Radiology
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Published online before print January 25, 2002, 10.1148/radiol.2223001822
This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
Right arrow All Versions of this Article:
2223001822v1
222/3/619    most recent
Right arrow Submit a response
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Medverd, J. R.
Right arrow Articles by Dubinsky, T. J.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Medverd, J. R.
Right arrow Articles by Dubinsky, T. J.
(Radiology 2002;222:619-627.)
© RSNA, 2002

Evidence-based Practice

Cost Analysis Model: US versus Endometrial Biopsy in Evaluation of Peri- and Postmenopausal Abnormal Vaginal Bleeding1

Jonathan R. Medverd, MD and Theodore J. Dubinsky, MD

1 From the Department of Radiology, University of Washington (J.R.M.); and the Department of Radiology, University of Washington Harborview Medical Center (T.J.D.), Box 359728, 325 Ninth Ave, Seattle, WA 98104-2499. From the 2000 RSNA scientific assembly. Received November 13, 2000; revision requested January 9, 2001; final revision received August 27; accepted September 19. Address correspondence to T.J.D. (e-mail: tdub@u.washington.edu).


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 APPENDIX A
 APPENDIX B
 REFERENCES
 
PURPOSE: To develop a cost minimization analysis model from the societal perspective of Medicare reimbursement to determine whether endometrial biopsy or transvaginal ultrasonography (US) is less expensive in evaluating peri- and postmenopausal women with abnormal vaginal bleeding and to assess whether this strategy is equally effective in populations at low and high risk for endometrial carcinoma.

MATERIALS AND METHODS: Clinical algorithms were constructed that detailed diagnostic evaluation of the target population by using office-based endometrial biopsy versus transvaginal US as starting points. An economic model based on Medicare reimbursement and average wholesale drug price data and using disease prevalences and modality sensitivities from the scientific literature was then created to examine common bleeding causes in this population. All models included the cost of obtaining a tissue diagnosis for focal or diffuse endometrial thickening found at US. Modality sensitivities and prevalences of disease states were varied within the model to discover limits at which each modality became cheaper versus the other for assessing a population of women.

RESULTS: Population prevalence of neoplastic disease is the principal factor governing total cost between competing diagnostic algorithms. In populations with 31% or less combined prevalence of endometrial carcinoma/atypical adenomatous hyperplasia, algorithms utilizing transvaginal US as the initial test are most cost minimizing. At combined endometrial carcinoma/atypical adenomatous hyperplasia prevalence of 10%, savings of up to 11% and 16% over pathways initiated with endometrial biopsy are predicted. In populations with a high incidence of neoplastic disease (>31%), biopsy-based algorithms should become least costly.

CONCLUSION: Transvaginal US–initiated triage predicts substantial cost savings versus biopsy-based algorithms in evaluating typical populations of peri- and postmenopausal women with abnormal vaginal bleeding seen in clinical practice.

© RSNA, 2002

Index terms: Economics, medical, 854.1261, 854.12989 • Uterus, biopsy, 854.1261 • Uterus, endometrium, 854.1261, 854.12989 • Uterus, US, 854.12989


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 APPENDIX A
 APPENDIX B
 REFERENCES
 
Postmenopausal vaginal bleeding is a common complaint, estimated as the reason for approximately 5% of all gynecologic visits (1–3). A majority of cases are caused by benign endometrial histologic factors. Endometrial carcinoma has been assumed to be the cause of bleeding in 10% or fewer of these women (4–6). Nonetheless, malignancy must be promptly excluded because abnormal vaginal bleeding is the presenting sign in 80%–85% of these cancers (7). Timely diagnosis of cancer is imperative, since therapy for local disease has higher associated disease-free survival and lower morbidity versus treatments for regional-stage disease (8,9). Between 70% and 80% of uterine cancers occur in postmenopausal women (8,10). Only 5% of women with endometrial cancer are younger than 40 years of age (8). Cost efficiencies applied to clinical evaluation of this commonly presented symptom predict substantial societal cost savings.

Office-based endometrial biopsy, transvaginal ultrasonography (US), and hysteroscopy have supplanted traditional dilation and curettage for initial evaluation of abnormal vaginal bleeding. More recently, saline-infusion hysterosonography also has been performed for office-based evaluation of these patients (11,12). Although endometrial biopsy is a sensitive and relatively inexpensive test for identifying endometrial atypical adenomatous hyperplasia and carcinoma, it is a poor test for diagnosing benign endometrial abnormalities such as atrophy, polyps, and submucosal fibroids, which are far more common causes of bleeding (4–6,13–15). US techniques have superior sensitivity in the identification of benign conditions (1,15,16). Hysteroscopy, while accurate in diagnosing nearly all conditions, is both invasive and expensive (17,18). Substantial cost savings and a potential reduction in patient morbidity would be predicted if benign findings could be reliably diagnosed in a minimally invasive cost-minimizing manner (4). However, authors of few studies (19,20) to date to our knowledge have rigorously evaluated cost efficiencies of competing diagnostic techniques.

The purpose of this study was to develop a cost minimization analysis from a societal perspective to assess the intermediate outcome of the total diagnostic cost of competing clinical pathways, by using office-based endometrial biopsy or transvaginal US as the initial diagnostic test in the evaluation of abnormal vaginal bleeding in peri- and postmenopausal women (21). The model incorporates reasonable ranges of disease prevalence, diagnostic procedure sensitivities, and Medicare reimbursement rates.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
 DISCUSSION
 APPENDIX A
 APPENDIX B
 REFERENCES
 
Six clinical algorithms detailing the diagnostic evaluation of the target population were constructed by using office-based endometrial biopsy versus transvaginal US as starting points (Appendix A). The diagnostic costs associated with each algorithm—inclusive of all appropriate outpatient facility, professional, pathology, and pharmacy services—based on Medicare reimbursement schedules and average wholesale price drug data were obtained (22,23) (Table 1).


