David G Greenhouse
7.021 Section B
Out of Class Exercise 2
Organization of Study
We designed a computer simulation of the movement of a population through
various disease states related to ESH, atrial fibrillation, and their treatment. By assigning
each disease state a cost and a utility we can determine the total number of discounted
quality adjusted life years (QALY’s) accumulated by the population. We analyze two
treatment strategies: radioiodine ablation of the thyroid and long-term observation (a wait
and see strategy).
Using Data 4.0 (Tree Age Software, Inc., Williamstown, MA) we created a
Markov decision tree and implemented a Monte Carlo simulation with a yearly time step.
A schematic of Markov state diagrams for both treatments is shown below. Patients
moved between disease states at probabilities determined by the literature. Costs were
accumulated from a societal perspective given public financial data. Each year’s costs
and benefits were discounted 3% to account for the decreased value of distant costs (e.g.-
inflation). Each simulation ended once all patients had entered the dead clinical state. A
sensitivity analysis was then performed on all costs and transition probabilities in order to
determine the stability of the decision.
Base Case Demographics
Male and female populations were considered separately. Initially 10,000 patients
were divided into preliminary disease states given their prevalence in the Framingham
data. These patients all begin at age 60 due to the lack of data regarding ESH in patients
younger than 60 years. We also considered all patients to already have ESH confirmed
by third generation TSH assay level less than 0.1 mU/L on two separate occasions at least
two months apart.
The initial probability of having repeatedly high TSH assays was estimated from
the Wickham survey among others. These values range from 0.2-11.8%, but is likely
between 0.6-1.2%. Disease prevalence data for atrial fibrillation for the general
community were taken from the Framingham study. In order to determine the probability
of developing atrial fibrillation from the ESH disease state we utilized the data of Sawin,
et al and Scheibner, et al. Scheibner’s data showed a 12.3% chance of developing atrial
fibrillation (n=613), a marked increase over the 2.3% chance in the rest of the population
(n=22,300). We assumed the chance of developing atrial fibrillation to be uniformly
distributed over the lifespan of the population. Unfortunately, this study defined ESH as
those with a TSH <0.4 mU/L. As well, they did not require a repeated positive result.
Nevertheless, this suggests that the actual rate of development of atrial fibrillation in
those with ESH is higher than 12.3% since many “normal” patients without ESH (and
most likely a lower incidence of atrial fibrillation) were included in the population. The
probability of overt hyperthyroidism following diagnosis of TSH was taken from
Wiersinga as 5% per year when untreated. Age related all-cause mortalities were taken
from the CDC’s National Vital Statistics Report, 2002.
Treatments, Costs, and Efficacy
For all costs, we took a societal perspective. Therefore, whenever possible, we
compiled average medicare costs for treatments and physician consultations. For
radioiodine, we assumed an average treatment of 12mCi and recommend a standardized
dose regimen based on palpated thyroid size (Kok SW, et al). We assume 85% of
subclinically hyperthyroid patients treated with iodine will no longer be hyperthyroid
after one year. The remaining 15% will require re-administration of treatment (after
which no patients are still hyperthyroid). 15% of those treated once and 20% of those
treated twice will become hypothyroid from too high a dosage. The costs of treatment for
hypothyroid patients included treatment with levathyroxine at $340 per year (Osby’s,
2000). Patients with atrial fibrillation incur a cost of $1,458 per year using the data of
GB Lumber, et al.
When available quality of life weights were taken from Tengs, et al’s
compilation: “One thousand health-related quality-of-life estimates”. For the ESH
untreated state with no symptoms we assigned it a QALY of 1.0, since by definition there
is no loss of quality of life. For the inconvenience of radioiodine therapy we assigned
that year a QALY of 0.8. For mixed disease states, such as ESH with atrial fibrillation
we assigned the lower QALY score.