---|
|
The one-way sensitivity analysis showed that strategy 2 is the best strategy over the whole range of values considered except in four instances (Appendix 1). First, if the sensitivity of MR imaging is greater than or equal to 0.94, or if the sensitivity of CT (10 H) is less than or equal to 0.62, or if the specificity of CT (10 H) is less than or equal to 0.91, then the preferred strategy is CT (0 H) followed by MR imaging, if needed (strategy 1). Second, if the cost of MR imaging is less than or equal to $211, or if the cost of CT is greater than or equal to $320, then the preferred strategy is MR imaging followed by CT (10 H), if needed (strategy 6). Third, if the cost of surgery is greater than or equal to $40,902, then the preferred strategy is either CT (0 H) followed by MR imaging or CT (0 H) followed by biopsy, depending on the cost of surgery. Fourth, if the pretest probability of an adenoma is 0.81-0.98, then CT (10 H) followed by MR imaging remains the best strategy (strategy 2). If the pretest probability is less than or equal to 0.8, then one of several other strategies is preferable, depending on the exact value. For the range of 0.23-0.80, then CT (0 H) followed by MR imaging is best (strategy 1). If the pretest probability is greater than or equal to 0.99, then proceeding directly to surgery is best (strategy 9).
The assumption of conditional independence (no correlation) between MR imaging and CT was eliminated to see how a positive correlation between the tests would affect the results. If conditional independence holds, then the sensitivity of either MR imaging or CT does not change from the baseline, regardless of the order of the examinations. If there is correlation between MR imaging and CT, then the sensitivity of the second test would change if the first could not diagnose and adenoma is negative. CT (10 H) followed by MR imaging remains the best test if conditional independence holds or if it does not hold and there is moderate correlation between CT and MR imaging, such that after a negative CT (10 H) the sensitivity of MR imaging is greater than or equal to 0.49, after a negative CT (0 H) the sensitivity of MR imaging is greater than or equal to 0.67, after a negative MR imaging study the sensitivity of CT (10 H) is greater than or equal to 0.30 and the sensitivity of CT (0 H) is greater than or equal to 0.20. If MR imaging and CT are highly correlated, then the best strategy depends on the exact correlation. For example, the best strategy is CT (0 H) followed by MR imaging when after a negative CT (10 H) the sensitivity of MR imaging is 0.46, after a negative CT (0 H) the sensitivity of MR imaging is 0.66, after a negative MR imaging study the sensitivity of CT (10 H) is 0.26, and after a negative MR imaging study the sensitivity of CT (0 H) is 0.16. At the extreme where the test results of MR imaging and CT are perfectly positively correlated, MR imaging followed by biopsy (strategy 7) is the preferred test strategy; it dominates CT (10 H) followed by MR imaging (i.e., both cheaper and better outcome). The second-best strategy is CT (10 H) followed by biopsy; the incremental cost-effectiveness ratio is $15,316 per year of life expectancy.
A two-way sensitivity analysis was performed involving the sensitivity of CT (10 H) and the cost of MR imaging. These results are summarized in Figure 2; the plotted line represents the threshold point at which one strategy is favored over another, using the principles of simple dominance and incremental cost effectiveness. Figure 2 indicates which strategy is most cost-effective for different values of the sensitivity of CT (10 H) and the cost of MR imaging. Over the majority of the ranges considered, CT (10 H) followed by MR imaging is favored; however, if the cost of MR imaging was very low or the sensitivity of CT (10 H) diminished, then other strategies would become more cost-effective.
|
In the circumstance of having an initial unenhanced CT available, the strategy of CT (10 H) followed by MR imaging, if indicated, remained the most cost-effective. This strategy is most effective, but not the least costly. The next most effective strategy was CT (0 H) followed by MR imaging. The incremental cost effectiveness ratio between the two was $16,370 per life-year gained. In decreasing order of cost-effectiveness, the remaining strategies were CT (10 H) followed by biopsy, CT (0 H) followed by biopsy, and the default strategy of assuming that the adrenal mass was a metastatsis.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionAppendicesReferences |
|---|
The decision analysis performed shows that using unenhanced helical CT with a threshold of 10 H, followed by MR imaging and biopsy, if necessary, (strategy 2), to distinguish adenoma from nonadenoma is the most cost-effective strategy. Although the life expectancy gained and cost saved are small in comparison with other strategies, strategy 2 is best over the majority of the parameters varied in the one-way sensitivity analysis.