View this table:
[in this window]
[in a new window]

  		TABLE 1. Diagnostic Costs Associated with Each Clinical Algorithm

 
Total costs for each algorithm were evaluated for each of the following common histologic categories in the target population: atypical adenomatous hyperplasia; carcinoma; polyps; submucosal fibroids; and atrophic, secretory, and proliferative endometrium. Endometrial carcinoma and atypical adenomatous hyperplasia were evaluated as a single category because US sensitivity and the usual treatment of either finding are similar. Disease prevalence and test sensitivity values for the detection of each condition were based on a review of the pertinent published literature (Table 2) (1,4–6,10–16,18,19, 24–49). It is notable that the reported prevalence of the examined causes accounting for abnormal vaginal bleeding has varied over time. The values selected attempt to reflect disease prevalence encountered in modern clinical practice (4,11,13,18,42,47–49). To account for differences in the literature, prevalence and sensitivity values were varied over reasonable ranges as part of this analysis.


View this table:
[in this window]
[in a new window]

  		TABLE 2. Prevalence of Endometrial Histologic Finding and Sensitivity of Diagnostic Method

 
An economic model to examine the cost associated with each diagnostic approach was then created by using a computer spreadsheet program (Excel; Microsoft, Redmond, Wash) and applying the reimbursement, prevalence, and sensitivity data to the clinical algorithms presented in Appendix A. All models include the cost of obtaining a tissue diagnosis for focal or diffuse endometrial thickening found at US examination. The cost of the initial clinic visit and definitive treatment were constant between models and, therefore, have been omitted. Costs related to ancillary patient expenses (eg, hygiene pads, lost income) and the full range of short and long-term patient outcomes were not examined. Hence, this is not a cost-effectiveness study (21).

Modeling assumptions were as follows:

1. Each subject had a single cause of endometrial bleeding.

2. The sample population included no women receiving HRT or tamoxifen therapy prior to presentation.

3. Tests were performed sequentially until diagnosis was made or the algorithm completed.

4. Cost values were based on one run through each model of the same hypothetical population of 1,000 patients.

5. Both diagnostic hysteroscopy and saline-infusion hysterosonography were assumed to be 95% sensitive for detecting endometrial conditions, except for endometrial carcinoma/typical adenomatous hyperplasia and secretory/proliferative categories. This analysis required tissue diagnosis for these histologic factors. To satisfy this demand and maintain consistency, sensitivity for these categories was set to that of endometrial biopsy, 85%.

6. Blind endometrial biopsy was performed at the conclusion of all office hysteroscopic examinations, as is common practice at our institution. Blind endometrial biopsy was also performed after any saline infusion hysterosonographic examination with findings suggestive of neoplastic disease.

7. No patients were lost to follow-up.

8. Follow-up clinic appointments assume Medicare level 2 encounters.

9. HRT trials represented 3 months of conjugated estrogens/medroxyprogesterone acetate tablet (Prempro; Wyeth-Ayerst, St. Davids, Penn) therapy.

10. HRT was 100% successful for all atrophic, secretory, and proliferative endometria and failed for polyps, submucosal fibroids, atypical adenomatous hyperplasia, and carcinoma.

11. There was a 10% false-positive rate for transvaginal US (43). This was in women whose endometrium appeared thick at transvaginal US but who were found to have secretory or proliferative endometrium at pathologic examination.

12. The diagnosis of a submucosal fibroid at transvaginal US was assumed to be 100% specific. Such women were assumed to undergo treatment without further diagnostic testing.

13. Formulas used to calculate cost according to algorithm and histology are presented in Appendix B.

Each model was evaluated for a range of endometrial carcinoma/atypical adenomatous hyperplasia prevalence to determine the prevalence at which biopsy-based algorithms would become more cost minimizing than US-based models. The sensitivity of biopsy was then varied within each model to discover limits at which each modality became cheaper versus the other for each histologic category. Reimbursement values were then varied to evaluate the performance of each model with respect to this variable.


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
DISCUSSION
 APPENDIX A
 APPENDIX B
 REFERENCES
 
The prevalence analysis for endometrial carcinoma/atypical adenomatous hyperplasia is shown in the Figure. In this exhibit, as the combined prevalence of carcinoma and atypical adenomatous hyperplasia is varied, the sum of the other disease prevalences is also changed in equal proportions for each disease entity (see Figure legend for example). At combined population prevalence rates of endometrial carcinoma/atypical adenomatous hyperplasia greater than 7% but less than 31%, an algorithm (model 6) beginning with transvaginal US followed by endometrial biopsy and then saline-infusion hysterosonography as necessary if the cause of bleeding remains uncertain is predicted to be most cost minimizing. If the prevalence of carcinoma/atypical adenomatous hyperplasia is less than 7%, substitution of saline-infusion hysterosonography for blind endometrial biopsy as the next test when a thickened endometrium is discovered at initial transvaginal US (model 4) is predicted to be cheapest. The combined population prevalence of endometrial carcinoma/atypical adenomatous hyperplasia would need to exceed 31% for an algorithm initiated with endometrial biopsy (model 2) to become more cost minimizing than US-based models.



