Assessment of the Cost of Fine-Needle Aspiration ... - Semantic Scholar

Anatomic Pathology / Cost of Diagnosis of Thyroid Nodules

Assessment of the Cost of Fine-Needle Aspiration Cytology as a Diagnostic Tool in Patients With Thyroid Nodules Isabelle Borget, MD,1 Philippe Vielh, MD, PhD,2 Sophie Leboulleux, MD,3 Mark Allyn,4 Sylviane Iacobelli,1 Martin Schlumberger, MD, PhD,3 and Gérard de Pouvourville, PhD1,5 Key Words: Fine-needle aspiration cytology; Cost; Thyroid nodules; Diagnosis DOI: 10.1309/H86KM785Q9KBWPW5

Abstract Fine-needle aspiration cytology (FNAC) is the reference diagnostic tool in patients with thyroid nodules. Because the true diagnosis is based on histopathologic findings, the final diagnosis of nodules not operated on is postponed, impacting the cost. We aimed to determine the cost of FNAC by taking account of diagnostic mistakes, failures, and follow-up of patients who did not have surgery. A Markov model described the management of patients according to initial cytopathologic results. Estimates for accuracy values and follow-up were derived from a retrospective study of 624 patients. Costs were computed from the hospital perspective. Sensitivity analyses were performed. Of the lesions, 381 were cytopathologically classified as benign, 15 as malignant, 57 as “suspicious,” and 171 as unsatisfactory. The cost of diagnosis was 1,145 euros (z). Cost was significantly affected by the unsatisfactory specimen percentage (27.4%), without which the cost would be reduced by 35.5%, to z738. Cost depends on cytopathologist performance and the unsatisfactory rate. In the future, routine ultrasound guidance and on-site assessment of cytopathologic adequacy would help reduce costs.

© American Society for Clinical Pathology

The prevalence of thyroid nodules is estimated at 4% to 7% in the general population.1 Because they are asymptomatic, most nodules are found incidentally. Only 5% are malignant,1,2 and the challenge for physicians is to identify the few malignant nodules among a large number of benign nodules. Fine-needle aspiration cytology (FNAC) is widely recommended for diagnosis of thyroid nodules2-4 because it is simple and safe. Cytopathologic results may be classified into 4 categories: benign, malignant, “suspicious,” and unsatisfactory. Surgery is performed when FNAC proves the lesion to be malignant or suspicious or when the nodule is large or causes compression symptoms. Histopathologic results permit estimation of the diagnostic accuracy of the method. In a review of 18,183 FNAC cases from 7 large series,5 the mean sensitivity and specificity of the procedure were 83% and 92%, respectively, with low falsepositive (mean, 3%) and false-negative (mean, 5%) results. The diagnostic accuracy of the method6-8 may be improved by ultrasound guidance (US-FNAC). Moreover, FNAC is a costeffective diagnostic method9-14 compared with scintigraphy and biopsy, and this was studied in patients who underwent operation.5 FNAC generated a lower unit cost, significantly reduced medical costs by avoiding surgery in 25% of patients with benign thyroid disease (true-negative results), increased the percentage of malignant lesions found at surgery (from 14% to 39%), and shortened the postponement of treatment for thyroid cancer. An extensive body of literature exists on the diagnostic accuracy and the cost of FNAC in patients who undergo operation,5,15-17 but the management of patients who do not has rarely been taken into account. Patients who do not undergo operation are usually followed up with a clinical

763

Am J Clin Pathol 2008;129:763-771 DOI: 10.1309/H86KM785Q9KBWPW5

763 763

Borget et al / Cost of Diagnosis of Thyroid Nodules

examination, ultrasound examination of the neck, or repeated FNAC, until the final diagnosis. Unsatisfactory specimens represent 10% to 30% of the prepared smears and lead to repeated FNAC procedures and delayed diagnoses. These additional procedures influence the cost, and no study has yet evaluated the cost of follow-up of patients who do not undergo operation. The present study was, therefore, conducted to assess the cost of FNAC as a diagnostic tool for patients with thyroid nodules. We developed a Markov model that takes into account the accuracy of the procedure and unsatisfactory specimens and the follow-up of patients who do not undergo surgery.