A sensitivity analysis was performed to assess how uncertainty in the parameters used in the analysis could affect the results. There are four circumstances in which the uncertainty of our estimates could cause other strategies to become preferable. The first circumstance relates to the sensitivities of MR imaging and CT (10 H), and the specificity of CT (10 H). If these are altered enough, CT (0 H) followed by MR imaging becomes the preferable strategy. Each of these possibilities lies near the boundary of the range of plausible values (Table 2). The second circumstance relates to the relative costs of CT and MR imaging. If MR imaging becomes less costly compared with CT, then MR imaging becomes the preferred first test and MR imaging followed by CT (10 H) the best strategy. Because MR imaging is more sensitive than CT, only a disparity in cost between MR imaging and CT prevents it from being the preferred first test. The third circumstance relates to a very high cost of surgery. Should this occur, CT (0 H) followed by MR imaging or biopsy becomes preferable. The range of values considered for the cost of surgery was intentionally very broad because surgical costs are more difficult to estimate than other costs included in the analysis. The value at which the CT (0 H) strategies are preferred ($40,902) is almost twice the baseline value. This is the only situation when an imaging test followed directly by biopsy could become the most cost-effective strategy. Lastly, the pretest probability of adenoma has an impact on the best strategy. If the probability of adenoma remains in the range of 81-98%, then the results are unchanged. If it were to drop to 23-80%, CT (0 H) then MR imaging becomes the preferred strategy.
Abandoning the assumption of conditional independence also affects the results of the analysis. This is a standard assumption in this type of analysis, but previous studies have raised the possibility that it may not hold in this setting [28, 29]. These researchers reported that CT attenuation values and MR imaging chemical-shift ratios are highly correlated (correlation coefficient, 0.85 [28]) and both are correlated with the amount of histologic lipid in adenomas [29]. Thus, they are indeterminate in a similar subset of patients [28]. CT attenuation values and chemical-shift ratios are continuous variables and have a linear relationship; CT and MR imaging examinations are interpreted in a dichotomous (positive or negative) fashion. Thus, this correlation cannot be directly translated into an exact degree of correlation of the MR imaging and CT results; however, it does suggest that if the first study performed was unable to confirm an adrenal adenoma, then the sensitivity for the second imaging study may be lower than that in the general population. Our sensitivity analysis showed that if the correlation is moderate, then the baseline results are unchanged. If the tests are highly correlated, then several other strategies may be best, depending on the degree of correlation. A further study using serial CT and MR imaging is necessary to determine the true degree of correlation.
The analysis was also performed using the perspective of a hospital, which might be important in a setting such as the negotiation of a managed care contract (Appendix 2). This secondary analysis, using cost data particular to our institution, was performed to highlight the cost factors determining differential cost-effectiveness. However, the results of this supplemental analysis are the same as those using a societal perspective: CT (10 H) followed by MR imaging was the most cost-effective strategy. This result suggests that the particulars of the analysis and imaging tests override the differences in perspective relating to costs.
The two-way sensitivity analysis (Fig. 2) evaluated the strategies on the basis of the sensitivity of CT using a 10 H threshold and the cost of MR imaging. This analysis shows how changes in both parameters affect the determination of the most cost-effective test at any combination of values. CT (10 H) then MR imaging is favored over most of the range evaluated; however, at low sensitivities of CT (10 H) and low costs of MR imaging, other strategies become more cost-effective.
Several limitations with our approach exist. The decision tree has been intentionally simplified. Patients are assumed to have locally resectable disease, which is only true in approximately 73% of these patients [30]. This decision node was excluded from the analysis of mediastinoscopy, bronchoscopy, and thoracoscopy because of differences in the use, timing (intraoperative before thoracotomy or outpatient), and, thus, costs among surgeons and institutions.
We decided to exclude nonoperative therapy from the analysis. This was done for several reasons. The determination of distant metastatic disease, although essential to exclude operative resection in most patients, is insufficient to customize nonoperative therapy. Many centers choose to give local radiotherapy for symptomatic patients who have advanced local disease, but no randomized studies have proven an increased length of survival for asymptomatic patients [24]. Data on chemotherapy in patients with metastatic disease are conflicting [24]. Whether chemotherapy is cost-effective in metastatic non-small cell carcinoma of the lung has also been debated [31]. Attempting to add nonoperative therapy to the decision tree would add unmanageable complexity.
Delayed-enhanced CT has been evaluated to characterize adrenal masses. One study evaluated a 1-hr delay [32] and a second study evaluated an approximately 14-min delay [33]. Both studies found sensitivity and specificity of 95% or more; however, these results are preliminary and not included in this analysis.