View larger version (35K):
[in this window]
[in a new window]
[Download PPT slide]
  		Graph shows prevalence analysis data for endometrial carcinoma/atypical adenomatous hyperplasia. Total costs according to model at varying combined prevalence rates of endometrial carcinoma/atypical adenomatous hyperplasia are shown. Calculations assume that alternate histologic categories increase and decrease in equal proportion to neoplasia. (For example, if the carcinoma/atypical adenomatous hyperplasia prevalence is increased from 10% to 15%, the prevalence of each of the other four examined histologic categories is decreased by 1.25%.) Test sensitivities are fixed at values presented in Table 2 for this analysis. At an endometrial carcinoma/atypical adenomatous hyperplasia prevalence of 7% or less, model 4 is least costly (short solid arrow). For endometrial carcinoma/atypical adenomatous hyperplasia prevalence of 8%-31%, model 6 is cheapest. Both models utilize transvaginal US as the initial test. It is not until population prevalence for endometrial carcinoma/atypical adenomatous hyperplasia exceeds 31% (model 2, long solid arrow) or 35% (model 1, dotted arrow) that an algorithm beginning with endometrial biopsy becomes cost minimizing. {blacksquare} with solid line = model 1, {blacksquare} with dashed line = model 2, {blacktriangleup} with solid line = model 3, {blacktriangleup} with dashed line = model 4, {bullet} with solid line = model 5, and {bullet} with dashed line = model 6.

 
Table 3 displays cost according to model and histologic category. US-based algorithms are cost minimizing for detecting most benign endometrial disease—including atrophy, submucosal fibroids, and polyps—in the target population, because of superior sensitivity. Biopsy in these patient groups is frequently nondiagnostic, and either a therapeutic trial with HRT or another diagnostic procedure is undertaken in this circumstance. In contrast, endometrial biopsy–based models will always be more cost minimizing for diagnosing neoplastic disease. This is due to the similar sensitivities of biopsy and transvaginal US for detecting neoplasia and to the requirement of tissue acquisition to establish the diagnosis. The extra costs associated with tissue collection and analysis to confirm suspicious findings at US are not encountered with biopsy-initiated models. This cost behavior illustrates that the efficiency of each model largely depends on the prevalence of malignant disease in the population of women being evaluated.


View this table:
[in this window]
[in a new window]

  		TABLE 3. Cost according to Model and Histologic Category

 
Table 3 also illustrates the total cost for each model, as well as the savings garnered if model 6, the least costly model for a combined endometrial carcinoma/atypical adenomatous hyperplasia prevalence of 10%, were used instead. Cost savings of 11% and 16% over endometrial biopsy–initiated pathways (models 2 and 1, respectively) are predicted.

The last column in Table 3 details the predicted percentage of patients in the sample population who would remain undiagnosed at the conclusion of one run through each clinical algorithm. This number is calculated by summing the product of each histologic subpopulation size by the false-negative rates of each of the tests used to make a diagnosis.

Analysis was performed to assess the sensitivity necessary for an endometrial biopsy–initiated algorithm to become most cost minimizing versus transvaginal US–initiated pathways in pure populations of endometrial polyps, submucosal fibroids, and atrophy (Table 4). Transvaginal US sensitivity was held constant at the values presented in Table 2 for this analysis. Endometrial biopsy sensitivity analysis was not undertaken for carcinoma/atypical adenomatous hyperplasia and secretory/proliferative categories because biopsy is predicted to be cost minimizing for these histologic factors at presently observed performance values (as displayed in Table 3). Within models 1 and 2, the sensitivity of biopsy for detection of endometrial polyps would need to exceed approximately 40% and 34%, respectively, to be less costly than model 4, the most efficient algorithm for the diagnosis of pure polyp populations. Similarly, biopsy would need to be greater than 31% sensitive for the detection of atrophy to outperform US-initiated algorithms, a likely unattainable value with use of the current technique. Because of the substantially lower cost of transvaginal US versus endometrial biopsy and because the detection of a submucosal leiomyoma is an algorithm termination point, the sensitivity of endometrial biopsy for submucosal leiomyomas would need to be unrealistically high, greater than 90%, for endometrial biopsy-based algorithms to become more efficient than models 4 and 6 for evaluating this subpopulation.


View this table:
[in this window]
[in a new window]

  		TABLE 4. Endometrial Biopsy Sensitivity for Cost Minimization

 
Reimbursement variance analysis was performed (Table 5), with test sensitivity and histologic prevalence fixed at values presented in Table 2. The cost of transvaginal US or endometrial biopsy or both was incrementally increased or decreased until inflexion points at which biopsy-initiated models became cheapest were discovered. If transvaginal US reimbursement is held constant, endometrial biopsy reimbursement (inclusive of pathology components) would need to be 60% less for biopsy-based triage to become most cost minimizing. If reimbursement for transvaginal US increased by 34%, while that for endometrial biopsy decreased by 34%, biopsy-based triage would become most cost minimizing. Alternatively, if endometrial biopsy reimbursement were held constant, transvaginal US reimbursement could increase by 64% before a biopsy-based model became most cost minimizing.