Materials and Methods Decision Model Structure and Assumptions Determining the cost of a true (final) diagnosis was accomplished by using a decision-analytic model. We developed a Markov model, ie, a multistate transient model in which patients make transitions among various collectively exhaustive and mutually exclusive health states at different rates for extended periods (cycle).18 The aim of the model is to describe the management and the care process of patients with thyroid nodules according to initial cytopathologic results. Each clinically important event for the diagnosis, such as a repeated FNAC, thyroid surgery, or a clinical examination, is modeled as a transition from one state to another. Each Markov cycle corresponds to a follow-up visit. In the decision-analysis model, the first decision node for a given patient is the first FNAC result: patients are then distributed into 4 transient states according to cytopathologic results, ie, malignant, benign, suspicious, and unsatisfactory. If a patient undergoes thyroid surgery, the histologic result provides the final diagnosis (benign or malignant) that is considered the “gold standard,” and the patient makes a transition into 1 of 4 absorbent states. These 4 absorbent states are defined, corresponding to the cytopathologic result, when the histopathologic diagnosis is known, ie, true-positive, true-negative, false-positive, and false-negative. When a second FNAC is performed in a patient who does not undergo surgery, the patient makes a transition into the state corresponding to the cytopathologic result of the second FNAC. The patient who has only a clinical examination remains in the same transient state. The simulation is then run for a hypothetical cohort of patients, up to 3 cycles. The diagnosis is defined as true (or final) when the status of the nodule can definitively be classified as benign or malignant by histologic examination for patients who undergo operation or when patients who do not undergo surgery remain in the same transient state during 3 cycles. At the end of these 3 cycles, the 764 764


assumption is made that all patients remaining in the suspicious or unsatisfactory transient states will end up in an absorbent state following surgery. Data We derived estimates for diagnostic variables and accuracy values from a retrospective study reviewing all consecutive patients who underwent a first thyroid FNAC examination between January 2003 and December 2005 at the Institut Gustave Roussy (IGR), Villejuif, France. Data on patient characteristics (age, sex, nodule size, number and type of nodules) and surgical and follow-up procedures were extracted from the hospital computerized medical record. Data concerning FNAC examination (date, name of practitioner, under guidance or not) and cytopathologic (coded into 1 of the following 4 categories: malignant, benign, suspicious, and unsatisfactory) and histologic results were extracted from the database of the pathology department at IGR. Data on the follow-up of patients, repeated FNAC, and clinical examinations performed for patients who did not undergo surgery were also collected until June 30, 2006. Cytopathology According to Faquin,19 lesions derived from follicular cells may be classified as follows: (1) benign when aspirates are hypocellular to moderately cellular with moderate to abundant colloid and follicular cells have round nuclei of uniform size; (2) suspicious when aspirates suggest a follicular neoplasm, ie, hypercellular sample with scant colloid and a significant proportion of microfollicules, trabeculae, or crowded overlapping clusters of follicular cells (also includes lesions consisting of oncocytic [Hürthle cell] neoplasms); (3) malignant when cellular features of papillary, insular, undifferentiated, or medullary thyroid carcinoma are present; and (4) unsatisfactory for cytopathologic examination. Costs Costs were full costs estimated from the hospital perspective, which in France, corresponded to the payer’s perspective. A microcosting study was performed to determine the unit cost of FNAC examination at IGR. It included costs for all visits with a clinician and for performing the FNAC examination and processing and interpreting the sample. Resources included medical devices and supplies, personnel costs, and the depreciation cost of medical equipment, plus overhead costs. They were recorded at IGR via a prospective study of 167 consecutive thyroid FNAC examinations performed between January and October 2005. The time taken by each category of personnel (clinician, radiologist, technician, secretary, and cytopathologist) to perform the FNAC examination and process and interpret the sample was noted. The unit acquisition prices for consumables were based on 2005 rates © American Society for Clinical Pathology

Anatomic Pathology / Original Article

expressed in euros (z). Personnel costs were determined by using the average hourly wage of each professional category at IGR. The cost of a consultation was obtained by multiplying the mean duration of the consultation by the average hourly wage of a consultant at IGR. The cost of the hospital stay for a thyroidectomy was extracted from the hospital cost accounting system. It included direct medical costs (medical supplies, laboratory tests, and radiology) and overhead costs and represents actual costs. Because the study aimed at computing the cost of the diagnosis alone (rather than the cost of thyroid cancer treatment), we decided to use the following rules for attributing the cost of surgery. For patients who had malignant or suspicious cytologic findings and ultimately had a malignant tumor diagnosed by histologic examination (true-positive results), surgery was the treatment of choice of the carcinoma and its cost was, therefore, not considered a cost of the diagnosis. In patients with benign cytopathologic findings who had undergone surgery to relieve compression symptoms or for aesthetic reasons but not to confirm the cytopathologic diagnosis, the cost of surgery was not attributed to the failure of the FNAC procedure. For patients with an unsatisfactory FNAC specimen, the decision to treat the patient included the need to know the nature of the nodule (certainties of diagnosis) and the need for care (owing to a large nodule and/or compression symptoms). Because it is difficult to attribute the cost to one or another, in the basal model, the cost of surgery was applied to all patients with unsatisfactory cytopathologic results who had undergone surgery, whatever the final histopathologic result. Based on this critical assumption, a sensitivity analysis was performed. We considered that during the first 2 cycles of the model, surgery was performed for reasons unrelated to diagnosis (to relieve large nodules or for cosmetic reasons). Conversely, during the last (or third) cycle, FNAC failed to provide a diagnosis and the specimen was assessed to know the nature of the nodule (production of knowledge). Under this assumption, the cost of surgery was applied only for the last cycle. Indeed, a complementary analysis was performed to incorporate into the model the indirect costs related to day losses in active patients (ie, employed patients or those seeking employment), from the societal perspective. In France, loss of income is partially compensated by the National Sickness Fund, and the difference is the responsibility of the patient. Thus, by adding total hospital cost and sick leave costs, the total budgetary impact of the diagnosis was estimated. The estimation of the duration of sick leave was extracted from a study in process at our center, which involved the evaluation of the length and the cost of sick leave in patients with thyroid cancer treated by total thyroidectomy and radioiodine ablation. Data on sick leave duration was collected in active patients from the week before surgery to 3 months after radioiodine ablation. The value of 1 day of sick leave © American Society for Clinical Pathology

