Annual financial impact of well-differentiated thyroid cancer care in the United States
Carrie C. Lubitz MD, MPH,
Harvard Medical School, Boston, Massachusetts
Department of Surgery, Massachusetts General Hospital, Boston, Massachusetts
Institute for Technology Assessment, Massachusetts General Hospital, Boston, Massachusetts
Corresponding author: Carrie C. Lubitz, MD, MPH, Assistant Professor of Surgery, Harvard Medical School, Division of General Surgery, Massachusetts General Hospital, 55 Fruit Street, Yawkey 7B, Boston, MA 02114-3117; Fax: (617) 724-3895; email@example.com
Well-differentiated thyroid cancer (WDTC) is a prevalent disease, which is increasing in incidence faster than any other cancer. Substantial direct medical care costs are related to the diagnosis and treatment of newly diagnosed patients as well as the ongoing surveillance of patients who have a long life expectancy. Prior analyses of the aggregate health care costs attributable to WDTC in the United States have not been reported.
A stacked cohort cost analysis was performed on the US population from 1985 to 2013 to estimate the number of WDTC survivors in 2013. Incidence rates, and cancer-specific and overall survival were based on Surveillance, Epidemiology, and End Results (SEER) data. Current and projected direct medical care costs attributable to the care of patients with WDTC were then estimated. Health care–related costs and event probabilities were based on Medicare reimbursement schedules and the literature.
Estimated overall societal cost of WDTC care in 2013 for all US patients diagnosed after 1985 is $1.6 billion. Diagnosis, surgery, and adjuvant therapy for newly diagnosed patients (41%) constitutes the greatest proportion of costs, followed by surveillance of survivors (37%), and nonoperative death costs attributable to thyroid cancer care (22%). Projected 2030 costs (in 2013 US dollars) based on current incidence trends exceed $3.5 billion.
Thyroid cancer is a common disease whose incidence is increasing by 6.6% annually in the United States, a rate which is more than any other cancer.[1, 2] Since 1973, the annual incidence has increased by more than 500%. It is estimated that 60,220 new cases of thyroid cancer will be diagnosed in 2013, and that it is the fifth most common cancer diagnosed among women. There are more than 530,000 thyroid cancer survivors living in the United States today[1, 3-7]; 80% of new thyroid cancer patients are < 65 years of age and the 20-year disease-specific survival is more than 90%. Given these data and the aging trends in the population, it is likely that the prevalence of thyroid cancer will continue to rise.[3, 8, 9]
The current and projected costs of caring for patients with other cancers in the United States are substantial, with overall cancer care costs estimated at $157 billion in 2010. Prior research has shown that the majority of these cancer care costs are related to the beginning and end of care, with a relatively lower cost of continuing care.[10-13] Given the unique trends in incidence and survival in the majority of patients with thyroid cancer, we hypothesized that thyroid cancer costs would have a different pattern.
Well-differentiated thyroid cancer (WDTC; comprising papillary thyroid carcinoma, follicular thyroid carcinoma, and Hürthle cell thyroid carcinoma) makes up about 95% of all thyroid cancers. Our aim was to determine current and projected health care–related costs attributable to WDTC care. To our knowledge, the aggregate national cost of WDTC care has not been reported. Given resource constraints, costs of thyroid cancer care may have important policy implications. Therefore, understanding these costs is essential for performance of future comparative effectiveness analyses.
MATERIALS AND METHODS
Using the Surveillance, Epidemiology, and End Results (SEER) SEER*Stat version 8.0.4 and observed sex-specific rates for incidence, death from WDTC, and survival with WDTC, we estimated the prevalence of patients with WDTC in the United States in 2013. We excluded patients with anaplastic and medullary thyroid carcinomas given the distinctive natural history, treatment, and costs of these rare, more aggressive subtypes of thyroid cancer. We included papillary thyroid microcarcinomas (PTMCs; AJCC/UICC TNM classification T1a), because, in practice, > 99% of patients with PTMC reported in SEER undergo surgery for their disease, purporting the same costs as papillary thyroid carcinoma (PTC) > 1 cm. Moreover, PTMCs do exhibit aggressive behavior and metastasize. Next, we used observed all-cause survival for included patients, which is an estimate of the probability of surviving all causes of death from 1985 to 2010 (survival estimates from SEER assume that the general population dies of competing causes at the same rate as patients with WDTC), and cause-specific survival conditional on year of diagnosis to identify the number of WDTC-related deaths in 2013.
