Prediction of mortality in overt and subclinical hyperthyroidism

Dr Kristien Boelaert, University of Birmingham.

Final report

Our study proposal aimed to define those patients with overt hyperthyroidism at particular risk of increased mortality, especially form cardiovascular disease. Using the largest thyroid research clinical database in the UK we set out to investigate mortality in a cohort of subjects with overt hyperthyroidism and in a euthyroid cohort with benign goitre or thyroid nodules. Specifically we investigated the influence of severity, duration and treatment (medical, radioiodine) of hyperthyroidism and of subsequent hypothyroidism on mortality. Furthermore we aimed to clarify the influence of potential vascular risk factors such as previous cardiovascular history, cardiovascular medications, structural heart disease, hypertension, family and smoking history, hyperlipidaemia and presence of AF/dysrhythmias on mortality in order to identify high risk groups.

Identification of cohort

We identified the following cohorts:

  • Hyperthyroid cohort: 2397 thyrotoxic patients (1919 females and 478 males, mean age 47.14 years).
  • Control cohort: 1552 patients (1351 females and 201 males, mean age 60 years)

All these patients attended our joint thyroid clinic between 02.06.1975 and 19.05.2003. We chose 01.06.2003 as the cut off date to allow adequate follow-up time in clinic.

The demographic details of patients in both of these cohorts were sent to the Office of National Statistics (ONS) for determination of their vital status.

Results from ONS - update since Interim Report

All the records for both the hyperthyroid and the control cohort have now been returned from ONS including those requiring operator searches. All the data regarding severity, duration and treatment of hyperthyroidism, subsequent hypothyroidism, and cardiovascular risk factors have been collated for both cohorts. To date 199 patients in the thyrotoxic cohort have died and their death certificates and ICD-10 codes have been returned to us. In the control group 178 patients have died and similarly the causes of death and the ICD-10 codes have been returned to us.

Data analysis to date

We have decided to first analyse the data from the thyrotoxic cohort.

  1. Effects of treatment of thyrotoxicosis on mortalilty
    Preliminary analysis of the data so far indicates that the mortality rate is higher in patients treated with antithyroid drugs only when compared with those receiving radioactive iodine (p< 0.001). A small minority of patients underwent surgery for hyperthyroidism (1.84%) and only one of these patients has died.
  2. Effects of development of hypothyroidism post 131I treatment on mortality
    Of the 1343 patients treated with radioactive iodine, 649 (48.3%) patients have become hypothyroid whereas 694 (51.7%) patients remain hyperthyroid or euthyroid. Death rates are significantly higher (p=0.01) in those who have not developed hypothyroidism compared with patients who are hypothyroid and on thyroxine replacement following treatment with 131I.

Further data analysis to be undertaken

Data sheets compiling all the data obtained for this study are currently being analysed by Dr Maissoneuve with whom we have previously collaborated to generate high impact factor papers. In particular we are calculating the standardised mortality ratio in patients with hyperthyroidism to compare this with control populations of the West Midlands and England and Wales. In addition we are evaluating the influence of severity, duration and treatment of thyrotoxicosis, subsequent hypothyroidism and cardiovascular factors on mortality. It is anticipated that the results of these analysis will result in several high impact factor papers.

Since the statistical analysis is still ongoing, we would like to request that the duration of the project is extended for a further 6 months. This will allow us to spend the funds remaining for this project (£2,000 + £2,500 to be obtained on submission of this report) which will be used for outstanding ONS fees and statistician fees.