compensation was estimated at z27. Patients’ related costs represented uncompensated loss of income, and they were computed for active patients. In France, the daily amount can be calculated as the difference of the mean net income per day minus daily compensation received. Statistics and Sensitivity Analysis Cases that had been cytopathologically classified as malignant or suspicious and finally diagnosed as malignant at histologic analysis were considered true-positives. Cases with benign or unsatisfactory cytopathologic results and with a benign histologic confirmation or a benign clinical course were considered true-negatives. The false-positive category comprised cases with suspicious or malignant cytologic results and benign histologic findings, whereas false-negative diagnoses were cases with benign or unsatisfactory cytopathologic findings and malignant histologic findings. We performed several 1-way sensitivity analyses for key model parameters, including the rate of initial unsatisfactory cytologic results, the rate of malignancy among suspicious cases, and the rates of false-negative and false-positive results. To test the impact of a lower rate of unsatisfactory specimens on the cost of a true diagnosis, we assumed that the same proportion of patients would be distributed between the 3 other transient states (malignant, benign, and suspicious) as in our retrospective study. If Ti0 was the initial rate of unsatisfactory results and Ti1 a lower rate we wanted to test, then Ti1 could be defined as Ti1 = Ti0 × k, with 0 < k ≤ 1. The difference (Ti0 – Ti1) was distributed among the 3 other transient cytopathologic states. The corresponding new rate of malignant (Tm1), benign (Tb1), and suspicious (Ts1) was calculated as follows: Tm1 = Tm0 + [Ti0 × Tm0 × (1 – k)/(Tb0 + Tm0 + Ts0)] Tb1 = Tb0 + [Ti0 × Tb0 × (1 – k)/(Tb0 + Tm0 + Ts0)] Ts1 = Ts0 + [Ti0 × Ts0 × (1 – k)/(Tb0 + Tm0 + Ts0)] Retrospective and prospective data were collected on Epi Info software (version 3.3.2) (Centers for Disease Control and Prevention, Atlanta, GA). The model was developed using decision-analysis software from TreeAge Data 4.0 Software (TreeAge Software, Williamstown, MA).

Results Primary Data Sources and Identification of Baseline and Transition Probabilities Between January 2003 and December 2005, 624 patients with thyroid nodules underwent FNAC examination. The mean ± SD age of patients was 48.6 (± 14.8) years, and 508 (81.4%) were female. The mean ± SD diameter of the largest nodule was 26.8 (± 11.5) mm. Each FNAC sample was processed by 1 of 6 practitioners (4 endocrinologists and 2 surgeons), but

765


765 765


none was performed in the presence of a cytopathologist or in a 1-stop clinic. FNAC examination was initially performed under ultrasound guidance in 186 patients (29.8%) with a smaller nodule (22 vs 29 mm for classic FNAC). Baseline Data The model and its initial transition probabilities are given in zFigure 1z. The first node of the model for each patient corresponds to the results of the first FNAC, which were malignant in 15 cases (2.4%), benign in 381 (61.1%), suspicious in 57 (9.1%), and unsatisfactory in 171 (27.4%). These rates represented the initial probabilities. The 15 patients with a malignant FNAC result underwent surgery, and all lesions were diagnosed as malignant by histologic examination. Consequently, all of these patients made a transition into the absorbent state called true-positive in the model. Of the 381 patients with a benign FNAC result, 93 (24.4%) underwent surgery because of compression symptoms or a recent increase in nodule size. At histologic examination, 87 specimens (94%) were benign (true-negative) and 6 (6%) were malignant (false-negative). Follow-up of the 288 patients who did not undergo surgery mainly involved clinical examination (204 [70.8%]) or a repeated FNAC examination (84 [29.2%]). The results of the second FNAC examination were as follows: malignant, 0 (0%); benign, 64 (76%), suspicious, 6 (7%); and unsatisfactory specimen, 14 (17%). Patients were then assumed to transit to the state corresponding to the results of this second FNAC examination until a new diagnostic event occurred. The first FNAC result was suspicious in 57 patients who underwent surgery (46 [81%]) or were followed up (11 [19%]). Among the 46 patients who underwent surgery, 16 (35%) had a malignant tumor diagnosed by histologic examination and, thus, made a transition to the true-positive state, whereas the other 30 (65%) patients with benign histologic findings made a transition into the false-positive state. Follow-up of patients who did not undergo surgery involved a clinical examination, and these patients remained in the state called suspicious. Of the 171 patients with initially unsatisfactory cytopathologic specimens, 128 (74.9%) had a subsequent clinical examination (47 [36.7%]) or another FNAC examination (81 [63.3%]) with the following results: benign, 48 (59%); suspicious, 4 (5%); or unsatisfactory specimen, 29 (36%); none were malignant. Patients were then assumed to transit to a state corresponding to the results of this second FNAC examination. The other 43 patients underwent surgery, and histologic examination showed a malignant tumor in 3 and a benign lesion in 40. Valuation of Cost Our prospective microcosting study included data for 167 patients. At IGR, the cost of a consultation (without FNAC) 766 766