View this table:
[in this window]
[in a new window]

  		TABLE 5. Reimbursement Variance Analysis Findings

 

   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
APPENDIX A
 APPENDIX B
 REFERENCES
 
Although a great deal of variability exists in the diagnostic evaluation of women with abnormal vaginal bleeding, our models are based on combinations of diagnostic examinations and the use of HRT trials. Our results predict that transvaginal US as the initial examination yields cost savings in comparison with biopsy-based models at prevalence rates of neoplastic disease typically encountered in modern clinical practice. We have also demonstrated the sensitivity and reimbursement requirements necessary for biopsy to become more cost minimizing than US. Savings achieved with US are primarily due to more efficient diagnosis of benign endometrial conditions. The sensitivity of endometrial biopsy for benign disease would need to be much higher than is commonly experienced to make biopsy-based models competitive. It is by no means clear that endometrial atrophy can be diagnosed with office-based endometrial biopsy, since pathologists often interpret tissue in these women as insufficient for diagnosis.

Endometrial biopsy is efficient for diagnosing neoplasia but relatively costly for identifying the benign conditions that account for a majority of peri- and postmenopausal women being evaluated for abnormal vaginal bleeding. For this reason, the combined population prevalence of endometrial carcinoma/atypical adenomatous hyperplasia must exceed 31% before a biopsy-based algorithm becomes cost minimizing versus US-based models in the evaluation of the heterogeneous patient population encountered in outpatient settings.

Population prevalence of the various causes of abnormal vaginal bleeding studied in this analysis has changed in the recent medical literature. For example, an approximate 20% prevalence for neoplastic disease has been established in the older scientific literature (5). In reality, this number is much higher than in the more recent experience of most investigators (46). Similarly, findings in newer literature (47,48) suggest smaller atrophy and greater polyp prevalence figures than do findings of studies of the not so distant past (3,5). Our analysis uses prevalence data that reflect the typical patient mix encountered in the modern outpatient environment. However, the model is flexible. Cost forecasts can be easily derived for any desired levels of reimbursement, test sensitivity, or histology prevalence by substituting values into the spreadsheet model.

An unexpected discovery of this analysis was the apparent undervalued status of outpatient diagnostic hysteroscopy versus saline-infusion hysterosonography within the Federal Register. As currently reimbursed by Medicare, insertion of a catheter for saline-infusion hysterosonography (only one portion of the examination) is more valuable than an entire diagnostic hysteroscopic examination (combined professional plus technical relative value unit, or RVU, components for "hysteroscopy, diagnostic," CPT code 58555, are 3.33 + 6.30 = 9.63; professional and technical RVU components for "catheter for sonohysterography," CPT code 58340, are 0.88 + 9.67 = 10.55).

We have created a cost minimization analysis from a societal perspective to determine which of several commonly practiced diagnostic algorithms is least costly. However, the perspective is not all encompassing and has really been limited to Medicare reimbursements, whereas other costs, including those accrued by the patients, those to industry for lost productivity, etcetera, which are beyond the scope of our study, have not been addressed. Our results demonstrate that, at current reimbursement levels and prescription prices, costs are largely dependent on the prevalence of endometrial carcinoma/atypical adenomatous hyperplasia within the study population. The analysis is not a "cost-effectiveness" model because costs associated with short-and long-term outcomes, including treatment procedures and complications, have not been included, and no attempt to measure treatment effectiveness has been made.

Our models assume that for any intermediate outcome, a given further diagnostic test will be performed until definitive imaging or tissue diagnosis is complete, or until the patient becomes asymptomatic by means of HRT. The number of women with atrophy treated with HRT, the number with carcinoma/atypical adenomatous hyperplasia that undergo hysterectomy, the number with polyps that undergo surgical hysteroscopy, and the number with submucosal fibroids that receive treatment should be in direct proportion to the prevalence of each histologic condition if each receives an accurate diagnosis. Numerous treatment options including embolization are available for fibroids, and evaluation of treatment costs for this cause will vary considerably among institutions.

In Table 3, it is predicted that the disease of 1.4%–2.4% of a population being evaluated with any of the models will remain undiagnosed at the conclusion of one run through at the test sensitivities and cause prevalences listed in Table 2. Similarity between algorithm pairs (models 1 and 2, 3 and 4, and 5 and 6) is due to the assumption of equal sensitivity of saline-infusion hysterosonography and hysteroscopy. The 1.4% rate for models 3 and 4 is nearly half the 2.2% and 2.4% rates for models 1 and 2 and 5 and 6, respectively, suggesting an increased diagnostic yield by proceeding directly to hysterography or saline-infusion hysterosonography after indeterminate or abnormal results are obtained at initial screening US. Such an approach would also be predicted to decrease time to diagnosis. If it may be assumed that time to treatment would also be minimized, then ancillary patient expenses (eg, hygiene pads, lost income) and other societal costs may be reduced. However, these variables were not studied in this analysis. A dedicated cost-effectiveness study would be required to examine these hypotheses (21).

This analysis predicts 10 per 1,000 more patients who are remaining undiagnosed after evaluation with model 5 or 6 versus models 3 or 4 in the same population. It is conceivable that a significant portion or all of the predicted $12,310 savings per 1,000 patients (in Table 3, total cost with model 4 minus total cost with model 6) could be potentially consumed in subsequent efforts to assign diagnoses to these women. Economic modeling of any additional evaluation is extremely difficult, since little consistency in approach is observed between practitioners (or from case to case by the same practitioner) for these challenging cases. Significantly, if these patients undergo hysterectomy for control of symptoms without undergoing additional work-up, as often occurs at our institution, no further diagnostic costs are incurred (since this would be considered a treatment cost), thereby negating discrepancy between algorithms. Our model predicts that differences in the number of women left undiagnosed after each algorithm converges to parity as saline-infusion hysterosonography and hysteroscopy sensitivity approaches 100%.

