It is universally accepted that patients on thyroxine replacement should have regular clinical and biochemical surveillance to ensure compliance and adjust doses in relation to changing requirements. However the ideal screening interval has not been ascertained and there are no published studies on the appropriate frequency of follow-up. Surveillance by annual TFT is recommended by most of the published thyroid guidelines [9–11]. The recommendations are largely based on consensus, and there is limited evidence to support this practice. Vanderpump et al  recommended further audit/study to clarify some of the uncertainties in the follow-up of hypothyroid patients.
Grampian Automated Thyroid Register (GAFUR) was established in 1967 , and is one of the oldest thyroid registers in the world. Thyroid registers improve records, ensure regular reviews, and promote early detection of changes in the patient's clinical status. Thyroid registers have proved to be beneficial in the long-term follow-up of patients with thyroid dysfunction. They are cost-effective  and achieve better biochemical control  in comparison to conventional follow-up. From our thyroid register we were able to identify a large cohort of hypothyroid patients on thyroxine with a minimum of 10 years of follow-up. As expected the majority of patients were women who had developed spontaneous hypothyroidism. As there were no set guidelines, individuals were allocated to either follow-up plan by the referring physician on a non-selective basis. The baseline characteristics were well matched and there did not appear to be any bias in relation to allocation to either group. The system of recall and follow-up was well managed through good co-operation and links between primary care and GAFUR. Individuals in both the groups were compliant with follow-up arrangements. Similar benefits of thyroid registers in maintaining long term surveillance with minimal defaults was reported by Jung et al. in 1991.
During follow-up thyroid function tests were undertaken at each review appointment and 18 monthly follow-up was not associated with an increase in adverse outcomes. On the other hand, patients on annual surveillance were more likely to have an abnormal test. Both the groups were well matched at baseline (table.1) in relation to age and sex distribution, thyroxine dose and duration of follow-up, and we cannot readily account for the difference in outcome. One possible explanation could be physician bias at the time of registration. There were no set guidelines or criteria and it is possible that there was physician bias while allocating individuals to annual screening taking into account factors such as compliance. It is also likely that more frequent testing will find more abnormal results. Another conclusion that can be drawn is that annual surveillance should be recommended as there was an increased chance of detecting an abnormal result. This approach may however lead to unnecessary tests in some individuals as thyroxine dose remained unchanged in 35% of patients during long term follow-up. On the other hand, although 18 monthly surveillance cannot be applied universally, it may be an option in individuals whose thyroxine requirement is less likely to change or have less abnormal tests during follow-up. We undertook further sub-group analysis to identify individuals who may be suitable for less frequent testing.
Patients aged over 60 years at the time of registration were more likely to have an abnormal test during follow-up. This may be due to decreased thyroxine requirements with increasing age. This has been reported in previous studies [17, 18] and probably reflects the progressive decrease in thyroxine degradation rate that occurs with advancing age. The fall in thyroxine requirement with age also corresponds to the loss of lean body mass . Hence people aged over 60 years require annual surveillance.
Individuals taking 100–150 μg/day required fewer dose changes during long term follow-up compared to the rest. We would recommend annual testing in individuals who are on either <100 or >150 μg of thyroxine per day. Once established on the appropriate dose of thyroxine, follow-up interval can be extended to 18 months in individuals aged less than 60 years. There were insufficient patients on the register with a regular follow-up interval of two years to evaluate even less frequent monitoring.
Individuals on annual surveillance had more frequent reviews and required more TFTs compared to the 18 monthly group, and this will have cost implications. The Grampian automated thyroid register currently has over 10,000 registered hypothyroid patients, and the majority are being followed-up 18 monthly. Allocating everyone to annual follow up would mean an extra 3,000 thyroid tests per year. A thyroid function test costs around £15 for the NHS (personal communication with Dr Heather Watson, clinical biochemistry, NHS Grampian on 24/04/06). This would amount to an extra spending of around £45,000 per year without taking into account the cost of additional administrative time for GAFUR and primary care, and inconvenience to the patient. Our study has shown that 18 monthly follow-up may be adequate in the majority of patients under 60 years of age on a stable thyroxine dose of 100–150 μg/day. Implementing these recommendations has potential benefits in terms of savings and reduction in the workload.
The Grampian automated thyroid register was set up in the late 1960s and traditionally TSH and FT4 estimations were undertaken. The recently published UK guidelines  on the use of thyroid function tests recommend TSH estimation alone in the long-term follow-up of treated hypothyroid patients. We are considering changing our practice to achieve additional cost savings in relation to thyroid tests.