is z17.3. The mean ± SD FNAC costs, including costs for an endocrinology consultation, performing the FNAC procedure, processing specimens, and interpreting the results are z107.7 ± z9.6 or z142.2 ± z21.8 for US-FNAC zTable 1z. Personnel costs represented 67% for both methods, whereas medical device and equipment costs were negligible. The FNAC examination was performed under US guidance in 37% of cases, and the mean cost of the FNAC examination inserted in the model was z120.5. The cost for a thyroidectomy, including the cost of surgery and the hospital stay, was z3,440. In the 60% of active patients, the mean duration of sick leave was estimated to be 32 days after surgery. For active patients who underwent operation, the compensated sick leave cost amounted to z863 for the National Sickness Fund, whereas the uncompensated cost was z1,088 for patients. Application of the Markov Model The distribution of patients among the different health states per follow-up period determined the baseline values and transition probabilities and their changes over time. At the end of the third cycle of the model, 47% of patients would not undergo surgery and their nodule would have been definitively classified as benign (because it remained in the benign state), whereas the other 53% of patients would undergo surgery. Of these 53%, 9.4% would have malignant nodules, corresponding to 6.6% of true-positives and 2.8% of false-negatives (ie, first classified as benign by cytologic findings and found to be malignant by histologic examination). Of the other 43.6% of patients with benign nodules by histopathologic examination, 35.7% would have true-negative results and 7.9% false-positive results (all found to be suspicious at cytologic examination and benign at histopathologic examination). When the Markov model was run for 3 cycles, the expected cost of a diagnosis by FNAC was z1,145 per patient, of which z271 (23.7%) and z46 (4.0%) were attributable to false-positive and false-negative results, respectively. Followup amounted to z194, or 16.9% of the cost of the diagnosis. The contribution of each cytologic result to the cost was then estimated. Unsatisfactory cytopathologic specimens amounted to z690 (690/1,145 = 60.3%), of which z630 corresponded to the cost of surgery. Suspicious results amounted to z282 (24.6%) and malignant and benign results amounted to z3 and z168, respectively zFigure 2z. By including indirect costs related to sick leave and patients’ related costs, the total cost would increase to z1,467 per patient. Sensitivity Analyses One-way sensitivity analyses consisted of decreasing the rate of initial unsatisfactory cytopathologic specimens. zFigure 3z shows a linear relationship between the rate of unsatisfactory specimens and the cost, and that a 35.1% reduc© American Society for Clinical Pathology

Anatomic Pathology / Original Article Histology + Malignant 2.4%

100% Histology – 0%

True-positive False-positive Histology +

Surgery

6.6% Histology –

25%

93.4%

False-negative True-negative Malignant 0% Benign

Benign 61%

2nd FNAC 29%

7.5% Unsatisfactory

Follow-up 75%

17% Clinical examination 71% Histology +

Surgery Suspicious 9.2% First FNAC

35% Histology –

81%

65% Follow-up 19%

75.5% Suspicious

Malignant Benign Suspicious Unsatisfactory

Benign True-positive False-positive

Suspicious Histology +

Surgery 25%

8.3% Histology – 91.7%

False-negative True-negative Malignant 0% Benign

Unsatisfactory 27.4%

2nd FNAC 63.2%

59.7% Suspicious 4.5% Unsatisfactory

Follow-up 75%

35.8% Clinical examination 36.8%

Malignant Benign Suspicious Unsatisfactory

Unsatisfactory

True-positive 0 True-negative 0 False-positive 0 False-negative 0

zFigure 1z Model of the management of patients with thyroid nodules according to initial cytopathologic results. FNAC, fineneedle aspiration cytology.

tion in the cost (from z1,145 to z743) of the procedure would be obtained with an absence of unsatisfactory specimens. The sensitivity analysis, which consisted of applying the cost of surgery only during the last cycle for unsatisfactory cytopathologic results, showed the total cost of FNAC examination would be © American Society for Clinical Pathology

reduced to z516 per patient. This important reduction of the cost confirms that the cost of diagnosis is highly dependent on the rate of unsatisfactory specimens and on the clinician’s judgment of whether to perform surgery. A sensitivity analysis involved the univariate variation of

767


767 767


zTable 1z Unit Cost of FNAC and US-FNAC* FNAC

Direct medical costs Medical devices Equipment Personnel Endocrinologist Radiologist Laboratory technician Cytopathologist Secretary Total direct costs Overhead and logistics Total cost

3.8 6.5 20.0 ± 7.3 — 18.4 27.4 5.0 81.0 ± 7.3 26.7 ± 2.4 107.7 ± 9.6

US-FNAC

3.8 6.5

23.6 ± 8.4 22.3 ± 8.0 18.4 27.4 5.0 106.9 ± 16.4 35.3 ± 5.4 142.2 ± 21.8

FNAC, fine-needle aspiration cytology; US, ultrasound guided. * Data are given as cost or mean ± SD cost in euros.