Complications and costs related to delay in diagnosis have not been assessed. However, it is clear that time to diagnosis will be shorter with US-based algorithms, since they provide same-day diagnosis for several benign conditions. Although some information regarding procedural complications can be inferred from the data, none have been formally compared with each other. Hence, a paucity of information exists on which to base such an analysis. It seems reasonable to infer that biopsy will have a higher complication rate than will US and that hysteroscopic procedures will have a higher complication rate than will saline-infusion hysterosonography. In reality, many patients with the worst outcomes, including anemia and hospitalization for severe bleeding, have not been evaluated by using any systematic algorithm. The evaluation of these severe long-term outcome costs would require a well-controlled prospective randomized study before any meaningful long-term outcome cost-effectiveness study could be performed.

No loss of patients to follow-up was assumed to avoid estimation bias and simplify the model. It is unclear from the literature whether there would be increased rates of failure to return to the clinic or noncompliance with medical recommendations between women initially evaluated with office-based biopsy versus transvaginal US (1,3,6,8,10,12–16, 19,25–27,31–34,37,39,41,43,46–48).

This analysis has been biased against US to some degree. We have set US sensitivities lower than many published in the literature (27,33,43,44) while using the more optimistic for biopsy, particularly those for benign endometrial disease (1). We have assumed equal sensitivities for saline-infusion hysterosonography and hysteroscopy, despite some publications (49) indicating that the sensitivity of saline-infusion hysterosonography is higher. Raising sensitivity values for transvaginal US and/or saline-infusion hysterosonography would further improve cost minimization versus biopsy-initiated pathways, as well as result in fewer nondiagnostic cases at the conclusion of an algorithm using these techniques.

Procedural charges vary between geographic regions and individual institutions within the same region. Accordingly, economic models built from local charge data can be difficult to meaningfully translate, either regionally or nationally. Medicare resource–based resource value unit and average wholesale drug price data serve as surrogates of true costs (50). We used these data to provide a reproducible analysis that predicts cost behavior nationally. By studying the effect that varying reimbursements have on the model, accounting for some regional variability can be inferred. Single-institution cost data can be substituted into the model to analyze local behavior.

US has its limitations. According to studies in the literature (44), in approximately 10% of peri- and postmenopausal women, a thick endometrium is seen at US, and either secretory or proliferative endometrium is shown at biopsy. We have considered these to be false-positive US findings at a rate of 10% to more accurately depict this limitation of US. Multiple uterine leiomyomata can obscure endometrial visualization at transvaginal US in 2%–3% of patients (37,51). These patients represent a subpopulation at relatively high risk for discovery of a uterine abnormality as the cause of bleeding (6,41). In our algorithms, women with equivocal US results advance to hysteroscopy, biopsy, or saline-infusion hysterosonography.

Although minimally invasive, the procedures we have examined are occasionally hampered by inability to access the vagina or uterus for anatomic or pathologic reasons. Cervical stenosis or unacceptable procedural pain can prevent performance of office-based endometrial biopsy, saline-infusion hysterosonography, or hysteroscopy 4%–10% of the time (13,26,33,35,43). The percentage of women with cervical stenosis would be the same for each algorithm. Most of these women would probably be referred to undergo US followed by one of the other procedures, biopsy or hysteroscopy, under local or general anesthesia. However, because practice patterns can vary in the evaluation of women with cervical stenosis, and because reimbursement and/or costs for surgical suite procedures are difficult to predict, a separate clinical assignment for these patients was not specifically studied in this analysis.

In conclusion, we have developed a cost minimization analysis demonstrating that at current endometrial carcinoma/atypical adenomatous hyperplasia prevalence and over a range of Medicare reimbursement levels and test sensitivities, societal cost savings are predicted with use of transvaginal US–based triage algorithms. We have not evaluated cost-effectiveness, since short- and long-term outcomes are not included in our model. Since, to our knowledge, few data with which to do so have been published, a prospective randomized study would be necessary to create such an evaluation.


   APPENDIX A
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 APPENDIX A
APPENDIX B
 REFERENCES
 
Analysis is based on six common diagnostic algorithms. These combine endometrial biopsy, HRT, transvaginal US, saline-infusion hysterosonography, and diagnostic office-based hysteroscopy.

In models 1 and 2 (Fig A1), endometrial biopsy is performed. If malignancy, atypical hyperplasia, or histology of another bleeding cause is found, the patient would proceed to treatment and leave the model. If biopsy findings are negative, however, then a 3-month HRT trial is undertaken. If the patient responds and bleeding resolves, she leaves the model. If HRT fails, then the patient would be further evaluated with diagnostic hysteroscopy in model 1 or with saline-infusion hysterosonography in model 2.



View larger version (14K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure A1. Models 1 and 2. See Appendix A for description. SIS = saline-infusion sonohysterography.

 
In models 3 and 4 (Fig A2), transvaginal US is performed as the initial step. If atrophy is found (homogeneous endometrium <5 mm thick), the patient receives 3 months of HRT and a level 2 follow-up clinic appointment to confirm resolution of bleeding. If a submucosal fibroid is detected, the patient goes on to other therapy. Both groups of patients drop out of the model at this point. If US findings are positive for an endometrium thicker than 5 mm or bleeding does not resolve after an HRT trial, the patient undergoes further evaluation with hysteroscopy in model 3 and with saline-infusion hysterosonography in model 4.



