Analytical approach:
The analysis was based on a Markov model, with a five-year time horizon. The authors stated that it was carried out from the perspective of society, but limited to the perspective of the US health care system.
Effectiveness data:
Most of the clinical inputs were from a variety of published sources, including literature reviews, government publications, professional society guidelines, and clinical trials. The key data were from published decision analyses on similar questions in thyroid nodule management. Some data were from a panel of six medical and surgical thyroid experts working at the Johns Hopkins Medical Institutions. The sensitivity and specificity of the diagnostic tests were the main inputs.
Monetary benefit and utility valuations:
The utility values for the model states were from published literature or the opinions of the expert panel. The panel used the time trade-off method to estimate the utility weights.
Measure of benefit:
Quality-adjusted life-years (QALYs) were the summary benefit measure and were discounted at an annual rate of 3%.
Cost data:
The economic analysis included the costs of surgery, in-patient care, out-patient care (clinic visits, ultrasound scans, and radioiodine scans), the treatment of complications, and follow-up (additional clinic visits, laboratory tests, imaging procedures, and medications). Most of these costs were from nationally representative sources, such as the Centers for Medicare and Medicaid Services, and the Agency for Healthcare Research and Quality. Some data, such as the costs of complications, were based on published estimates. All costs were in US dollars ($) and were discounted at an annual rate of 3%. The price year was 2010.
Analysis of uncertainty:
One-way sensitivity analyses were carried out to examine how robust the model outcomes were to variations in all the inputs; the ranges of values were from published sources and expert opinion. A probabilistic sensitivity analysis was performed, using 10,000 Monte Carlo simulations and assigning normal or beta distributions to the parameters.