There are two regimens to treat hyperthyroidism with drugs. Give thyroid blockers alone in doses that may or may not control thyroxine levels in a given patient (standard) or Give maximum dose of thyroid blockers and replace thyroxine(block and replace).
We don’t routinely use block and replace regimen, since, there is no evidence to suggest that it increases remission rate. However, there are other metrics we may care about, apart from long term remission rates.
Quality of life
Therapeutic certainty that there won’t be hypothyroidism or hyperthyroidism
Speed of attainment of euthyroidism
The best way to identify this is to do a randomized controlled trial. However, since we are testing QOL and due to differences in perception between individuals, we may get noisy data.
I can think of two options.
? Use mixed effects regression
Use a cross over design
Are there any better ways to approach this problem, if we care about things other than remission rate?
In psychiatry we often assess subjective endpoints like quality of life. Mixed effects regression is the gold-standard approach in the field for RCTs. Primarily because they handle missing data well, but also because comparing average slopes between groups smooths out some of the noise and is less vulnerable to regression to the mean. Here’s an example picked at random: https://jamanetwork.com/journals/jamanetworkopen/fullarticle/2770049
This may be a bit off topic but the utility of block and replace is only when sequential thyroid function is unavailable such as for a remote rural patient who has limited access to testing. There is no other utility for this regimen and I can probably say, with fair certainty, that QoL will be worse, therapeutic certainty is better judged through testing (thyroxine has a half life of a week and thus six weekly testing is sufficient) and there is no difference in speed of attainment of euthyroidism because the standard practice in terms of tapering therapy is to start with the maximum effective dose (e.g. carbimazole 20mg bd) and then taper based on patient evaluation and lab testing.
In clinical research, we need to be careful to design a study whose answers are not already obvious or that we cannot derive from first principles.
Here’s an interesting N-of-1 trial that I offer up to expand your differential.
Riggare S, Unruh KT, Sturr J, et al. Patient-driven N-of-1 in Parkinson’s Disease. Lessons Learned from a Placebo-controlled Study of the Effect of Nicotine on Dyskinesia. Methods Inf Med. 2017;56(99):e123-e128. doi:10.3414/ME16-02-0040 Open-access PDF
We have patients from remote rural areas. Mostly poor. Qol , not sure whether the existing questionnaires are applicable. Anyway, thanks for your inputs
Agree, it is a problem in countries with large remote areas like Australia and India. Although the death of B&R for Graves’ was “announced” in the early 2000’s (except for rural/remote) there have still been relatively recent studies and opinions e.g. this by Duntas.
What has not been done (I may not be aware as most centers like mine only see urban patients) is what is the best combination when it is needed and how can that be decided. If you have a lot of remote patients this would really be an interesting study and may help the under-served communities get better care.