1,200 Consultation Repeated FNAC Surgery First FNAC

Total Cost per Patient ( )

1,000 800

Discussion

600 400 200 0 Malignant

Benign

Suspicious

Unsatisfactory

All Specimens

Results of the Initial FNAC

zFigure 2z Contribution of each cytopathologic result to the cost. FNAC, fine-needle aspiration cytology.

Basal model With systematic US-FNAC

25

19

14

11

5

0

Cost ( )

27

1,400 1,200 1,000 800 600 400 200 0

Rate of Unsatisfactory Specimens (%)

zFigure 3z Sensitivity analysis showing the effect of the reduction in the unsatisfactory specimen rates on the cost of diagnosis in patients with thyroid nodules. US-FNAC, ultrasound-guided fine-needle aspiration cytology.

768 768


FNAC examination has proved to be a simple, accurate, safe, and cost-effective method for the preoperative diagnosis of benign and malignant thyroid nodules. However, studies have considered the management, cost, and diagnostic accuracy values only for patients who undergo surgery. To our knowledge, this is the first study aimed at determining the cost of a true diagnosis of thyroid nodules using FNAC, by taking into account FNAC diagnostic mistakes and follow-up of patients who did not undergo surgery. Our model shows that after 3 cycles, 9.4% of the hypothetical cohort would have had a final diagnosis of thyroid cancer. This rate is higher than that generally reported, which usually ranges from 3% to 5%.5 The difference may be

Cost ( )

the rate of malignancy among patients whose lesions had first been classified as suspicious by cytopathologic examination and who subsequently underwent surgery. As shown in zFigure 4z, a 32.2% reduction (from z1,290 to z874) in the cost of the diagnosis could be obtained if the cytopathologist was always able to give an ascertained diagnosis. It is interesting that the cost was less impacted by the number of false-negative and false-positive results. For example, varying the rate of false-negative results observed in the literature (from 1.3% to 11.5%) would marginally increase the cost of diagnosis (from z1,108 to z1,179). If all patients had a routine US-FNAC, the cost of diagnosis would only increase from z1,145 to z1,177. This small extra cost would easily be compensated by a slight reduction from 27.4% to 25.2% in the rate of unsatisfactory specimens with the use of US-FNAC (Figure 3). This could easily be achieved because the rate of unsatisfactory specimens in our series was 20% for patients who underwent US-FNAC.

1,400 1,200 1,000 800 600 400 200 0

0

20 40 60 80 100 Rate of Specimens Suspicious at Cytopathology and Malignant at Histology (%)

zFigure 4z Sensitivity analysis showing the effect of an increased rate of malignancy on the cost of diagnosis in patients with thyroid nodules classified as “suspicious” by fine-needle aspiration cytology and who underwent surgery.



related to the effects of later detection of malignancy (followup) and diagnostic mistakes and also to the referral nature of IGR. False-negative results are a major pitfall in thyroid cancer diagnosis. In our model, the rate of false-negatives was estimated at 2.8%. This rate is not comparable to the 1.3% to 11.5% rate5,20-22 reported in studies that evaluated the diagnostic accuracy of FNAC examination, which reported incomplete results because they considered only surgically treated patients. In fact, our study takes into account follow-up data to estimate the rate of false-negative results and integrates the a posteriori detection of malignant nodules that were not detected at the first FNAC examination. Consequently, the false-negative percentage reported concerns the entire cohort of patients who underwent FNAC examination, including patients who did and did not undergo surgery. However, our results are consistent with the false-negative rates ranging from 0% to 2.3%23-28 reported in a few studies that investigated the long-term follow-up of patients with initially benign cytologic diagnoses. Finally, our false-negative rate may be overestimated given the potential development of a malignancy in a previously benign lesion, but this possibility seems rare, if it even occurs. This low rate explains why false-negative results represented only 4.0% (z46) of the cost of the diagnosis and that variation in the percentage did not significantly impact the cost. Moreover, the cost attributed to false-negative lesions is probably also underestimated because we failed to consider the economic consequences of a delayed diagnosis. Patients with false-negative results on FNAC examination and histopathologically proven thyroid carcinoma had a delay in surgical treatment for about 2 years, resulting in higher rates of vascular and capsular invasion and a 2-fold increased risk of persistent disease during follow-up,29 impacting the cost of the management of treatment and follow-up. In our study, suspicious findings accounted for 9.2% of all specimens obtained by FNAC, of which 35% and 65%, respectively, proved to be malignant and benign by histologic examination (in the 81% of patients who underwent operation). From an economic perspective, specimens found to be suspicious by FNAC examination and benign by histologic examination were classified as false-positives because they corresponded to unnecessary surgeries for patients and avoidable expenses for the payer. False-positive results were found in 7.9% of patients. This rate is higher than the 0% to 7.7% rate previously reported among cases with malignant cytopathologic results and benign histologic findings. The sensitivity analysis showed that better diagnosis of malignant nodules in patients with suspicious results would decrease the cost from z1,140 to z868 per patient. In the future, the use of new techniques, such as immunocytochemical analysis with thyroid peroxidase or galectin 3 antibodies or gene expression studies30-32 should improve this rate, but these techniques © American Society for Clinical Pathology