WDTC incidence, disease-specific survival, and all-cause mortality were obtained from the SEER registry research data from patients diagnosed from 1985 to 2010 based on cancer site and histological type, sex, and year of diagnosis. The SEER program comprises 18 cancer registries that collect data on approximately 28% of the US population. US annual population data were obtained from the US Census Bureau. SEER annual incidence rates were then multiplied by US Census Bureau estimates to approximate yearly incident cases.
Clinical care practices were based on American Thyroid Association guideline recommendations, reported practice patterns within the appropriate literature, and the expert opinion of one of the authors (G.H.D.).[15-18] For the base-case analysis, patients were assumed to present for an initial consult with an endocrinologist, have a diagnostic fine-needle aspiration biopsy (FNA) under ultrasound guidance, and undergo a total thyroidectomy as per standard of care. Rates of adjuvant radioactive iodine (RAI) as well as the proportion of patients receiving recombinant thyroid stimulating hormone (rTSH) (versus thyroid hormone withdrawal) in preparation for RAI were abstracted from reported clinical practice patterns in the literature (Table 1).[16, 19, 20] Data sources used for rates of tumor recurrence, surgical deaths, and surgical complications are outlined in Table 1.
Table 1. Input Parameters and Probabilities for Base Case Analysis
Hypoparathyroidism and recurrent laryngeal nerve injury were defined as temporary when symptoms lasted less than 6 months following surgery, according to convention (Table 2). Patients were assumed to have 2 follow-up visits within the first year consisting of serum thyroglobulin and TSH tests and neck ultrasound, and once per year thereafter. Those with a local recurrence were assumed to undergo repeat FNA for confirmation and additional surgery. The cost of surgical deaths from thyroidectomy were estimated from the National Inpatient Sample by Vashishta et al and converted to an estimated cost using the 2011 cost-to-charge ratio for short stay by Diagnosis-Related Group for complicated thyroidectomy.[21-23] Costs attributable to WDTC care do not differ significantly by age at diagnosis or current age, but by length of time with the disease and phase of care. A conservative cost of WDTC (nonoperative) death was used in the base-case analysis based on lowest cost estimates from other solid tumors and cost of advanced thyroid cancer in the literature.[10, 24] For consistency with prior cost of cancer care studies, we also define phases of care as initial (within the first year following diagnosis), continuing, and last year of life.[10-12] Because the majority of thyroid cancer tumor recurrences occur within 5 years of initial diagnosis, when estimating recurrence costs for patients in 2013, we accounted only for tumor-recurrence related costs for patients diagnosed after 2009.[25-27]
Table 2. Base Case Cost Estimates
Cost in 2013 US Dollars (Sensitivity Analysis Range)
Physician and facility costs specific for WDTC care were estimated from the Healthcare Common Procedure Coding System (HCPCS), Diagnosis-Related Group (DRG), and/or Ambulatory Payment Classification (APC) as appropriate using Medicare national reimbursement data for 2012 through 2013 (Table 2). Anesthesiology fees were based on average anesthesia time (15-minute increments), 2013 HCPCS Anesthesia Base Units, and the 2013 national anesthesia conversion factor (21.9243). Average wholesale drug prices were obtained from the RED BOOK online via Micromedex 2.0 (Truven Health Analytics, Greenwood Village, Colo). All costs are measured in 2013 US dollars. The costs attributed to WDTC survivors alive in 2013 and the costs of diagnosis and treatment of newly diagnosed and recurrent WDTC patients in 2013 were analyzed. We assumed patient time costs and productivity loss to be negligible and therefore were not included. Only costs associated with malignancy (ie, not biopsies for thyroid nodules found to be benign) were included.
