- Email: [email protected]
Routine prophylactic central neck dissection for low-risk papillary thyroid cancer: A cost-effectiveness analysis
Routine prophylactic central neck dissection for low-risk papillary thyroid cancer: A cost-effectiveness analysis Kyle Zanocco, MD,a,b Dina Elaraj, MD,a and Cord Sturgeon, MD,a Chicago, IL
Background. Routine prophylactic central neck dissection (pCND) after total thyroidectomy (TTX) for low-risk papillary thyroid cancer (PTC) offers the potential to decrease disease recurrence but may increase operative complications. We hypothesized that routine pCND is not cost-effective in low-risk PTC. Methods. A Markov transition-state model was constructed to compare TTX with and without pCND. Outcome probabilities, utilities, and costs were estimated on the basis of literature review. The threshold for cost-effectiveness was $100,000 per quality-adjusted life year. Sensitivity analysis was used to examine model uncertainty. Results. pCND cost $10,315 and produced an effectiveness of 23.785 quality-adjusted life years. This strategy was more costly and less effective than TTX without pCND and was therefore dominated. pCND became cost-effective when the probability of recurrence increased from 6% to 10.3%, cost of reoperation for recurrence increased from $8,900 to $26,120, or added probabilities of recurrent laryngeal nerve injury and hypoparathyroidism due to pCND were less than 0.20% and 0.18% during 2-way sensitivity analysis. Monte Carlo simulation showed that pCND was not cost-effective in 97.3% of iterations. Conclusion. Routine pCND for low-risk PTC is not cost-effective unless the recurrence rate is greater than 10.3%. Application of pCND should be individualized based on risk of recurrence and added complications. (Surgery 2013;154:1148-55.) From the Department of Surgery,a Section of Endocrine Surgery, and Center for Healthcare Studies,b Northwestern University Feinberg School of Medicine, Chicago, IL
CERVICAL LYMPH NODE METASTASES are common in papillary thyroid cancer (PTC). PTC-positive nodes in the central neck are identified frequently after operative resection in patients who appear to be clinically free of nodal disease. In these patients, an estimated 37 64% will have microscopically identifiable lymph node metastases.1 However, these metastases appear to be of limited importance, given that the locoregional recurrence rate for this group is much lower (approximately 6%)
Kyle Zanocco’s participation in this study was supported in part by the Northwestern University Feinberg School of Medicine Center for Healthcare Studies under an institutional award from the Agency for Healthcare Research and Quality, T-32 HS 000078 (PI: Jane L. Holl, MD, MPH). Accepted for publication June 21, 2013. Reprint requests: Cord Sturgeon, MD, Department of Surgery, Northwestern University Feinberg School of Medicine, 676 N St. Clair St. Ste 650, Chicago, IL 60611. E-mail: csturgeo@ nmh.org. 0039-6060/$ - see front matter Ó 2013 Mosby, Inc. All rights reserved. http://dx.doi.org/10.1016/j.surg.2013.06.016
1148 SURGERY
even if central neck dissection is not performed after initial thyroidectomy.1,2 Routine prophylactic central neck dissection (pCND) in clinically and intraoperatively nodenegative PTC is controversial. Proponents of this practice argue that pCND decreases the need for reoperative neck surgery by reducing locoregional recurrence and simplifies follow-up by lowering postoperative serum thyroglobulin.2-4 Opponents maintain that this procedure exposes patients to additional risk of recurrent laryngeal nerve (RLN) injury and hypoparathyroidism without proven benefit.2 A prospective, randomized trial comparing pCND versus no pCND is the ideal method for assessing these purported benefits and risks. However, this approach is not feasible given the prohibitively large patient sample size necessary to detect the hypothesized clinical differences in recurrence and complications between the 2 groups.2 Given the unavailability of randomized controlled data for the foreseeable future, we sought to model these outcomes using decision analysis. We hypothesized that routine pCND is not cost-effective in low-risk PTC.
Surgery Volume 154, Number 6
METHODS Reference case scenario. The reference case patient in this analysis was defined as a 40-year-old patient with a 2-cm, noninvasive PTC that was without preoperative or intraoperative evidence of nodal involvement. This patient was assumed to be without a history of previous neck surgery and an appropriate candidate for thyroidectomy via cervical incision. Decision model. A Markov transition-state decision model was constructed for the reference case using decision analysis software (TreeAge Pro; TreeAge Software, Inc, Williamstown, MA, 2012). Two alternatives were created: (1) total thyroidectomy alone or (2) total thyroidectomy with pCND. The outcome probabilities for these alternatives were estimated from literature review and are listed in Table I.5-11 The model had a 1-year cycle length and was terminated when the number of completed cycles exceeded patient life expectancy. The lifetime risk of recurrence without pCND was estimated to be 6% (1.2% per year for 5 years of follow-up).1,2 All recurrences were assumed to be definitively treated with reoperative neck dissection of affected compartments.4 The relative risk reduction of recurrence after pCND, compared with total thyroidectomy without pCND, was assumed to be 75%. A large-risk reduction was used to maintain a conservative overall model perspective given the hypothesis that pCND was not cost-effective. The strategy that provided the greatest quality-adjusted life expectancy (QALE) without exceeding an incremental costeffectiveness ratio (ICER) of $100,000 per qualityadjusted life year (QALY) was defined as optimal. We chose $100,000 per QALY as the threshold for cost-effectiveness on the basis of analysis of the cost of current health care resource allocation decisions in the United States and currentlyaccepted convention.12 Cost estimation. A combination of micro- and gross costing techniques was used to estimate the costs of resources consumed for each strategy from the societal perspective. Only costs that differed between the alternative strategies were included.13,14 All costs in the model were reported in 2010 U.S. dollars. An inflation rate of 4.1%, calculated as the mean of annual changes in the Consumer Price Index for Medical Care from 2000 to 2010, was used to adjust older costs to their 2010 amount when necessary.15 A discount rate of 3% was applied to all future costs in the model.13 Cost estimates for physician labor were determined by the 2010 national Medicare charge limits for the surgery, anesthesiology, and pathology
Zanocco, Elaraj, and Sturgeon 1149
services rendered in each strategy.16 Hospital costs were estimated by calculating weighted averages of 2009 Medicare cost-to-charge ratios for relevant diagnosis-related groups and multiplying these ratios by national median impatient charge estimates for thyroid surgery and neck lymph node dissection reported by the 2009 Nationwide Inpatient Sample.17,18 Thyroid hormone-replacement costs were based on average U.S. wholesale prices.19 The costs for the thyroidectomy-related complications of permanent RLN injury and permanent hypoparathyroidism have been estimated in previous work and were adjusted to 2010 costs.20 Missed patient work due to surgery was assigned an hourly cost based on 2010 annual US hourly earnings.21 All of the costs used in the model are listed in Table I. Effectiveness. QALE was used as the measure of effectiveness. Each of the model’s different outcomes was assigned a corresponding quality adjustment factor based on literature review. The time elapsed in a particular outcome was multiplied by that outcome’s quality adjustment factor to yield effectiveness in QALYs. Patients who developed recurrent disease were assigned a one-time treatment-related reduction in quality-adjusted life expectancy.22 No other survival difference was assumed between the 2 management strategies in the reference case.1,2,23,24 All future QALYs were discounted at 3%.13 The quality adjustment factors used in the model are listed with supporting references in Table I. Sensitivity analysis. Threshold analysis was performed on each variable in the reference case to identify values at which the ICER exceeded $50,000/QALY and $100,000/QALY during oneway sensitivity analysis. Two-way sensitivity analysis was performed to examine the combined impact of changes in lifetime recurrence after total thyroidectomy without pCND and recurrence risk reduction with additional pCND. The combined effects of the additional probabilities of RLN injury and permanent hypoparathyroidism with pCND were also evaluated with 2-way analysis. Probabilistic sensitivity analysis was performed with Monte Carlo simulation, where triangular frequency distributions for each variable were simultaneously sampled during 1,000 consecutive iterations. In this simulation, all variables were allowed to vary up to ±50% of the reference case estimate or, if not possible in the case of probability and utility estimates, to the greatest extent such that the reference case value remained at the center of a range containing the boundary 0 or 1.