Fluorodeoxyglucose positron emission tomography is increasingly used as an adjunct to CT for staging patients with newly diagnosed non-small cell carcinoma of the lung. It can accurately diagnose indeterminate adrenal masses [34, 35], but was not included in this analysis.
Biopsy complications were not factored into the analysis. Unfortunately, because the types of complications that occur are heterogeneous (pneumothorax, hemorrhage, seeding of a needle tract), we have found it impossible to predict accurate Medicare reimbursements for each of these. Furthermore, the first two would not be expected to affect our outcome measure (life-years), and the third would be very difficult to quantify. We believe that trying to quantify the costs and life-expectancy change from needle-tract seeding would add even more uncertainty to the model.
Finally, the patient or referring clinician may not wish to accept two imaging studies after an initial CT revealing the lung mass before proceeding to biopsy. Both time and psychosocial issues may prevail in a patient with a newly diagnosed malignancy. However, if the initial test is an unenhanced CT, then the number of noninvasive tests is reduced to two, which would possibly be more acceptible to the patient.
From a strict decision-analysis approach, one strategy is the most cost-effective, but the differences in cost-effectiveness are moderate. This is highlighted by the base case analysis, for which the incremental cost-effectiveness difference between the best strategy and the least costly is $16,370. This difference is less than the $19,000 median cost for a medical intervention [36]. One could argue that from a practical standpoint, other issues should be considered. For instance, local practice patterns, examination charges, or patient preference may impact these choices.
We conclude that, based on this model, CT with a 10-H threshold, followed by MR imaging and CT-guided biopsy, if needed, is the most cost-effective strategy to evaluate a patient with newly diagnosed non-small cell carcinoma of the lung and an adrenal mass. This strategy is best using both Medicare reimbursements as a proxy for cost and our actual hospital costs (Appendix 2) and when the decision tree is modified to incorporate an initial unenhanced CT. Future studies are needed to further evaluate MR imaging and CT because this strategy is best only when there is little to moderate correlation between them.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionAppendicesReferences |
|---|
|
|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionAppendicesReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  Y. Ohno, H. Koyama, Y. Onishi, D. Takenaka, M. Nogami, T. Yoshikawa, S. Matsumoto, Y. Kotani, and K. Sugimura Non-Small Cell Lung Cancer: Whole-Body MR Examination for M-Stage Assessment--Utility for Whole-Body Diffusion-weighted Imaging Compared with Integrated FDG PET/CT Radiology, August 1, 2008; 248(2): 643 - 654. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. Chong, K. S. Lee, H. Y. Kim, Y. K. Kim, B.-T. Kim, M. J. Chung, C. A Yi, and G. Y. Kwon Integrated PET-CT for the Characterization of Adrenal Gland Lesions in Cancer Patients: Diagnostic Efficacy and Interpretation Pitfalls RadioGraphics, November 1, 2006; 26(6): 1811 - 1824. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. Bunyaviroch and R. E. Coleman PET Evaluation of Lung Cancer J. Nucl. Med., March 1, 2006; 47(3): 451 - 469. [Full Text] [PDF]
|
|
|
|
 |
|
 C. C. Earle Outcomes Research in Lung Cancer J Natl Cancer Inst Monographs, October 1, 2004; 2004(33): 56 - 77. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. A. Haider, S. Ghai, K. Jhaveri, and G. Lockwood Chemical Shift MR Imaging of Hyperattenuating (>10 HU) Adrenal Masses: Does It Still Have a Role? Radiology, June 1, 2004; 231(3): 711 - 716. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. G. Pfister, D. H. Johnson, C. G. Azzoli, W. Sause, T. J. Smith, S. Baker Jr, J. Olak, D. Stover, J. R. Strawn, A. T. Turrisi, et al. American Society of Clinical Oncology Treatment of Unresectable Non-Small-Cell Lung Cancer Guideline: Update 2003 J. Clin. Oncol., January 15, 2004; 22(2): 330 - 353. [Full Text] [PDF]
|
|
|
|
 |
|
 S. G. Spiro and J. C. Porter Lung Cancer--Where Are We Today?: Current Advances in Staging and Nonsurgical Treatment Am. J. Respir. Crit. Care Med., November 1, 2002; 166(9): 1166 - 1196. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 N. Hollings and P. Shaw Diagnostic imaging of lung cancer Eur. Respir. J., April 1, 2002; 19(4): 722 - 742. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. R. Medverd and T. J. Dubinsky Cost Analysis Model: US versus Endometrial Biopsy in Evaluation of Peri- and Postmenopausal Abnormal Vaginal Bleeding Radiology, March 1, 2002; 222(3): 619 - 627. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