Projected WDTC Prevalence and Cost Estimates
WDTC incidence for years beyond available SEER data (2010) was extrapolated by continuing increasing, linear incidence trends based on Joinpoint regression analysis extrapolation (Joinpoint Regression Program, version 4.0.4.).[28, 29] Numbers of incident cases in the US population were estimated by the product of US population forecasts by sex as projected by the US Census Bureau and the projected incidence rates. All-cause death rates by age were estimated using the Berkeley Mortality Database life tables.[30, 31] All costs, including projected costs, were estimated for a single year, 2013, and were therefore not discounted. Given the relatively stable survival over the study period, survival was not varied in projected estimates.
Given variation in clinical practice, we assessed the effect on cost of a broad range of input values. Sensitivity analyses were performed to assess the stability of results to uncertainty surrounding key parameters, including recurrence rates, surgical complication rates, and costs of surgical and cancer deaths. We assessed reported ranges when available, or 0.5 to 1.5 times the base-case estimate when ranges were unavailable (Tables 1 and 2).
WDTC Incidence, Prevalence, and Survival
After accounting for deaths from other causes, there are an estimated 670,687 individuals in the United States today who were diagnosed with WDTC from 1985 to 2013. These individuals continue to require treatment or surveillance for their disease. Of these individuals 76% are women. The observed overall survival is worse for men with WDTC (Table 3). In the base-case analysis, we estimate 2173 women and 1522 men will die of WDTC in 2013.
Table 3. Well-Differentiated Thyroid Cancer Survivors by Sex and Year of Diagnosis
Alive in 2013
Alive in 2013
Alive in 2013
Estimated 2013 Health Care–Related Costs Associated With Treatment of Thyroid Cancer
The total estimated 2013 costs associated with WDTC care exceed $1.6 billion. Initial treatment, including work-up, surgery, and adjuvant RAI (proportional to reported practice patterns) ($658 million, 41% of costs) and continuing phases ($595 million, 37% of costs) constituted the great majority of total cost (Table 4, Fig. 1). The 3695 WDTC 2013 deaths are estimated to cost $351 million. Although men compose 24% of the total WDTC population, 28% of the costs are attributed to their care, perhaps reflecting the lower disease-specific survival in men and increased proportion of costs attributable to thyroid cancer deaths.
Table 4. Current Cost by Phase of Care and Sex in 2013
% of Cost
% of Cost
Thyroid cancer deaths
Sensitivity Analysis of Key Input Parameters
Overall, estimated costs remained stable across a reasonable range of input values. The tornado plot (Fig. 2) illustrates that overall costs do not change significantly when ranges of surgical morbidities and mortality are tested over a broad range of inputs. Given the changing practice patterns and recent evidence that RAI does not benefit survival outcomes in Stage I patients and observance of the risks of RAI, it was important to show the minimal impact on cost of decreasing use of RAI in practice. Varying cost of cancer death had the greatest impact on total costs; but as is shown, even with decreasing the cost of cancer death to half of the cost of similar cancer deaths reported in the literature, annual costs still exceed $1.4 billion. Surgical morbidity and mortality had minimal impact on the overall attributable cost over extensive ranges of complication rates.
Projected 2030 Thyroid Cancer Incidence and Cost
Joinpoint regression analysis indicates a significant change in the rate of increase in annual incidence for patients with WDTC in 1994 (Fig. 3). After adjusting for projected all-cause mortality in the population, we estimate a prevalence of 393,730 male and 1,251,705 female patients with thyroid cancer in 2030, more than double current survivorship. The total projected 2013 costs, assuming stable thyroid-cancer–specific survival and thyroid-cancer–related costs, exceeds $3.55 billion (in 2013 US dollars). Proportion of costs of attributable to initial treatment is predicted to decrease to 27%; with end-of-life treatment making up a greater relative proportion (34%) (Table 5).
Table 5. Projected Cost by Phase of Care and Sex in 2030 (in 2013 US Dollars)
% of Cost
% of Cost
Thyroid cancer deaths
The incidence of WDTC has increased more than any other cancer over the past 4 decades, while survival has remained relatively unchanged (>90% 20-year survival).[2, 33] Overdiagnosis likely accounts for much of this increased incidence, given that smaller PTCs account for much of the increased incidence.[4, 6] Regardless, the increased rate of diagnosis in addition to an aging population results in an increase in the number of WDTC survivors and their associated cost of care.