1150 Zanocco, Elaraj, and Sturgeon
Surgery December 2013
Table I. Model assumptions Variables Probabilities Lifetime recurrence Relative risk reduction of recurrence after pCND RLN injury after total thyroidectomy Additional risk of RLN injury after pCND Permanent hypoparathyroidism after total thyroidectomy Additional risk of permanent hypoparathyroidism after pCND RLN injury after reoperation for recurrence Permanent hypoparathyroidism after reoperation for recurrence Costs in U.S. dollars Total thyroidectomy Additional cost of pCND Reoperation for recurrence Treatment of RLN injury Annual cost of permanent hypoparathyroidism Annual cost of thyroid hormone replacement Hourly cost of missed patient work Quality-of-life adjustment factors Hypothyroid on thyroid hormone replacement Hypothyroid with RLN injury Hypothyroid with hypoparathyroidism Hypothyroid with hypoparathyroidism and RLN injury 1-time reduction in quality-adjusted life expectancy for undergoing reoperation for recurrence, d Time Work missed to undergo reoperation for recurrence, h Life expectancy, y
Reference case value 6% 75%
Discount rate
Range of values used in Monte Carlo simulation distributions 3% to 9% 50% to 100%
Sources 1,2 3
5
2% 0.25% 1%
1% to 3% 0.125% to 0.375% 0.5% to 1.5%
0.25%
0.125% to 0.375%
2,26
1.5% to 4.5% 1.5% to 4.5%
24
$5,500 $513 $8,887 $10,367 $863
$2,750 to $8,250 $256 to $770 $4,444 to $13,331 $5,183 to $15,550 $432 to $1,295
16,18
$152 $23
$76 to $227 $11 to $34
19
0.990
0.980 to 1
10
0.881 0.893 0.775
0.762 to 1 0.786 to 1 0.550 to 1
10,22
11
5.5 to 16.5
Expert opinion
16
8 to 24
Expert opinion
40
20 to 60
3,11
3%
1.5% to 4.5%
13
3% 3%
2,26 6-9
24
16,18 16-18 20 20
21
10,22 10,22
pCND, Prophylactic central neck dissection; RLN, recurrent laryngeal nerve.
RESULTS Reference case. Total thyroidectomy without pCND was the less costly strategy with an expected cost of $10,149 and an effectiveness of 23.791 QALYs. The strategy of total thyroidectomy with routine pCND had an expected cost of $10,315 and an effectiveness of 23.785 QALYs. The addition of routine pCND resulted in an expected incremental cost of $166 and a loss of 0.006 QALYs. Routine pCND was therefore dominated because this strategy was both more costly and less effective than total thyroidectomy alone.
Sensitivity analysis. The threshold values of all the model variables that were necessary to make routine pCND cost-effective during one-way sensitivity analysis are listed in Table II. The model was most sensitive to lifetime recurrence rate, with pCND becoming the optimal strategy if recurrence after total thyroidectomy alone exceeded 10.3% (Fig 1). pCND also became cost-effective if the cost of treatment for recurrence exceeded $26,120 (Fig 2) or the reduction in QALE due to recurrence exceeded 76 days. At the $100,000/ QALY threshold level, the model was not sensitive
Zanocco, Elaraj, and Sturgeon 1151
Surgery Volume 154, Number 6
Table II. $100,000/QALY threshold values for model variables that make prophylactic central neck dissection cost-effective Variable Probabilities Lifetime recurrence Relative risk reduction of recurrence after pCND RLN injury after total thyroidectomy Additional risk of RLN injury after pCND Permanent hypoparathyroidism after total thyroidectomy Additional risk of permanent hypoparathyroidism after total thyroidectomy RLN injury after reoperation for recurrence Permanent hypoparathyroidism after reoperation for recurrence Costs in U.S. dollars Total thyroidectomy Additional cost of pCND Reoperation for recurrence Treatment of RLN injury Annual cost of permanent hypoparathyroidism Annual cost of thyroid hormone replacement Hourly cost of missed patient work Quality-of-life adjustment factors Hypothyroid on thyroid hormone replacement Hypothyroid with RLN injury Hypothyroid with hypoparathyroidism Hypothyroid with hypoparathyroidism and RLN injury 1-time reduction in quality-adjusted life expectancy for undergoing reoperation for recurrence (d) Time Work missed to undergo reoperation for recurrence (h) Life expectancy (y) Discount rate
$50,000/QALY threshold
$100,000/QALY threshold
>9.6% None None None None None
>10.3% None None None None None
>18.8% >16.5%
>15.5% >15.1%
None <$52 >$19,480 None None None >$680
None None >$26,120 None None None >$1,100
<0.858 None None None >90
<0.882 None None None >75.9
>482 None None
>780 None None
pCND, Prophylactic central neck dissection; RLN, recurrent laryngeal nerve.
to life expectancy, the quality adjustment factors for RLN injury or hypoparathyroidism, the additional cost of pCND, or the relative risk reduction of recurrence after pCND. Two-way sensitivity analysis demonstrated that the decision model was sensitive to simultaneous changes in several clinically relevant variables. If a sufficiently large risk of recurrence after total thyroidectomy without pCND was combined with a substantial risk reduction after pCND, routine pCND was cost-effective (Fig 3). Figure 4 shows that routine pCND also became cost-effective when combinations of the additional risk of RLN injury and permanent hypoparathyroidism with pCND were sufficiently low (eg, when the additional risks were both 0.08%). Monte Carlo simulation showed total thyroidectomy without pCND to be the optimal strategy in 973 (97.3%) of the iterations. Routine pCND was both more costly and less effective (ie, dominated)
in 845 iterations (84.5%). In an additional 118 iterations (11.8%), forgoing pCND resulted in gains in quality of life that had an incremental cost that was low enough to produce an ICER of less than $100,000 QALY. Of the 27 (2.7%) cases in which routine pCND was cost-effective, this strategy was less costly and less effective compared with total thyroidectomy alone in 1 (0.1%) and dominant (ie, less costly and more effective) in 7 (0.7%). All of the iterations from this simulation are plotted in Fig 5. DISCUSSION This is the first cost-effectiveness analysis of routine pCND for PTC. We found that routine pCND for low-risk PTC is more costly and less effective than total thyroidectomy alone. However, pCND would be cost-effective if the PTC recurrence rate exceeds 10.3%, the cost of treatment for recurrence exceeds $26,120, or the additional risks
1152 Zanocco, Elaraj, and Sturgeon
Fig 1. One-way sensitivity analysis of the ICER of total thyroidectomy without routine pCND as a function of lifetime recurrence rate. The dashed line represents the 100,000/QALY threshold for cost-effectiveness. When the recurrence rate after total thyroidectomy alone exceeded 10.3%, routine pCND became costeffective.