are not yet recognized standards, nor are they widely available. False-positives amounted to z271, ie, 23.7% of the cost of diagnosis, and this cost is probably underestimated. The inclusion of indirect costs of sick leave increased costs up to z1,467. Our sensitivity analyses showed that the cost may be significantly reduced, conditional to the performance of the cytopathologist and the rate of unsatisfactory specimens. Reduction of this rate from 27% to 0% would lead to a cost reduction of 35.1% (from z1,145 to z743 per patient). The results of the sensitivity analysis showed that the extra cost of routine US-FNAC would be offset if it permitted a relative reduction of the rate of unsatisfactory results of 2.2% or more (from 27.4% to 25.2%; Figure 3).The application of the Markov model to the subgroup of patients who had a US-FNAC as the first test showed a reduction of the cost (from z1,145 to z1,084 per patient). According to the literature, the routine use of US-FNAC allows a decrease in the rate of unsatisfactory specimens from 15% to 20% down to 5%.33-36 US-FNAC makes the diagnosis possible in 50% to 63% of initially unsatisfactory FNAC results.37-39 In addition, the on-site assessment of FNAC specimen adequacy increases the likelihood of accurate diagnosis40-42 because the risk of initial unsatisfactory cytologic specimens remains significant (5%-15%),43 and the rate of cancer in surgically resected nodules is higher in unsatisfactory than in benign FNAC results (7.1% and 2.1%, respectively).44 It was estimated that on-site evaluation allowed a benefit of z55 per case, based on a reduction of repeated FNAC owing to unsatisfactory specimens (1% vs 20%) and despite the additional cost of z177 per case for the immediate professional service.45 The important distinguishing feature of the model is the consideration not only of the cost of FNAC examination but also the entire diagnosis process, combining specific characteristics of the test, relative skill of the cytopathologist, and the clinician’s judgment as to whether surgery should be performed. The model had 2 limitations. First, the model described joint production of diagnosis process and care, for which it is difficult to separate the costs. Two scenarios were calculated: if the cost of surgery was applied to all unsatisfactory results that resulted in surgery, the cost of diagnosis reached z1,145, whereas it reached only z516 when the cost was attributed only to a fraction of the lesions. Second, the contribution of the cost of surgery to the total cost of diagnosis is very important (Figure 2). The clinical judgment of whether to operate, which determined the surgery frequency by center, led to great variations in the diagnosis cost across institutions. Data on the care process of patients and resources consumed were collected retrospectively from the computerized medical records of the hospital. Consequently, the perspective of the economic evaluation was restricted to the hospital per-

769


769 769


spective because of the absence of data on extrahospital costs. The hospital perspective was adopted because we wanted to inform clinicians and managers of the budgetary impact of the diagnosis of thyroid nodules. Certain costs, such as patient-related costs, loss of work income, litigation costs in case of diagnostic errors, and transport costs, were excluded from the basal analysis, leading to an underestimation of the cost of diagnosis. The hospital perspective can be considered a limit of the model: nations with different cost structures may prefer other perspectives. In the United States, a more appropriate perspective may be from the payer viewpoint because hospitals and services would generate more revenue based on unsatisfactory FNAC specimens because of repeated FNAC procedures. Nevertheless, a complementary analysis was completed by adding compensated and uncompensated sick leave costs. Data collection and the model were assessed at only 1 study site, so the generalization of our results remains questionable. However, the aim of this study was not to generalize the results from our particular center to practices throughout France, but to provide practitioners with a model allowing them to evaluate their own costs. More generally, our model can also be used by other centers to assess the performance and the cost of their own diagnostic process, from the initial FNAC examination to the final assessment of the patient’s status.