In our analysis, we estimate the aggregate national cost of WDTC care in 2013 to be $1.6 billion dollars. This estimate was stable with respect to a wide range of input parameters. Compared to estimates of site-specific costs of cancer care in 2010, this amount is comparable to the cost of other solid tumors such as cervical, gastric, and esophageal cancer. The cost of initial diagnosis and treatment comprises the highest percentage of cost in WDTC as it is in most other cancers. However, unlike other reported distributions of costs of cancer care, the “last year of life phase” in thyroid cancer makes up a smaller proportion (22%) of total cost, compared to 60% to 70% of costs for other cancers. This is undoubtedly due to the fact that most WDTC patients die of other, unrelated causes, rather than from their malignancy. The distinct cost distribution is an important consideration in cost-containment and quality improvement strategies. In contrast to the majority of other cancers, assessing the effectiveness of diagnostic, treatment, and surveillance strategies that comprise 78% of annual WDTC costs will be essential. It is important to note that the costs represented in this study include exclusively patients with a definitive diagnosis of cancer, who comprise only a portion of patients who undergo diagnostic work-up of thyroid nodules. In a recent meta-analysis of more than 25,000 pooled FNA biopsies, only 5.4% of FNAs were found to be malignant, and of the 25% of patients that went on to have surgery for indeterminate cytology (ie, requiring surgery for diagnosis), only 34% were malignant on histology.
Current clinical practice guidelines for WDTC patients are largely based on retrospective data and expert opinion, with few data available from large-scale randomized controlled trials. Although there are many potential reasons for the absence of these trials, the net result is that the comparative risks and value of many of our interventions are not well documented. Moreover, the magnitude of overdiagnosis and the consequence and costs of overtreatment of WDTC have not been evaluated but are likely substantial. Estimating the cost of care for WDTC is an important initial step before performing comparative effectiveness research to evaluate the effectiveness and efficiency of our current therapies.
This analysis has a number of limitations. Due to the limited clinical trial data to inform our input parameters, we used simplifying assumptions to model complex disease processes, varying clinical practice preference, and care pathways. Indeed, it has been shown that practice variations exist among physicians treating patients with thyroid cancer. We addressed this limitation using sensitivity analysis for key input parameters (Fig. 2). Estimates of thyroid cancer incidence and prevalence were based on SEER-18 areas and not the entire US population, and geographic variations in costs may exist. We are not able to discern from the SEER database if PTMC were found incidentally during thyroidectomy for benign disease or if the PTMC was indication for surgery and, therefore, may have overestimated the costs of preoperative cost of cancer diagnosis in this subset of patients. Although we did include a conservative cost of thyroid cancer deaths, we did not include the present value of future deaths from other causes. Given that we used observed survival with WDTC (including deaths from other causes) for available time points (ie, until 2010) and the relatively low mortality from WDTC, we feel that it was reasonable to omit this variable. Lastly, given the inaccuracies in projecting fluctuations, usually increases, in medical care costs above inflation, and/or potentially more effective, but more expensive innovations, we chose to report projected costs in 2013 US dollars by assuming stable treatment technologies and unit costs. Although recent reports indicate a deceleration in health care costs, health care inflation still exceeds average inflation in the US economy.[35, 36] Therefore, our projections are likely an underestimate compared to inflation-adjusted projected costs in 2030.
Given the relatively low disease-specific mortality of WDTC patients, WDTC is frequently not considered a “relevant” cancer. This is underscored by the fact that thyroid cancer has been omitted from cost-of-cancer-care analyses to date. This analysis illustrates that the care of patients with WDTC has significant economic and societal impact. The increasing prevalence and cost make it essential that we identify appropriate and effective strategies for diagnosis, treatment, and surveillance of patients with WDTC. Specifically, we hope that these findings provide the groundwork and focus for future research including improved risk-stratification and comparative-effectiveness research.
Funding is provided by a Program in Cancer Outcomes Research Training Grant (NCI R25CA092203), Massachusetts General Hospital Department of Surgery, and Massachusetts General Hospital American Cancer Society Institutional Research Grant.