View larger version (15K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure A2. Models 3 and 4. See Appendix A for description. SIS = saline-infusion sonohysterography.

 
In models 5 and 6 (Fig A3), transvaginal US is also performed as the initial step. Patients with submucosal fibroids and atrophy drop out of the model, as in models 3 and 4. However, patients with thickened endometrium (>5 mm) undergo blind biopsy. If findings are positive for malignancy or another cause of bleeding, the patient drops out of the model. If blind biopsy findings are inconclusive or HRT trial fails, the patient is further evaluated with hysteroscopy in model 5 or with saline-infusion hysterosonography in model 6.



View larger version (18K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure A3. Models 5 and 6. See Appendix A for description. SIS = saline-infusion sonohysterography.

 
For all pathways, examinations or HRT trials continue until a diagnosis is established, bleeding stops, or the algorithm is completed.

Blind endometrial biopsy follows all hysteroscopic examinations in these models, as is the common practice at our institution. For models in which saline-infusion hysterosonography is used, subsequent blind endometrial biopsy is performed for only suspected endometrial carcinoma/atypical adenomatous hyperplasia or secretory/proliferative endometrium at imaging or to confirm an appearance of atrophy.


   APPENDIX B
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 APPENDIX A
 APPENDIX B
REFERENCES
 
The total cost for each model equals the sum of the examined costs for diagnosing each histologic cause of bleeding. Hence, total cost equals the sum of costs for atrophy, secretory/proliferative, submucosal fibroids, polyps, and carcinoma/atypical adenomatous hyperplasia. We calculated our costs on the basis of an arbitrary population size of 1,000 women. Therefore, the cost for each histologic cause equals 1,000 times the prevalence of that histologic cause, multiplied by the summed costs for that histologic cause, as described subsequently (B1–B4).



View larger version (26K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure B1. Formulas for polyp subpopulation. EB = reimbursement for office-based endometrial biopsy, FN = false-negative, HRT = average wholesale price for 3 months of HRT + reimbursement for Medicare level 2 follow-up clinic appointment, HS = reimbursement for office-based diagnostic hysteroscopy, SIH = reimbursement for office-based saline-infusion hysterosonography, Subpop = histologic subpopulation (ie, "atrophy," or "cancer/atypical adenomatous hyperplasia"), TP = true-positive, and TVUS = reimbursement for office-based transvaginal US.

 


View larger version (45K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure B2. Formulas for submucosal fibroid subpopulation. See Figure B1 for abbreviations.

 


View larger version (29K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure B3. Formulas for carcinoma and atypical adenomatous hyperplasia subpopulations. See Figure B1 for abbreviations.

 


View larger version (30K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure B4. Formulas for atrophy and secretory/proliferative subpopulations. See Figure B1 for abbreviations.

 
For each histologic cause, cost equals the sum of the number of patients undergoing each examination, multiplied by the cost of that examination. In this analysis, all women undergo the first test in each model, that is, accrued cost equals 1,000 multiplied by the population prevalence of the histologic cause in question, multiplied by the reimbursement value for the first test. Women who successfully receive a diagnosis drop out of the algorithm. Those who remain are the false-negative (1 - sensitivity) and false-positive results from the first test. Hence, the cost of women undergoing the second test or trial in each model equals 1,000 multiplied by the population prevalence of the histologic factor in question, multiplied by the rate of false-negative and false-positive findings from the first test, multiplied by the cost of the second test or trial. Again, women who successfully receive a diagnosis or treatment drop out, and those remaining represent the false-negative and false-positive results of the second test. Accordingly, the cost of the third test equals the number of women whose disease is undiagnosed from the second test, multiplied by the cost of the third test. Total cost equals the sum of the costs for all three tests or trials.

By varying disease prevalence, test sensitivity, or reimbursement values within the model, the effects on total cost can then be studied.


   ACKNOWLEDGMENTS
 
We appreciate the assistance of Donna Younes, BS, CPC, whose familiarity with Medicare reimbursement was invaluable.


   FOOTNOTES
 
See also the editorial by Bree in this issue.

Abbreviation: HRT = hormone replacement therapy

Author contributions: Guarantor of integrity of entire study, T.J.D.; study concepts and design, J.R.M., T.J.D.; literature research, J.R.M.; data acquisition, J.R.M.; data analysis/interpretation, J.R.M., T.J.D.; statistical analysis, J.R.M.; manuscript preparation, J.R.M.; manuscript definition of intellectual content, T.J.D.; manuscript editing, revision/review, and final version approval, J.R.M., T.J.D.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 APPENDIX A
 APPENDIX B
 REFERENCES
 