Conclusion This study demonstrates that the cost of a given procedure for a true diagnosis exceeds its unit cost of production. The cost study of a diagnostic procedure should take into account its predictive value, ie, its errors and its failures (ie, unsatisfactory specimens). Thyroid FNAC failures generate a much higher cost than errors. Routine use of US-FNAC and on-site assessment of the adequacy of cytopathologic specimens should dramatically reduce the rate of failures and, thus, the cost of a true diagnosis. From the Departments of 1Health Economics and 2Pathology, Institut de Cancérologie Gustave Roussy, and 3Nuclear Medicine and Endocrine Oncology, Institut de Cancérologie Gustave Roussy and University Paris Sud-11, Villejuif, France; 4Genzyme, Cambridge, MA; and 5Health Economics, ESSEC Business School, Cergy Pontoise, France. Supported by an unrestricted grant from Genzyme. Presented in part as a poster at the American Society of Clinical Oncology meeting; June 1-5, 2007; Chicago, IL. Address reprint requests to Dr Borget: Dept of Health Economics, Institut de Cancérologie Gustave Roussy, 39 rue Camille Desmoulins, 94805 Villejuif, France. Acknowledgment: We thank Lorna Saint Ange for editing the manuscript.

770 770


References 1. Hegedus L. Clinical practice: the thyroid nodule. N Engl J Med. 2004;351:1764-1771. 2. Singer PA, Cooper DS, Daniels GH, et al, for the American Thyroid Association. Treatment guidelines for patients with thyroid nodules and well-differentiated thyroid cancer. Arch Intern Med. 1996;156:2165-2172. 3. Cooper DS, Doherty GM, Haugen BR, et al. Management guidelines for patients with thyroid nodules and differentiated thyroid cancer. Thyroid. 2006;16:109-142. 4. AACE/AME Task Force on Thyroid Nodules. American Association of Clinical Endocrinologists and Associazione Medici Endocrinologi medical guidelines for clinical practice for the diagnosis and management of thyroid nodules. Endocr Pract. 2006;12:63-102. 5. Gharib H, Goellner JR. Fine-needle aspiration biopsy of the thyroid: an appraisal. Ann Intern Med. 1993;118:282-289. 6. Mittendorf EA, Tamarkin SW, McHenry CR. The results of ultrasound-guided fine-needle aspiration biopsy for evaluation of nodular thyroid disease. Surgery. 2002;132:648-654. 7. Cesur M, Corapcioglu D, Bulut S, et al. Comparison of palpation-guided fine-needle aspiration biopsy to ultrasoundguided fine-needle aspiration biopsy in the evaluation of thyroid nodules. Thyroid. 2006;16:555-561. 8. Izquierdo R, Arekat MR, Knudson PE, et al. Comparison of palpation-guided versus ultrasound-guided fine-needle aspiration biopsies of thyroid nodules in an outpatient endocrinology practice. Endocr Pract. 2006;12:609-614. 9. Hamberger B, Gharib H, Melton LJ III, et al. Fine-needle aspiration biopsy of thyroid nodules; impact on thyroid practice and cost of care. Am J Med. 1982;73:381-384. 10. Ng EH, Lim-Tan SK, Nambiar R. Fine needle aspiration cytology in the management of solitary thyroid nodules: a comparison with other diagnostic modalities in costeffectiveness. Singapore Med J. 1989;30:557-560. 11. Caplan RH, Strutt PJ, Kisken WA, et al. Fine needle aspiration biopsy of thyroid nodules. Wis Med J. 1991;90:285-288. 12. Rimm DL, Stastny JF, Rimm EB, et al. Comparison of the costs of fine-needle aspiration and open surgical biopsy as methods for obtaining a pathologic diagnosis. Cancer. 1997;81:51-56. 13. Bajaj Y, De M, Thompson A. Fine needle aspiration cytology in diagnosis and management of thyroid disease. J Laryngol Otol. 2006;120:467-469. 14. Khalid AN, Hollenbeak CS, Quraishi SA, et al. The costeffectiveness of iodine 131 scintigraphy, ultrasonography, and fine-needle aspiration biopsy in the initial diagnosis of solitary thyroid nodules. Arch Otolaryngol Head Neck Surg. 2006;132:244-250. 15. Sangalli G, Serio G, Zampatti C, et al. Fine needle aspiration cytology of the thyroid: a comparison of 5469 cytological and final histological diagnoses. Cytopathology. 2006;17:245-250. 16. Amrikachi M, Ramzy I, Rubenfeld S, et al. Accuracy of fine-needle aspiration of thyroid. Arch Pathol Lab Med. 2001;125:484-488. 17. Hooft L, Hoekstra OS, Boers M, et al. Practice, efficacy, and costs of thyroid nodule evaluation: a retrospective study in a Dutch university hospital. Thyroid. 2004;14:287-293. 18. Sonnenberg FA, Beck JR. Markov models in medical decision making: a practical guide. Med Decis Making. 1993;13:322-338. 19. Faquin WC. Aspiration of the thyroid. In: Atkinson BF, ed. Atlas of Diagnostic Cytopathology. 2nd ed. Philadelphia, PA: Saunders; 2003:459-496. © American Society for Clinical Pathology


