This study is the first report of the incidence of GD in Thailand. The age-adjusted incidence of GD in this study was 26.57 per 100,000, similar to other countries about 20–40 per 100,000, but the female to male ratio of 1.6:1 is lower than others of 10:1 [1–4].
The first-line treatment was an ATD of 99.7%, higher than the recent survey study of 90.8%  and retrospective review of 70.8% in Thailand . The RAI treatment of 0.3% was lesser than 21% of the previous study. Like other countries in Asia and Europe , ATD is the most popular treatment. The median duration of ATD treatment was 32 months (IQR 22–47), almost twice longer than the recommended duration of 12–18 months . Due to RAI unavailability in the local site, the leftover treatments were prolonged ATD treatment and surgery. This logistic problem probably explained the causes of much more ATD preference and duration. MMI was the most preferred ATD in general and PTU was usually used in severe disease and pregnancy. TRAB was done only 2.2%, probably due to high cost and long turn-around time.
The remission rate of ATD was 21.5%, similarly, the remission rate from the retrospective study of Thai patients in the private hospital was 30.7% . This ATD remission rate was drastically different from the studies in the US and Europe of 50% [17–19], even though ATD duration was prolonged. The author suspected high iodine diet in general and genetic components were the cause of refractory disease, even though the patients were instructed to take a low iodine diet for a week before RAI. More intensive treatment such as higher dose RAI or surgery might be proper in this circumstance. RAI showed the most effectiveness in combined with remission rate and time to failure. Surgery showed the most remission rate but the worst time to failure and complication. These were probably due to non-specialized surgeons in the countryside hospital where thyroidectomy was usually performed by Ear-Nose-Throat (ENT) or general surgeons. The small sample size of 9 out of 355 patients who underwent surgery could cause inaccurate outcomes as well. Although long-term ATD could be considered in a refractory disease , this strategy showed better results compared to RAI in hypothyroidism and GO [20, 21]. The remission rates were 77.3% for RAI and 88.9% for surgery, less successful than the nationwide study in the US of 93% for RAI and 99% for surgery .
RAI had the most safety profile without any complications, except the adverse condition of hypothyroidism. Surgery showed most complications of 33% with 3 out of 9 patients, followed by ATD of 1.1%. The risk of surgery was varied in previous studies, hypocalcemia occurred 9.4–54.4% and recurrent laryngeal nerve injury occurred 0.9–33% [22–25]. To decrease these risks, the procedure should be done by a skilled high-volume surgeon. Risk of agranulocytosis occurred far less about 0.1-1.0% [26–28].
The cost-effective treatment was seen most in RAI with about 100 USD (3,000 Thai Baht) per a single treatment, followed by subtotal thyroidectomy with 830 USD (25,000 Thai Baht) per single surgery and MMI with 54 USD (1,620 Thai Baht) per year for a dose of 15 mg per day. Thyroxin replacement cost was 12 USD (360 Thai Baht) per year for a dose of 100 mcg per day. These costs were used in public hospitals and covered by the Thai universal health care system. On the other hand, in private hospitals, the costs were usually three times higher. Costs of treatment were various in different countries, the most cost-effective modality was RAI in the UK but total thyroidectomy in the US .
We analyzed the predictors of the ATD time to failure and found TSH < 0.01 uIU/ml as a significant factor. We did not find any association between time to failure and FT3 and FT4. However, younger age, high FT3, FT4, orbitopathy, and large goiter size were significant factors predicting treatment failure in the past studies [15, 17, 29, 30].
Several limitations could have influenced this study. The incidence analysis included only a small number of patients due to a lack of good data management. The retrospective scheme might have confounder and bias. This study was not categorized as a controlled disease and might underestimate remission patients in a different definition.