  1. Dubinsky TJ, Parvey HR, Maklad N. The role of transvaginal sonography and endometrial biopsy in the evaluation of peri- and postmenopausal bleeding. AJR Am J Roentgenol 1997; 169:145-149.
  2. Kazadi-Buanga J, Jurado-Chacon M. Etiologic study of 275 cases of endo-uterine hemorrhage by uterine curettage. Rev Fr Gynecol Obstet 1994; 89:129-133.
  3. Nasri MN, Coast GJ. Correlation of ultrasound findings and endometrial histopathology in postmenopausal women. Br J Obstet Gynaecol 1989; 96:1333-1338.
  4. O’Connell LP, Fries MH, Zeringue E, Brehm W. Triage of abnormal postmenopausal bleeding: a comparison of endometrial biopsy and transvaginal sonohysterography versus fractional curettage with hysteroscopy. Am J Obstet Gynecol 1998; 178:956-961.
  5. Holst J, Koskela O, von Schoultz B. Endometrial findings following curettage in 2018 women according to age and indications. Ann Chir Gynaecol 1983; 72:274-277.
  6. Karlsson B, Granberg S, Wikland M, et al. Transvaginal ultrasonography of the endometrium in women with postmenopausal bleeding: a Nordic multicenter study. Am J Obstet Gynecol 1995; 172:1488-1494.
  7. Fiorica JV, Hoffman MS, Roberts WS, et al. Detection of endometrial carcinoma: clinical judgement versus histologic examination. South Med J 1990; 83:759-760.
  8. Osmers RGW, Kuhn W. Endometrial cancer screening. Curr Opin Obstet Gynecol 1994; 6:75-79.
  9. Landis SH, Murray T, Bolden S, Wingo PA. Cancer statistics 1999. CA Cancer J Clin 1999; 49:8-31.
  10. Goldstein SR, Nachtigall M, Snyder JR, Nachtigall L. Endometrial assessment by vaginal ultrasonography before endometrial sampling in patients with postmenopausal bleeding. Am J Obstet Gynecol 1990; 163:119-123.
  11. Dubinsky TJ, Stroehlein K, Abu-Ghazzeh Y, et al. Prediction of benign and malignant endometrial disease: hysterosonographic-pathologic correlation. Radiology 1999; 210:393-397.
  12. Laughead MK, Stones LM. Clinical utility of saline solution infusion sonohysterography in a primary care obstetric-gynecologic practice. Am J Obstet Gynecol 1997; 176:1313-1318.
  13. Giusa-Chiferi MG, Gonçalves WJ, Baracat EC, et al. Transvaginal ultrasound, uterine biopsy and hysteroscopy for postmenopausal bleeding. Int J Gynecol Obstet 1996; 55:39-44.
  14. Buyuk E, Durmusoglu F, Erenus M, Karakoç B. Endometrial disease diagnosed by transvaginal ultrasound and dilation and curettage. Acta Obstet Gynecol Scand 1999; 78:419-422.
  15. Kavak Z, Ceyhan N, Pekin S. Combination of vaginal ultrasonography and Pipell sampling in the diagnosis of endometrial disease. Aust N Z J Obstet Gynaecol 1996; 36:63-66.
  16. Van den Bosch T, Vandendael A, Van Schoubroeck D, et al. Combining vaginal ultrasonography and office endometrial sampling in the diagnosis of endometrial disease in postmenopausal women. Obstet Gynecol 1995; 85:349-352.
  17. Goldstein SR. Use of ultrasonohysterography for triage of perimenopausal patients with unexplained uterine bleeding. Am J Obstet Gynecol 1994; 170:565-570.
  18. Widrich T, Bradley LD, Mitchinson AR, Collins RL. Comparison of saline infusion sonography with operative hysteroscopy for the evaluation of the endometrium. Am J Obstet Gynecol 1996; 174:1327-1334.
  19. Weber AM, Belinson JL, Bradley LD, Piedmonte MR. Vaginal ultrasonography versus endometrial biopsy in women with postmenopausal bleeding. Am J Obstet Gynecol 1997; 177:924-929.
  20. Blackmore CC, Magid DJ. Methodologic evaluation of the radiology cost-effectiveness literature. Radiology 1997; 203:87-91.
  21. Gold MR, Siegel JE, Russell LB, Weinstein MC. Cost effectiveness in health and medicine New York, NY: Oxford University Press, 1996; 61.
  22. Federal Register; 65:65517, 65539, 65555 2000.
  23. Drug Topics Red Book Montvale, NJ: Medical Economics, 1999; 506.
  24. Stovall TG, Photopulos GJ, Poston WM, et al. Pipelle endometrial sampling in patients with known endometrial carcinoma. Obstet Gynecol 1991; 77:954-956.
  25. Dorum A, Kristensen GB, Langebrekke A, et al. Evaluation of endometrial thickness measured by transvaginal ultrasound in women with postmenopausal bleeding. Acta Obstet Gynecol Scand 1993; 72:116-119.
  26. Moberger B, Nilsson S, Palmstierna S, et al. A multicenter study comparing two endometrial sampling devices: Medscand Endorette and Pipelle De Cornier. Acta Obstet Gynecol Scand 1998; 77:764-769.
  27. Bakour SH, Dwarakanath LS, Khan KS, et al. The diagnostic accuracy of ultrasound scan in predicting endometrial atypical adenomatous hyperplasia and cancer in postmenopausal bleeding. Acta Obstet Gynecol Scand 1999; 78:447-451.
  28. Cicinelli E, Romano F, Anastasio PS, et al. Sonohysterography versus hysteroscopy in the diagnosis of endouterine polyps. Gynecol Obstet Invest 1994; 38:266-271.
  29. Zorlu CG, Cobanoglu O, Isik AZ, et al. Accuracy of pipelle endometrial sampling in endometrial carcinoma. Gynecol Obstet Invest 1994; 38:272-275.
  30. Towbin NA, Gviazda IM, March CM. Operative hysteroscopy versus transvaginal ultrasonography in the evaluation of patients with excessive uterine bleeding. Am J Obstet Gynecol 1996; 174:1678-1682.
