Dr. Bohlmann described a case of thrombocytopenia in a pregnant woman with Graves disease who was treated with propylthiouracil. Although the association of Graves disease and thrombocytopenia is well known, it is not clear from the case that the decrease in the platelet count was related to the propylthiouracil therapy or, alternatively, the manifestation of coexistent autoimmune thrombocytopenia. If treatment with propylthiouracil was continued, it would be interesting to know if the patient was still receiving propylthiouracil when she responded to prednisone. Propylthiouracil-mediated agranulocytosis is presumably immunologically mediated. A report from Magner and Snyder [1] provides additional evidence for the potential usefulness of human granulocyte-colony-stimulating factor in the treatment of propylthiouracil-induced agranulocytosis. As we suggested [2], stimulation of granulocyte stem cells may hasten recovery from this side effect of treatment.

As Dr. Matfin knows, page constraints limit the depth and breadth of the discussion, even in review articles. Our main focus was to describe the management of the common forms of hyperthyroidism. We agree that for the treatment of the rare pituitary adenoma causing hyperthyroidism, adjunctive therapy beyond surgery may be required. The comments about octreotide therapy are potentially useful and may apply to individual patients.

The management of patients with hyperthyroidism who have atrial fibrillation carries additional considerations. We disagree with Dr. Woeber and adhere to our recommendations [2] that routine anticoagulation is not recommended for patients with hyperthyroidism for the following reasons: Many of these patients are young and have no underlying heart disease; atrial fibrillation resulting from hyperthyroidism is frequently of acute onset and will spontaneously revert to sinus rhythm without associated side effects [3]; the prevalence of systemic embolization in patients with hyperthyroidism who have atrial fibrillation has not been sufficiently studied to identify the actual risk [4]; and studies of groups larger than those in previous trials would indicate that the prevalence of systemic embolization is much lower than previously reported (Nakazawa H. Personal communication). In light of these considerations and in keeping with the conclusions reached by Presti and Hart [4], the decision to treat thyrotoxic patients with atrial fibrillation needs to be made on a case-by-case basis, addressing relevant variables such as advanced age, the presence of organic heart disease, and heart failure.

Dr. Gill questions our statement that radioiodine is the preferred treatment for patients with Graves disease. He suggests that a multicenter trial is necessary to arrive at such a conclusion. To the contrary, because radioiodine is the most frequently chosen form of treatment in the United States and because of its high likelihood of success and limited, if any, side effects, we find the term "preferable" appropriate. If other forms of therapy or modifications of existing treatments are developed in the future, their effectiveness must be compared with that of radioiodine treatment. Our opinion remains as stated.