20. Baloch ZW, Sack MJ, Yu GH, et al. Fine-needle aspiration of thyroid: an institutional experience. Thyroid. 1998;8:565-569. 21. Gardiner GW, de Souza FM, Carydis B, et al. Fine-needle aspiration biopsy of the thyroid gland: results of a five-year experience and discussion of its clinical limitations. J Otolaryngol. 1986;15:161-165. 22. Altavilla G, Pascale M, Nenci I. Fine needle aspiration cytology of thyroid gland diseases. Acta Cytol. 1990;34:251-256. 23. Mittendorf EA, McHenry CR. Follow-up evaluation and clinical course of patients with benign nodular thyroid disease. Am Surg. 1999;65:653-658. 24. Liel Y, Ariad S, Barchana M. Long-term follow-up of patients with initially benign thyroid fine-needle aspirations. Thyroid. 2001;11:775-778. 25. Orlandi A, Puscar A, Capriata E, et al. Repeated fineneedle aspiration of the thyroid in benign nodular thyroid disease: critical evaluation of long-term follow-up. Thyroid. 2005;15:274-278. 26. Grant CS, Hay ID, Gough IR, et al. Long-term follow-up of patients with benign thyroid fine-needle aspiration cytologic diagnoses. Surgery. 1989;106:980-986. 27. Julian JS, Pittman CE, Accettullo L, et al. Does fine-needle aspiration biopsy really spare patients thyroidectomy? Am Surg. 1989;55:238-242. 28. Mazzawi SJ, Rosen G, Luboshitzky R, et al. Management of benign thyroid nodules based on the findings of fine-needle aspiration. J Otolaryngol. 2000;29:95-97. 29. Yeh MW, Demircan O, Ituarte P, et al. False-negative fineneedle aspiration cytology results delay treatment and adversely affect outcome in patients with thyroid carcinoma. Thyroid. 2004;14:207-215. 30. Weber F, Shen L, Aldred MA, et al. Genetic classification of benign and malignant thyroid follicular neoplasia based on a three-gene combination. J Clin Endocrinol Metab. 2005;90:2512-2521. 31. Bartolazzi A, Gasbarri A, Papotti M, et al. Application of an immunodiagnostic method for improving preoperative diagnosis of nodular thyroid lesions. Lancet. 2001;357:1644-1650. 32. De Micco C, Vassko V, Henry JF. The value of thyroid peroxidase immunohistochemistry for preoperative fine-needle aspiration diagnosis of the follicular variant of papillary thyroid cancer. Surgery. 1999;126:1200-1204. 33. Carmeci C, Jeffrey RB, McDougall IR, et al. Ultrasoundguided fine-needle aspiration biopsy of thyroid masses. Thyroid. 1998;8:283-289.


34. Yang GC, Liebeskind D, Messina AV. Ultrasound-guided fine-needle aspiration of the thyroid assessed by ultrafast Papanicolaou stain: data from 1135 biopsies with a two- to six-year follow-up. Thyroid. 2001;11:581-589. 35. Takashima S, Fukuda H, Kobayashi T. Thyroid nodules: clinical effect of ultrasound-guided fine-needle aspiration biopsy. J Clin Ultrasound. 1994;22:535-542. 36. Danese D, Sciacchitano S, Farsetti A, et al. Diagnostic accuracy of conventional versus sonography-guided fine-needle aspiration biopsy of thyroid nodules. Thyroid. 1998;8:15-21. 37. Braga M, Cavalcanti TC, Collaco LM, et al. Efficacy of ultrasound-guided fine-needle aspiration biopsy in the diagnosis of complex thyroid nodules. J Clin Endocrinol Metab. 2001;86:4089-4091. 38. Castro MR, Gharib H. Continuing controversies in the management of thyroid nodules. Ann Intern Med. 2005;142:926-931. 39. Alexander EK, Heering JP, Benson CB, et al. Assessment of nondiagnostic ultrasound-guided fine needle aspirations of thyroid nodules. J Clin Endocrinol Metab. 2002;87:4924-4927. 40. Zhu W, Michael CW. How important is on-site adequacy assessment for thyroid FNA? an evaluation of 883 cases. Diagn Cytopathol. 2007;35:183-186. 41. Eedes CR, Wang HH. Cost-effectiveness of immediate specimen adequacy assessment of thyroid fine-needle aspirations. Am J Clin Pathol. 2004;121:64-69. 42. Ghofrani M, Beckman D, Rimm DL. The value of onsite adequacy assessment of thyroid fine-needle aspirations is a function of operator experience. Cancer. 2006;108:110-113. 43. de los Santos ET, Keyhani-Rofagha S, Cunningham JJ, et al. Cystic thyroid nodules; the dilemma of malignant lesions. Arch Intern Med. 1990;150:1422-1427. 44. Slowinnska-Klencka D, Sporny S, Klencki M, et al. Non-diagnostic cytological outcome of thyroid biopsy and the risk of thyroid malignancy. Endocr Pathol. 2004;15:65-75. 45. Nasuti JF, Gupta PK, Baloch ZW. Diagnostic value and costeffectiveness of on-site evaluation of fine-needle aspiration specimens: review of 5,688 cases. Diagn Cytopathol. 2002;27:1-4.

771


771 771