Fig 2. One-way sensitivity analysis displaying the ICER of total thyroidectomy without routine pCND as a function of the cost of reoperation for recurrence. The dashed line represents the 100,000/QALY threshold for costeffectiveness. When the cost of reoperation exceeded $26,120, routine pCND became cost-effective.
of RLN injury and permanent hypoparathyroidism with this procedure are negligible. Although to our knowledge, no previous authors have examined the costs of pCND, several recent retrospective analyses have compared outcomes of PTC patients who underwent total thyroidectomy with and without this additional procedure.3,23,25,26 One of these studies demonstrated that the 20 CLND procedures were required to prevent 1 central compartment reoperation.3 Two meta-analyses have demonstrated an increased rate of transient hypocalcemia after pCND without showing a difference in permanent hypoparathyroidism or RLN injury.23,26 Both of
Surgery December 2013
Fig 3. This 2-way sensitivity analysis demonstrates the combined impact of changes in lifetime recurrence after total thyroidectomy without pCND and recurrence risk reduction after the addition of pCND. The dashed threshold line represents combinations of these attributes that produce an ICER of $100,000 per QALY. The area to the right of the threshold line represents values that would make pCND a cost-effective strategy. The reference case assumptions intersect to the left of the threshold line in the area representing values where pCND is not cost-effective.
Fig 4. This 2-way sensitivity analysis demonstrates the combined effects of changes in the additional risk of RLN injury and permanent hypoparathyroidism with pCND. The dashed threshold line represents combinations of these risks that produce an ICER of $100,000 per QALY. The area to the left of the threshold line represents values that would make pCND a cost-effective strategy. The reference case risks intersect to the right of the threshold line in the area representing values where pCND is not cost-effective.
these studies had total sample sizes of 1,100 1,200 patients, far less than the estimated 2,000 5,000 needed to detect 1% or greater differences in the permanent complications with adequate statistical power.2 Given that the increased risk of these complications with pCND
Surgery Volume 154, Number 6
Zanocco, Elaraj, and Sturgeon 1153
Fig 5. A 1,000 iteration Monte Carlo simulation in which all model assumptions were simultaneously varied across their frequency distributions. The incremental cost and incremental effectiveness of total thyroidectomy without versus with pCND are plotted for each iteration. Points on the left of the dashed 100,000/QALY threshold line represent iterations where pCND was cost-effective. To the right of the line are iterations where pCND was not cost-effective. Total thyroidectomy without pCND was cost-effective in 97.3% of the iterations and the dominant strategy in 84.5% of the iterations, whereas total thyroidectomy with pCND was cost-effective in 2.7% of iterations and dominant in 0.7% of the iterations.
may be even smaller than 1%, an even larger sample size would be necessary before it would be appropriate to conclude that no difference exists with the addition of this procedure. Several limitations are evident in our study. The decision model omitted the operative complications of wound infection, hematoma formation, and the possibility of operative mortality. These omissions are unlikely to have significant impact on the results given that both strategies involved surgery. However, these events may be more likely during reoperation, making treatment of locoregional recurrence less favorable, and therefore improving the cost-effectiveness of the pCND strategy. The effects of temporary RLN injury and temporary hypocalcemia also were omitted because of their relatively low expected impact on overall cost and quality-of-life. If these temporary complications were more likely to occur with pCND, their inclusion in the model would have further increased the dominance of the thyroidectomy alone strategy. Although our model assumed all surgery was performed on an inpatient basis, the frequency of lower-cost outpatient thyroidectomy has markedly increased during the past decade.27 If the addition of pCND converted an outpatient thyroidectomy into an inpatient procedure, the additional cost of pCND would have been $1,173 instead of $513, based reimbursement data from the
Medicare Hospital Outpatient Prospective Payment System. This additional cost would have increased the dominance of the thyroidectomy alone strategy. Our model assumed no difference in postoperative treatment after initial operation between those who did and did not receive pCND; however, there is evidence that pCND leads to greater radioactive iodine (RAI) doses because of the detection of involved cervical lymph nodes.25,28 Adding the cost of additional RAI for these patients would have further diminished the costeffectiveness of pCND and would not have changed our conclusion. Conversely, if pCND led to less postoperative RAI administration through the demonstration of a node-negative central compartment, the overall cost of the pCND strategy would have been lower; however, the costsavings from less RAI would have needed to be greater than $750 per patient before changing the model’s overall conclusion. Although the model’s cost of recurrence included the procedural and ancillary costs of hospitalization for reoperation, some other costs were omitted for simplicity, including additional transportation to undergo reoperation and additional medical follow-up. Given that the overall cost of treatment for recurrence could have increased by more than $20,000 before pCND became costeffective, it is unlikely that these omissions would
1154 Zanocco, Elaraj, and Sturgeon
change the model’s conclusion. The model also did not allow for lower-cost alternative treatments for recurrence, including observation or RAI administration. Replacing some of the reoperations with these alternatives would have lowered the overall strategy cost of initial total thyroidectomy without pCND and increased this option’s dominance. The operative complications of RLN injury and permanent hypoparathyroidism have a wide spectrum of severity and treatment costs, but these considerations were simplified into single gross cost and quality of life estimates. We believe the impact of these simplifications is adequately addressed with sensitivity analysis. Comprehensive bilateral central compartment dissection may improve survival in higher-risk PTC patients.4 Our model did not assume a survival difference between low-risk PTC patients undergoing total thyroidectomy with or without pCND. However, it did show with sensitivity analysis that pCND would be cost-effective if disease recurrence reduced QALE by 76 days or greater. This finding suggests that central compartment dissection is cost-effective in cases where this technique provides patents a survival advantage. In conclusion, routine pCND for low-risk PTC is not cost-effective. Selective application of pCND should be individualized based on a balanced evaluation of the anticipated rate of recurrence versus additional risk of morbidity due to this procedure. REFERENCES 1. Randolph GW, Duh QY, Heller KS, et al. The prognostic significance of nodal metastases from papillary thyroid carcinoma can be stratified based on the size and number of metastatic lymph nodes, as well as the presence of extranodal extension. Thyroid 2012;22:1144-52. 2. Carling T, Carty SE, Ciarleglio MM, et al. American Thyroid Association design and feasibility of a prospective randomized controlled trial of prophylactic central lymph node dissection for papillary thyroid carcinoma. Thyroid 2012; 22:237-44. 3. Popadich A, Levin O, Lee JC, et al. A multicenter cohort study of total thyroidectomy and routine central lymph node dissection for cN0 papillary thyroid cancer. Surgery 2011;150:1048-57. 4. Cooper DS, Doherty GM, Haugen BR, et al. Revised American Thyroid Association management guidelines for patients with thyroid nodules and differentiated thyroid cancer. Thyroid 2009;19:1167-214. 5. Lo CY, Kwok KF, Yuen PW. A prospective evaluation of recurrent laryngeal nerve paralysis during thyroidectomy. Arch Surg 2000;135:204-7. 6. Dobrinja C, Trevisan G, Piscopello L, Fava M, Liguori G. Comparison between thyroidectomy and hemithyroidectomy in treatment of single thyroid nodules identified as indeterminate follicular lesions by fine-needle
Surgery December 2013
7.
8.
9.
10.
11.
12.
13. 14. 15.
16.
17.
18.
19. 20.
21.
22.
23.
24.