  31. Batool TA, Reginald PW, Hughes JH. Outpatient pipelle endometrial biopsy in the investigation of postmenopausal bleeding. Br J Obstet Gynaecol 1994; 101:545-546.
  32. Larson DM, Krawisz BR, Johnson KK, Broste SK. Comparison of the Z-Sampler and Novak endometrial biopsy instruments for in-office diagnosis of endometrial cancer. Gynecol Oncol 1994; 54:64-67.
  33. Kufahl J, Pedersen I, Eriksen PS, et al. Transvaginal ultrasound, endometrial cytology sampled by Gynoscann and histology obtained by Uterine Explora Curette compared to the histology of the uterine specimen. Acta Obstet Gynecol Scand 1997; 76:790-796.
  34. Goldchmit R, Katz Z, Blickstein I, et al. The accuracy of endometrial pipelle sampling with and without sonographic measurement of endometrial thickness. Obstet Gynecol 1993; 82:727-730.
  35. Dubinsky TJ, Parvey HR, Gormaz G, et al. Transvaginal saline assisted sonohysterography: comparison with biopsy in the evaluation of postmenopausal bleeding. J Ultrasound Med 1995; 14:887-893.
  36. Dijkuizen FPH, Brolmann HAM, Potters AE, et al. The accuracy of transvaginal ultrasonography in the diagnosis of endometrial abnormalities. Obstet Gynecol 1996; 87:345-349.
  37. Langer RD, Pierce JJ, O’Hanlan KA, et al. Transvaginal ultrasonography compared with endometrial biopsy for the detection of endometrial disease. N Engl J Med 1997; 337:1792-1798.
  38. Fistonic I, Hodek B, Klaric P, et al. Transvaginal sonographic assessment of premalignant and malignant changes in the endometrium in postmenopausal bleeding. J Clin Ultrasound 1997; 25:431-435.
  39. Indman PD. Abnormal uterine bleeding. J Reprod Med 1995; 40:545-548.
  40. Kekre AN, Jose R, Seshadri L. Transvaginal sonography of the endometrium in South Indian postmenopausal women. Aust N Z J Obstet Gynaecol 1997; 37:449-451.
  41. Gordon SJ, Westgate J. The incidence and management of failed pipelle sampling in a general outpatient clinic. Aust N Z J Obstet Gynaecol 1999; 39:115-118.
  42. Alcazar JL, Laparte C. comparative study of transvaginal ultrasonography and hysteroscopy in postmenopausal bleeding. Gynecol Obstet Invest 1996; 41:47-49.
  43. Tsuda H, Kawabata M, Yamamoto K, et al. Prospective study to compare endometrial cytology and transvaginal ultrasonography for identification of endometrial malignancies. Gynecol Oncol 1997; 65:383-386.
  44. Smith-Bindman R, Kerlikowske K, Feldstein VA, et al. Transvaginal ultrasound to exclude endometrial cancer and other endometrial abnormalities. JAMA 1998; 280:1510-1517.
  45. Loverro G, Bettocchi S, Cormio G, et al. Diagnostic accuracy of hysteroscopy in endometrial atypical adenomatous hyperplasia. Maturitas 1996; 25:187-191.
  46. Kramer DJ, Phinney AJ, Hashimoto BE, Cook MJ. How often can transvaginal ultrasound replace endometrial biopsy in the evaluation of the postmenopausal endometrium?. J Ultrasound Med 2000; 19:27.
  47. Bree RL, Bowerman RA, Bohn-Velez M, et al. Ultrasound evaluation of the uterus in patients with postmenopausal bleeding: a positive impact on diagnostic decision making. Radiology 2000; 216:260-264.
  48. Bronz T, Suter T, Rusca T. The value of transvaginal sonography with and without saline instillation in the diagnosis of uterine pathology in pre- and postmenopausal women with abnormal bleeding or suspect sonographic findings. Ultrasound Obstet Gynecol 1997; 9:53-58.
  49. Saidi MH, Sadler RK, Theis VD, Akright BD, Farhart SA, Villanueva GR. Comparison of sonography, sonohysterography, and hysteroscopy for evaluation of abnormal uterine bleeding. J Ultrasound Med 1997; 16:587-591.
  50. Remer EM, Obuchowski N, Ellis JD, et al. Adrenal mass evaluation in patients with lung carcinoma: a cost-effectiveness analysis. AJR Am J Roentgenol 2000; 174:1033-1039.
  51. Granberg S. Sonography of the endometrium in the post-menopausal woman. Ann Med 1994; 26:81-83.



This article has been cited by other articles:


Home page
 Am. J. Roentgenol.Home page
S.-H. Shen, Y.-Y. Chiou, J.-H. Wang, M.-S. Yen, R.-C. Lee, C.-R. Lai, and C.-Y. Chang
Diffusion-Weighted Single-Shot Echo-Planar Imaging with Parallel Technique in Assessment of Endometrial Cancer
Am. J. Roentgenol., February 1, 2008; 190(2): 481 - 488.
[Abstract] [Full Text] [PDF]

Home page
 RadiologyHome page
R. L. Bree and R. C. Carlos
US for Postmenopausal Bleeding: Consensus Development and Patient-centered Outcomes
Radiology, March 1, 2002; 222(3): 595 - 598.
[Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
Right arrow All Versions of this Article:
2223001822v1
222/3/619    most recent
Right arrow Submit a response
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Medverd, J. R.
Right arrow Articles by Dubinsky, T. J.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Medverd, J. R.
Right arrow Articles by Dubinsky, T. J.

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
RADIOLOGY RADIOGRAPHICS RSNA JOURNALS ONLINE