25.
aspiration cytology. Ann Ital Chir 2010;81:403-10; discussion 410-1. Page C, Strunski V. Parathyroid risk in total thyroidectomy for bilateral, benign, multinodular goitre: report of 351 surgical cases. J Laryngol Otol 2007;121:237-41. Sitges-Serra A, Ruiz S, Girvent M, Manjon H, Duenas JP, Sancho JJ. Outcome of protracted hypoparathyroidism after total thyroidectomy. Br J Surg 2010;97:1687-95. Youngwirth L, Benavidez J, Sippel R, Chen H. Parathyroid hormone deficiency after total thyroidectomy: incidence and time. J Surg Res 2010;163:69-71. Muennig P, Khan K. Designing and conducting costeffectiveness analyses in medicine and health care. 1st ed. San Francisco (CA): Jossey-Bass; 2002. Bell FC, Miller ML. Life Tables for the United States Social Security Area 1900-2100, Actuarial Study No. 120 2011. Available from http://www.socialsecurity.gov/ OACT/NOTES/as120/LifeTables_Tbl_6_2010.html. Braithwaite RS, Meltzer DO, King JT Jr, Leslie D, Roberts MS. What does the value of modern medicine say about the $50,000 per quality-adjusted life-year decision rule? Med Care 2008;46:349-56. Gold MR. Cost-effectiveness in health and medicine. New York (NY): Oxford University Press; 1996. Muennig P. Cost-effectiveness analyses in health: a practical approach. 2nd ed. San Francisco (CA): Jossey-Bass; 2008. Consumer Price Index - Medical Care. 2011. Available from http://data.bls.gov/timeseries/CUUR0000SAM? output_view=pct_12mths. Overview of the Medicare Physician Fee Schedule Search. 2011. Available from https://www.cms.gov/apps/physicianfee-schedule/overview.aspx. HCUPnet: A tool for identifying, tracking, and analyzing national hospital statistics. 2011. Available from http:// hcupnet.ahrq.gov/. MEDPAR Medicare Fee for Service for Parts A & B. 2011. Available from http://www.cms.gov/MedicareFeeforSvcPartsAB/ 03_MEDPAR.asp#TopOfPage. Thomson Healthcare. Red book. Montvale (NJ): Thomson PDR; 2010. v. Zanocco K, Angelos P, Sturgeon C. Cost-effectiveness analysis of parathyroidectomy for asymptomatic primary hyperparathyroidism. Surgery 2006;140:874-81; discussion 881-2. Employment, Hours, and Earnings from the Current Employment Statistics survey (National). 2011. Available from http://data.bls.gov/timeseries/CES0500000003?data_ tool=XGtable. Sejean K, Calmus S, Durand-Zaleski I, et al. Surgery versus medical follow-up in patients with asymptomatic primary hyperparathyroidism: a decision analysis. Eur J Endocrinol 2005;153:915-27. Chisholm EJ, Kulinskaya E, Tolley NS. Systematic review and meta-analysis of the adverse effects of thyroidectomy combined with central neck dissection as compared with thyroidectomy alone. Laryngoscope 2009;119:1135-9. Gyorki DE, Untch B, Tuttle RM, Shaha AR. Prophylactic central neck dissection in differentiated thyroid cancer: an assessment of the evidence. Annf Surg Oncol 2013;20: 2285-9. Hughes DT, White ML, Miller BS, Gauger PG, Burney RE, Doherty GM. Influence of prophylactic central lymph node dissection on postoperative thyroglobulin levels and radioiodine treatment in papillary thyroid cancer. Surgery 2010;148:1100-6; discussion 1006-7.
Surgery Volume 154, Number 6
26. Zetoune T, Keutgen X, Buitrago D, et al. Prophylactic central neck dissection and local recurrence in papillary thyroid cancer: a meta-analysis. Ann Surg Oncol 2010;17: 3287-93. 27. Sun GH, Demonner S, Davis MM. Epidemiological and economic trends in inpatient and outpatient thyroidectomy in the United States, 1996-2006. Thyroid 2013;23:727-33. 28. Moo TA, McGill J, Allendorf J, Lee J, Fahey T 3rd, Zarnegar R. Impact of prophylactic central neck lymph node dissection on early recurrence in papillary thyroid carcinoma. World J Surg 2010;34:1187-91.
DISCUSSION Dr Gerard M. Doherty (Boston, MA): My question or criticism is whether you could make this a little bit more complicated. You considered operation in isolation, as though what we do and the information that we gather does not affect the rest of the patient’s treatment. And, in fact, the American Thoracic Association guidelines dictate that the endocrinologist or whoever is deciding about radioactive iodine treatment and about the length and intensity of thyroid-stimulating hormone suppression considered the pathology results in that. Is there a way to make your model consider the value of CND in preventing unnecessary radioiodine treatment, for example, and considering the complications as to that? Dr Kyle Zanocco: That’s a fantastic question. The world-renowned statistician George Box just passed away approximately 2 weeks ago. He was famous for saying that ‘‘all models are wrong, some are useful.’’ I would say that our model is certainly wrong in the fact that there are a lot of nuances and variables that were not included, but I think it is a useful model in the sense that it captures the key decisions that are affected. When we compared pCND versus no pCND, we did assume that all patients received identical treatment after the initial operation, which is an imperfect assumption. I think that the key variables are captured. Dr Sareh Parangi (Boston, MA): I was wondering whether you did do evaluations in which you varied the patient age, for example, to older than 45, and whether that alone may increase the risk of recurrence to greater than 11%? Dr Kyle Zanocco: Excellent question. Yes, we did do sensitivity analysis on patient age. We looked at a variety of ages, and we did not find our conclusion to be sensitive to age. However, the staging criteria do change at 45, and that is a limitation of our model. We didn’t find a patient, even older than 45 years of age, who, given
Zanocco, Elaraj, and Sturgeon 1155
the clinical scenario studied, had a recurrence rate that was greater than our threshold value. Dr Orlo H. Clark (San Francisco, CA): Did you talk about unilateral or bilateral pCND, and does it make a difference? Dr Kyle Zanocco: In our analysis, we focused on a unilateral tumor with a unilateral pCND. The extent of pCND would make a difference. Although the data are not very good, we suspect that a bilateral compartment dissection would have had a greater rate of complications than what we modeled. Dr Thomas J. Fahey, III (New York, NY): I would like to echo what Dr Doherty said. I think it’s a very important omission to exclude whether or to give radioactive iodine as a consequence or as a result of what you find on CND in your model. And it would be very interesting to see how that affects that ultimate cost effectiveness. So you didn’t model that at all in the paper? Dr Kyle Zanocco: In this paper, we assumed identical treatment after the initial operation. I agree that that is an important consideration. And it is something that we should look into further with a more sophisticated model. Dr Jyotirmay Sharma (Atlanta, GA): What do you think about the utilization as far as the hospital utilization and cost because of duration of stay? You see, $1,000 or $1,500 a night really changes those numbers significantly. And the duration of stay is an important aspect of any cost analysis. Was that something that you considered at all in this analysis? Dr Kyle Zanocco: We used the diagnosis-related group (ie, DRG) system for the hospital costs of the two groups, and costs were calculated from the nationwide inpatient sample charges and a cost-to-charge ratio. We didn’t specifically look at duration of stay, but duration of stay is bundled into the average cost for the two different compared groups. Therefore, if there was a slightly increased duration of stay based on undergoing additional operation, the neck dissection, the additional duration of stay would have been included in the cost. Dr Lorend Aguilera (Philadelphia, PA): The part of the measuring quality is supposedly the hardest thing in cost effectiveness analysis. Was measurement of quality on the basis of expert judgment, or a prospective validated quality measure? Dr Kyle Zanocco: We obtained the quality adjustment factors from published literature. Those studies used the time tradeoff method to come up with them.
Background. Routine prophylactic central neck dissection (pCND) after total thyroidectomy (TTX) for low-risk papillary thyroid cancer (PTC) offers the potential to decrease disease recurrence but may increase operative complications. We hypothesized that routine pCND is not cost-effective in low-risk PTC. Methods. A Markov transition-state model was constructed to compare TTX with and without pCND. Outcome probabilities, utilities, and costs were estimated on the basis of literature review. The threshold for cost-effectiveness was $100,000 per quality-adjusted life year. Sensitivity analysis was used to examine model uncertainty. Results. pCND cost $10,315 and produced an effectiveness of 23.785 quality-adjusted life years. This strategy was more costly and less effective than TTX without pCND and was therefore dominated. pCND became cost-effective when the probability of recurrence increased from 6% to 10.3%, cost of reoperation for recurrence increased from $8,900 to $26,120, or added probabilities of recurrent laryngeal nerve injury and hypoparathyroidism due to pCND were less than 0.20% and 0.18% during 2-way sensitivity analysis. Monte Carlo simulation showed that pCND was not cost-effective in 97.3% of iterations. Conclusion. Routine pCND for low-risk PTC is not cost-effective unless the recurrence rate is greater than 10.3%. Application of pCND should be individualized based on risk of recurrence and added complications. (Surgery 2013;154:1148-55.) From the Department of Surgery,a Section of Endocrine Surgery, and Center for Healthcare Studies,b Northwestern University Feinberg School of Medicine, Chicago, IL
CERVICAL LYMPH NODE METASTASES are common in papillary thyroid cancer (PTC). PTC-positive nodes in the central neck are identified frequently after operative resection in patients who appear to be clinically free of nodal disease. In these patients, an estimated 37 64% will have microscopically identifiable lymph node metastases.1 However, these metastases appear to be of limited importance, given that the locoregional recurrence rate for this group is much lower (approximately 6%)
Kyle Zanocco’s participation in this study was supported in part by the Northwestern University Feinberg School of Medicine Center for Healthcare Studies under an institutional award from the Agency for Healthcare Research and Quality, T-32 HS 000078 (PI: Jane L. Holl, MD, MPH). Accepted for publication June 21, 2013. Reprint requests: Cord Sturgeon, MD, Department of Surgery, Northwestern University Feinberg School of Medicine, 676 N St. Clair St. Ste 650, Chicago, IL 60611. E-mail: csturgeo@ nmh.org. 0039-6060/$ - see front matter Ó 2013 Mosby, Inc. All rights reserved. http://dx.doi.org/10.1016/j.surg.2013.06.016
1148 SURGERY
even if central neck dissection is not performed after initial thyroidectomy.1,2 Routine prophylactic central neck dissection (pCND) in clinically and intraoperatively nodenegative PTC is controversial. Proponents of this practice argue that pCND decreases the need for reoperative neck surgery by reducing locoregional recurrence and simplifies follow-up by lowering postoperative serum thyroglobulin.2-4 Opponents maintain that this procedure exposes patients to additional risk of recurrent laryngeal nerve (RLN) injury and hypoparathyroidism without proven benefit.2 A prospective, randomized trial comparing pCND versus no pCND is the ideal method for assessing these purported benefits and risks. However, this approach is not feasible given the prohibitively large patient sample size necessary to detect the hypothesized clinical differences in recurrence and complications between the 2 groups.2 Given the unavailability of randomized controlled data for the foreseeable future, we sought to model these outcomes using decision analysis. We hypothesized that routine pCND is not cost-effective in low-risk PTC.
Surgery Volume 154, Number 6
METHODS Reference case scenario. The reference case patient in this analysis was defined as a 40-year-old patient with a 2-cm, noninvasive PTC that was without preoperative or intraoperative evidence of nodal involvement. This patient was assumed to be without a history of previous neck surgery and an appropriate candidate for thyroidectomy via cervical incision. Decision model. A Markov transition-state decision model was constructed for the reference case using decision analysis software (TreeAge Pro; TreeAge Software, Inc, Williamstown, MA, 2012). Two alternatives were created: (1) total thyroidectomy alone or (2) total thyroidectomy with pCND. The outcome probabilities for these alternatives were estimated from literature review and are listed in Table I.5-11 The model had a 1-year cycle length and was terminated when the number of completed cycles exceeded patient life expectancy. The lifetime risk of recurrence without pCND was estimated to be 6% (1.2% per year for 5 years of follow-up).1,2 All recurrences were assumed to be definitively treated with reoperative neck dissection of affected compartments.4 The relative risk reduction of recurrence after pCND, compared with total thyroidectomy without pCND, was assumed to be 75%. A large-risk reduction was used to maintain a conservative overall model perspective given the hypothesis that pCND was not cost-effective. The strategy that provided the greatest quality-adjusted life expectancy (QALE) without exceeding an incremental costeffectiveness ratio (ICER) of $100,000 per qualityadjusted life year (QALY) was defined as optimal. We chose $100,000 per QALY as the threshold for cost-effectiveness on the basis of analysis of the cost of current health care resource allocation decisions in the United States and currentlyaccepted convention.12 Cost estimation. A combination of micro- and gross costing techniques was used to estimate the costs of resources consumed for each strategy from the societal perspective. Only costs that differed between the alternative strategies were included.13,14 All costs in the model were reported in 2010 U.S. dollars. An inflation rate of 4.1%, calculated as the mean of annual changes in the Consumer Price Index for Medical Care from 2000 to 2010, was used to adjust older costs to their 2010 amount when necessary.15 A discount rate of 3% was applied to all future costs in the model.13 Cost estimates for physician labor were determined by the 2010 national Medicare charge limits for the surgery, anesthesiology, and pathology
Zanocco, Elaraj, and Sturgeon 1149
services rendered in each strategy.16 Hospital costs were estimated by calculating weighted averages of 2009 Medicare cost-to-charge ratios for relevant diagnosis-related groups and multiplying these ratios by national median impatient charge estimates for thyroid surgery and neck lymph node dissection reported by the 2009 Nationwide Inpatient Sample.17,18 Thyroid hormone-replacement costs were based on average U.S. wholesale prices.19 The costs for the thyroidectomy-related complications of permanent RLN injury and permanent hypoparathyroidism have been estimated in previous work and were adjusted to 2010 costs.20 Missed patient work due to surgery was assigned an hourly cost based on 2010 annual US hourly earnings.21 All of the costs used in the model are listed in Table I. Effectiveness. QALE was used as the measure of effectiveness. Each of the model’s different outcomes was assigned a corresponding quality adjustment factor based on literature review. The time elapsed in a particular outcome was multiplied by that outcome’s quality adjustment factor to yield effectiveness in QALYs. Patients who developed recurrent disease were assigned a one-time treatment-related reduction in quality-adjusted life expectancy.22 No other survival difference was assumed between the 2 management strategies in the reference case.1,2,23,24 All future QALYs were discounted at 3%.13 The quality adjustment factors used in the model are listed with supporting references in Table I. Sensitivity analysis. Threshold analysis was performed on each variable in the reference case to identify values at which the ICER exceeded $50,000/QALY and $100,000/QALY during oneway sensitivity analysis. Two-way sensitivity analysis was performed to examine the combined impact of changes in lifetime recurrence after total thyroidectomy without pCND and recurrence risk reduction with additional pCND. The combined effects of the additional probabilities of RLN injury and permanent hypoparathyroidism with pCND were also evaluated with 2-way analysis. Probabilistic sensitivity analysis was performed with Monte Carlo simulation, where triangular frequency distributions for each variable were simultaneously sampled during 1,000 consecutive iterations. In this simulation, all variables were allowed to vary up to ±50% of the reference case estimate or, if not possible in the case of probability and utility estimates, to the greatest extent such that the reference case value remained at the center of a range containing the boundary 0 or 1.
1150 Zanocco, Elaraj, and Sturgeon
Surgery December 2013
Table I. Model assumptions Variables Probabilities Lifetime recurrence Relative risk reduction of recurrence after pCND RLN injury after total thyroidectomy Additional risk of RLN injury after pCND Permanent hypoparathyroidism after total thyroidectomy Additional risk of permanent hypoparathyroidism after pCND RLN injury after reoperation for recurrence Permanent hypoparathyroidism after reoperation for recurrence Costs in U.S. dollars Total thyroidectomy Additional cost of pCND Reoperation for recurrence Treatment of RLN injury Annual cost of permanent hypoparathyroidism Annual cost of thyroid hormone replacement Hourly cost of missed patient work Quality-of-life adjustment factors Hypothyroid on thyroid hormone replacement Hypothyroid with RLN injury Hypothyroid with hypoparathyroidism Hypothyroid with hypoparathyroidism and RLN injury 1-time reduction in quality-adjusted life expectancy for undergoing reoperation for recurrence, d Time Work missed to undergo reoperation for recurrence, h Life expectancy, y
Reference case value 6% 75%
Discount rate
Range of values used in Monte Carlo simulation distributions 3% to 9% 50% to 100%
Sources 1,2 3
5
2% 0.25% 1%
1% to 3% 0.125% to 0.375% 0.5% to 1.5%
0.25%
0.125% to 0.375%
2,26
1.5% to 4.5% 1.5% to 4.5%
24
$5,500 $513 $8,887 $10,367 $863
$2,750 to $8,250 $256 to $770 $4,444 to $13,331 $5,183 to $15,550 $432 to $1,295
16,18
$152 $23
$76 to $227 $11 to $34
19
0.990
0.980 to 1
10
0.881 0.893 0.775
0.762 to 1 0.786 to 1 0.550 to 1
10,22
11
5.5 to 16.5
Expert opinion
16
8 to 24
Expert opinion
40
20 to 60
3,11
3%
1.5% to 4.5%
13
3% 3%
2,26 6-9
24
16,18 16-18 20 20
21
10,22 10,22
pCND, Prophylactic central neck dissection; RLN, recurrent laryngeal nerve.
RESULTS Reference case. Total thyroidectomy without pCND was the less costly strategy with an expected cost of $10,149 and an effectiveness of 23.791 QALYs. The strategy of total thyroidectomy with routine pCND had an expected cost of $10,315 and an effectiveness of 23.785 QALYs. The addition of routine pCND resulted in an expected incremental cost of $166 and a loss of 0.006 QALYs. Routine pCND was therefore dominated because this strategy was both more costly and less effective than total thyroidectomy alone.
Sensitivity analysis. The threshold values of all the model variables that were necessary to make routine pCND cost-effective during one-way sensitivity analysis are listed in Table II. The model was most sensitive to lifetime recurrence rate, with pCND becoming the optimal strategy if recurrence after total thyroidectomy alone exceeded 10.3% (Fig 1). pCND also became cost-effective if the cost of treatment for recurrence exceeded $26,120 (Fig 2) or the reduction in QALE due to recurrence exceeded 76 days. At the $100,000/ QALY threshold level, the model was not sensitive
Zanocco, Elaraj, and Sturgeon 1151
Surgery Volume 154, Number 6
Table II. $100,000/QALY threshold values for model variables that make prophylactic central neck dissection cost-effective Variable Probabilities Lifetime recurrence Relative risk reduction of recurrence after pCND RLN injury after total thyroidectomy Additional risk of RLN injury after pCND Permanent hypoparathyroidism after total thyroidectomy Additional risk of permanent hypoparathyroidism after total thyroidectomy RLN injury after reoperation for recurrence Permanent hypoparathyroidism after reoperation for recurrence Costs in U.S. dollars Total thyroidectomy Additional cost of pCND Reoperation for recurrence Treatment of RLN injury Annual cost of permanent hypoparathyroidism Annual cost of thyroid hormone replacement Hourly cost of missed patient work Quality-of-life adjustment factors Hypothyroid on thyroid hormone replacement Hypothyroid with RLN injury Hypothyroid with hypoparathyroidism Hypothyroid with hypoparathyroidism and RLN injury 1-time reduction in quality-adjusted life expectancy for undergoing reoperation for recurrence (d) Time Work missed to undergo reoperation for recurrence (h) Life expectancy (y) Discount rate
$50,000/QALY threshold
$100,000/QALY threshold
>9.6% None None None None None
>10.3% None None None None None
>18.8% >16.5%
>15.5% >15.1%
None <$52 >$19,480 None None None >$680
None None >$26,120 None None None >$1,100
<0.858 None None None >90
<0.882 None None None >75.9
>482 None None
>780 None None
pCND, Prophylactic central neck dissection; RLN, recurrent laryngeal nerve.
to life expectancy, the quality adjustment factors for RLN injury or hypoparathyroidism, the additional cost of pCND, or the relative risk reduction of recurrence after pCND. Two-way sensitivity analysis demonstrated that the decision model was sensitive to simultaneous changes in several clinically relevant variables. If a sufficiently large risk of recurrence after total thyroidectomy without pCND was combined with a substantial risk reduction after pCND, routine pCND was cost-effective (Fig 3). Figure 4 shows that routine pCND also became cost-effective when combinations of the additional risk of RLN injury and permanent hypoparathyroidism with pCND were sufficiently low (eg, when the additional risks were both 0.08%). Monte Carlo simulation showed total thyroidectomy without pCND to be the optimal strategy in 973 (97.3%) of the iterations. Routine pCND was both more costly and less effective (ie, dominated)
in 845 iterations (84.5%). In an additional 118 iterations (11.8%), forgoing pCND resulted in gains in quality of life that had an incremental cost that was low enough to produce an ICER of less than $100,000 QALY. Of the 27 (2.7%) cases in which routine pCND was cost-effective, this strategy was less costly and less effective compared with total thyroidectomy alone in 1 (0.1%) and dominant (ie, less costly and more effective) in 7 (0.7%). All of the iterations from this simulation are plotted in Fig 5. DISCUSSION This is the first cost-effectiveness analysis of routine pCND for PTC. We found that routine pCND for low-risk PTC is more costly and less effective than total thyroidectomy alone. However, pCND would be cost-effective if the PTC recurrence rate exceeds 10.3%, the cost of treatment for recurrence exceeds $26,120, or the additional risks
1152 Zanocco, Elaraj, and Sturgeon
Fig 1. One-way sensitivity analysis of the ICER of total thyroidectomy without routine pCND as a function of lifetime recurrence rate. The dashed line represents the 100,000/QALY threshold for cost-effectiveness. When the recurrence rate after total thyroidectomy alone exceeded 10.3%, routine pCND became costeffective.
Fig 2. One-way sensitivity analysis displaying the ICER of total thyroidectomy without routine pCND as a function of the cost of reoperation for recurrence. The dashed line represents the 100,000/QALY threshold for costeffectiveness. When the cost of reoperation exceeded $26,120, routine pCND became cost-effective.
of RLN injury and permanent hypoparathyroidism with this procedure are negligible. Although to our knowledge, no previous authors have examined the costs of pCND, several recent retrospective analyses have compared outcomes of PTC patients who underwent total thyroidectomy with and without this additional procedure.3,23,25,26 One of these studies demonstrated that the 20 CLND procedures were required to prevent 1 central compartment reoperation.3 Two meta-analyses have demonstrated an increased rate of transient hypocalcemia after pCND without showing a difference in permanent hypoparathyroidism or RLN injury.23,26 Both of
Surgery December 2013
Fig 3. This 2-way sensitivity analysis demonstrates the combined impact of changes in lifetime recurrence after total thyroidectomy without pCND and recurrence risk reduction after the addition of pCND. The dashed threshold line represents combinations of these attributes that produce an ICER of $100,000 per QALY. The area to the right of the threshold line represents values that would make pCND a cost-effective strategy. The reference case assumptions intersect to the left of the threshold line in the area representing values where pCND is not cost-effective.
Fig 4. This 2-way sensitivity analysis demonstrates the combined effects of changes in the additional risk of RLN injury and permanent hypoparathyroidism with pCND. The dashed threshold line represents combinations of these risks that produce an ICER of $100,000 per QALY. The area to the left of the threshold line represents values that would make pCND a cost-effective strategy. The reference case risks intersect to the right of the threshold line in the area representing values where pCND is not cost-effective.
these studies had total sample sizes of 1,100 1,200 patients, far less than the estimated 2,000 5,000 needed to detect 1% or greater differences in the permanent complications with adequate statistical power.2 Given that the increased risk of these complications with pCND
Surgery Volume 154, Number 6
Zanocco, Elaraj, and Sturgeon 1153
Fig 5. A 1,000 iteration Monte Carlo simulation in which all model assumptions were simultaneously varied across their frequency distributions. The incremental cost and incremental effectiveness of total thyroidectomy without versus with pCND are plotted for each iteration. Points on the left of the dashed 100,000/QALY threshold line represent iterations where pCND was cost-effective. To the right of the line are iterations where pCND was not cost-effective. Total thyroidectomy without pCND was cost-effective in 97.3% of the iterations and the dominant strategy in 84.5% of the iterations, whereas total thyroidectomy with pCND was cost-effective in 2.7% of iterations and dominant in 0.7% of the iterations.
may be even smaller than 1%, an even larger sample size would be necessary before it would be appropriate to conclude that no difference exists with the addition of this procedure. Several limitations are evident in our study. The decision model omitted the operative complications of wound infection, hematoma formation, and the possibility of operative mortality. These omissions are unlikely to have significant impact on the results given that both strategies involved surgery. However, these events may be more likely during reoperation, making treatment of locoregional recurrence less favorable, and therefore improving the cost-effectiveness of the pCND strategy. The effects of temporary RLN injury and temporary hypocalcemia also were omitted because of their relatively low expected impact on overall cost and quality-of-life. If these temporary complications were more likely to occur with pCND, their inclusion in the model would have further increased the dominance of the thyroidectomy alone strategy. Although our model assumed all surgery was performed on an inpatient basis, the frequency of lower-cost outpatient thyroidectomy has markedly increased during the past decade.27 If the addition of pCND converted an outpatient thyroidectomy into an inpatient procedure, the additional cost of pCND would have been $1,173 instead of $513, based reimbursement data from the
Medicare Hospital Outpatient Prospective Payment System. This additional cost would have increased the dominance of the thyroidectomy alone strategy. Our model assumed no difference in postoperative treatment after initial operation between those who did and did not receive pCND; however, there is evidence that pCND leads to greater radioactive iodine (RAI) doses because of the detection of involved cervical lymph nodes.25,28 Adding the cost of additional RAI for these patients would have further diminished the costeffectiveness of pCND and would not have changed our conclusion. Conversely, if pCND led to less postoperative RAI administration through the demonstration of a node-negative central compartment, the overall cost of the pCND strategy would have been lower; however, the costsavings from less RAI would have needed to be greater than $750 per patient before changing the model’s overall conclusion. Although the model’s cost of recurrence included the procedural and ancillary costs of hospitalization for reoperation, some other costs were omitted for simplicity, including additional transportation to undergo reoperation and additional medical follow-up. Given that the overall cost of treatment for recurrence could have increased by more than $20,000 before pCND became costeffective, it is unlikely that these omissions would
1154 Zanocco, Elaraj, and Sturgeon
change the model’s conclusion. The model also did not allow for lower-cost alternative treatments for recurrence, including observation or RAI administration. Replacing some of the reoperations with these alternatives would have lowered the overall strategy cost of initial total thyroidectomy without pCND and increased this option’s dominance. The operative complications of RLN injury and permanent hypoparathyroidism have a wide spectrum of severity and treatment costs, but these considerations were simplified into single gross cost and quality of life estimates. We believe the impact of these simplifications is adequately addressed with sensitivity analysis. Comprehensive bilateral central compartment dissection may improve survival in higher-risk PTC patients.4 Our model did not assume a survival difference between low-risk PTC patients undergoing total thyroidectomy with or without pCND. However, it did show with sensitivity analysis that pCND would be cost-effective if disease recurrence reduced QALE by 76 days or greater. This finding suggests that central compartment dissection is cost-effective in cases where this technique provides patents a survival advantage. In conclusion, routine pCND for low-risk PTC is not cost-effective. Selective application of pCND should be individualized based on a balanced evaluation of the anticipated rate of recurrence versus additional risk of morbidity due to this procedure. REFERENCES 1. Randolph GW, Duh QY, Heller KS, et al. The prognostic significance of nodal metastases from papillary thyroid carcinoma can be stratified based on the size and number of metastatic lymph nodes, as well as the presence of extranodal extension. Thyroid 2012;22:1144-52. 2. Carling T, Carty SE, Ciarleglio MM, et al. American Thyroid Association design and feasibility of a prospective randomized controlled trial of prophylactic central lymph node dissection for papillary thyroid carcinoma. Thyroid 2012; 22:237-44. 3. Popadich A, Levin O, Lee JC, et al. A multicenter cohort study of total thyroidectomy and routine central lymph node dissection for cN0 papillary thyroid cancer. Surgery 2011;150:1048-57. 4. Cooper DS, Doherty GM, Haugen BR, et al. Revised American Thyroid Association management guidelines for patients with thyroid nodules and differentiated thyroid cancer. Thyroid 2009;19:1167-214. 5. Lo CY, Kwok KF, Yuen PW. A prospective evaluation of recurrent laryngeal nerve paralysis during thyroidectomy. Arch Surg 2000;135:204-7. 6. Dobrinja C, Trevisan G, Piscopello L, Fava M, Liguori G. Comparison between thyroidectomy and hemithyroidectomy in treatment of single thyroid nodules identified as indeterminate follicular lesions by fine-needle
Surgery December 2013
7.
8.
9.
10.
11.
12.
13. 14. 15.
16.
17.
18.
19. 20.
21.
22.
23.
24.
25.
aspiration cytology. Ann Ital Chir 2010;81:403-10; discussion 410-1. Page C, Strunski V. Parathyroid risk in total thyroidectomy for bilateral, benign, multinodular goitre: report of 351 surgical cases. J Laryngol Otol 2007;121:237-41. Sitges-Serra A, Ruiz S, Girvent M, Manjon H, Duenas JP, Sancho JJ. Outcome of protracted hypoparathyroidism after total thyroidectomy. Br J Surg 2010;97:1687-95. Youngwirth L, Benavidez J, Sippel R, Chen H. Parathyroid hormone deficiency after total thyroidectomy: incidence and time. J Surg Res 2010;163:69-71. Muennig P, Khan K. Designing and conducting costeffectiveness analyses in medicine and health care. 1st ed. San Francisco (CA): Jossey-Bass; 2002. Bell FC, Miller ML. Life Tables for the United States Social Security Area 1900-2100, Actuarial Study No. 120 2011. Available from http://www.socialsecurity.gov/ OACT/NOTES/as120/LifeTables_Tbl_6_2010.html. Braithwaite RS, Meltzer DO, King JT Jr, Leslie D, Roberts MS. What does the value of modern medicine say about the $50,000 per quality-adjusted life-year decision rule? Med Care 2008;46:349-56. Gold MR. Cost-effectiveness in health and medicine. New York (NY): Oxford University Press; 1996. Muennig P. Cost-effectiveness analyses in health: a practical approach. 2nd ed. San Francisco (CA): Jossey-Bass; 2008. Consumer Price Index - Medical Care. 2011. Available from http://data.bls.gov/timeseries/CUUR0000SAM? output_view=pct_12mths. Overview of the Medicare Physician Fee Schedule Search. 2011. Available from https://www.cms.gov/apps/physicianfee-schedule/overview.aspx. HCUPnet: A tool for identifying, tracking, and analyzing national hospital statistics. 2011. Available from http:// hcupnet.ahrq.gov/. MEDPAR Medicare Fee for Service for Parts A & B. 2011. Available from http://www.cms.gov/MedicareFeeforSvcPartsAB/ 03_MEDPAR.asp#TopOfPage. Thomson Healthcare. Red book. Montvale (NJ): Thomson PDR; 2010. v. Zanocco K, Angelos P, Sturgeon C. Cost-effectiveness analysis of parathyroidectomy for asymptomatic primary hyperparathyroidism. Surgery 2006;140:874-81; discussion 881-2. Employment, Hours, and Earnings from the Current Employment Statistics survey (National). 2011. Available from http://data.bls.gov/timeseries/CES0500000003?data_ tool=XGtable. Sejean K, Calmus S, Durand-Zaleski I, et al. Surgery versus medical follow-up in patients with asymptomatic primary hyperparathyroidism: a decision analysis. Eur J Endocrinol 2005;153:915-27. Chisholm EJ, Kulinskaya E, Tolley NS. Systematic review and meta-analysis of the adverse effects of thyroidectomy combined with central neck dissection as compared with thyroidectomy alone. Laryngoscope 2009;119:1135-9. Gyorki DE, Untch B, Tuttle RM, Shaha AR. Prophylactic central neck dissection in differentiated thyroid cancer: an assessment of the evidence. Annf Surg Oncol 2013;20: 2285-9. Hughes DT, White ML, Miller BS, Gauger PG, Burney RE, Doherty GM. Influence of prophylactic central lymph node dissection on postoperative thyroglobulin levels and radioiodine treatment in papillary thyroid cancer. Surgery 2010;148:1100-6; discussion 1006-7.
Surgery Volume 154, Number 6
26. Zetoune T, Keutgen X, Buitrago D, et al. Prophylactic central neck dissection and local recurrence in papillary thyroid cancer: a meta-analysis. Ann Surg Oncol 2010;17: 3287-93. 27. Sun GH, Demonner S, Davis MM. Epidemiological and economic trends in inpatient and outpatient thyroidectomy in the United States, 1996-2006. Thyroid 2013;23:727-33. 28. Moo TA, McGill J, Allendorf J, Lee J, Fahey T 3rd, Zarnegar R. Impact of prophylactic central neck lymph node dissection on early recurrence in papillary thyroid carcinoma. World J Surg 2010;34:1187-91.
DISCUSSION Dr Gerard M. Doherty (Boston, MA): My question or criticism is whether you could make this a little bit more complicated. You considered operation in isolation, as though what we do and the information that we gather does not affect the rest of the patient’s treatment. And, in fact, the American Thoracic Association guidelines dictate that the endocrinologist or whoever is deciding about radioactive iodine treatment and about the length and intensity of thyroid-stimulating hormone suppression considered the pathology results in that. Is there a way to make your model consider the value of CND in preventing unnecessary radioiodine treatment, for example, and considering the complications as to that? Dr Kyle Zanocco: That’s a fantastic question. The world-renowned statistician George Box just passed away approximately 2 weeks ago. He was famous for saying that ‘‘all models are wrong, some are useful.’’ I would say that our model is certainly wrong in the fact that there are a lot of nuances and variables that were not included, but I think it is a useful model in the sense that it captures the key decisions that are affected. When we compared pCND versus no pCND, we did assume that all patients received identical treatment after the initial operation, which is an imperfect assumption. I think that the key variables are captured. Dr Sareh Parangi (Boston, MA): I was wondering whether you did do evaluations in which you varied the patient age, for example, to older than 45, and whether that alone may increase the risk of recurrence to greater than 11%? Dr Kyle Zanocco: Excellent question. Yes, we did do sensitivity analysis on patient age. We looked at a variety of ages, and we did not find our conclusion to be sensitive to age. However, the staging criteria do change at 45, and that is a limitation of our model. We didn’t find a patient, even older than 45 years of age, who, given
Zanocco, Elaraj, and Sturgeon 1155
the clinical scenario studied, had a recurrence rate that was greater than our threshold value. Dr Orlo H. Clark (San Francisco, CA): Did you talk about unilateral or bilateral pCND, and does it make a difference? Dr Kyle Zanocco: In our analysis, we focused on a unilateral tumor with a unilateral pCND. The extent of pCND would make a difference. Although the data are not very good, we suspect that a bilateral compartment dissection would have had a greater rate of complications than what we modeled. Dr Thomas J. Fahey, III (New York, NY): I would like to echo what Dr Doherty said. I think it’s a very important omission to exclude whether or to give radioactive iodine as a consequence or as a result of what you find on CND in your model. And it would be very interesting to see how that affects that ultimate cost effectiveness. So you didn’t model that at all in the paper? Dr Kyle Zanocco: In this paper, we assumed identical treatment after the initial operation. I agree that that is an important consideration. And it is something that we should look into further with a more sophisticated model. Dr Jyotirmay Sharma (Atlanta, GA): What do you think about the utilization as far as the hospital utilization and cost because of duration of stay? You see, $1,000 or $1,500 a night really changes those numbers significantly. And the duration of stay is an important aspect of any cost analysis. Was that something that you considered at all in this analysis? Dr Kyle Zanocco: We used the diagnosis-related group (ie, DRG) system for the hospital costs of the two groups, and costs were calculated from the nationwide inpatient sample charges and a cost-to-charge ratio. We didn’t specifically look at duration of stay, but duration of stay is bundled into the average cost for the two different compared groups. Therefore, if there was a slightly increased duration of stay based on undergoing additional operation, the neck dissection, the additional duration of stay would have been included in the cost. Dr Lorend Aguilera (Philadelphia, PA): The part of the measuring quality is supposedly the hardest thing in cost effectiveness analysis. Was measurement of quality on the basis of expert judgment, or a prospective validated quality measure? Dr Kyle Zanocco: We obtained the quality adjustment factors from published literature. Those studies used the time tradeoff method to come up with them.