The Diagnosis and Management of Subclinical Hypothyroidism Is Assay-Dependent

Implications for Clinical Practice

Tejas Kalaria; Anna Sanders; Jonathan Fenn; Helen L. Ashby; Pervaz Mohammed; Harit N. Buch; Clare Ford; Rousseau Gama


Clin Endocrinol. 2021;94(6):1012-1016. 

In This Article


We report that the diagnosis and management of subclinical hypothyroidism (SCH) would be strikingly different depending on whether patient samples are analysed for TSH and fT4 with assays provided by Abbott Laboratories and Roche Diagnostics, which are the two commonest thyroid assays used by approximately 75% of clinical laboratories in the UK. The between-assay diagnostic discordance for SCH would affect clinical management. None of the patients with Roche defined SCH who had normal Abbott TSH and fT4 results (75%) would require follow-up testing. In addition, 12% of patients with Abbott defined SCH would be considered for levothyroxine replacement if their samples were analysed using Roche assays. It, however, is unclear whether Roche assays lead to incorrect diagnosis and treatment of SCH or Abbott assays lead to missed diagnosis and under-treatment of SCH.

This may have significant clinical implications. SCH with TSH ≥10 mIU/L is associated with adverse changes in the lipid profile, increased carotid intima-media thickness, increased risk of heart failure and increased risk of cardiovascular mortality.[1] SCH with TSH ≥ 7 mIU/L is associated with risk of fatal stroke in younger patients and increased risk of coronary heart disease.[1,5] On the other hand, higher serum TSH with lower fT4 within the euthyroid range may be associated with lower risk of adverse events and mortality in older people[14] and unnecessary treatment with levothyroxine may result in iatrogenic thyrotoxicosis with increased risk of atrial fibrillation and osteoporosis[9] but this is unclear.[15] It may, therefore, be of importance not to over diagnose SCH or hypothyroidism requiring treatment especially in the elderly. At the same time, undertreating SCH with TSH ≥10 mIU/L, perhaps even ≥7 mIU/L especially in the young, could have multiple detrimental health effects. Although we excluded samples from pregnant women and children, SCH in pregnancy and childhood may be associated with adverse obstetric and neuro-intellectual outcomes.[16,17] Similarly, the misdiagnosis of SCH in these patient cohorts may, therefore, also have adverse health effects.

The two main reasons for the clinically discordant results are between-method bias of the TSH and fT4 assays and variation in the manufacturer-provided reference ranges. The Roche TSH and fT4 assays have a positive bias compared to the Abbott TSH and fT4 assays, respectively. In addition to the positive method bias, the upper reference limit (URL) for the Roche TSH assay is lower than that of the Abbott TSH assay (Figure 1) and would therefore be expected to report more TSH results above its URL compared to Abbott TSH assay. Although Roche fT4 has an approximate 15% positive bias, its lower reference limit (LRL) is 25% higher compared to the Abbott fT4 assay, which may explain the greater number of Roche fT4 results below its URL compared to Abbott fT4 results. This indicates dissociation of current TSH and fT4 assay performance of either or both the manufacturers' assays from their respective manufacturer-provided reference ranges.

The positive method bias of the Roche TSH assay over the Abbott TSH assay reported in this study is consistent with a previous study[8] and external quality assurance (EQA) data. EQA data have shown increasing divergence with the Roche TSH assay and Abbott TSH assay, respectively, showing an upward and downward trend in the last three years (commentary: UK NEQAS for Thyroid Hormones, distribution 444, October 2019). Despite efforts of IFCC C-STFT to standardize and harmonize TSH and fT4 assays,[10,18] substantial between-method variability persists (UK NEQAS for Thyroid Hormones). The difficulty in harmonizing TSH and standardizing fT4 assays has prevented the development of universal reference ranges for TSH and fT4.[19,20] It is too onerous for individual laboratories to derive in-house reference intervals for every single test offered and particularly to repeat the process with every change of the analytical system. Moreover, in-house assay-specific reference ranges derived by laboratories[21] may not remain valid with significant assay drift, as was observed with TSH, even on the same analytical platform. Most laboratories, therefore, use manufacturer-provided assay-specific reference ranges.

This study highlights the limitations of firstly manufacturer-provided reference ranges and particularly in relation to prevailing assay performance and secondly universal thyroid hormone cut-offs used in clinical guidelines for the management of functional thyroid disease[4] since they do not consider between-assay differences. Most guidelines on SCH recommend levothyroxine replacement if the TSH is ≥10 mIU/L, especially in younger patients.[2–4] On current assay performance, however, a Roche TSH of 10 mIU/L equates to Abbott TSH of 7.1 mIU/L, whereas an Abbott TSH of 10 mIU/L equates to Roche TSH of 14 mIU/L.

The study also raises other potentially important issues related to SCH. These assay-related differences may have affected the outcome of a multi-centre randomized controlled trial[22] and conclusions drawn from meta-analyses[23–27] of SCH as data from multiple analytical platforms were pooled, and additionally a single TSH cut-off was applied in some studies.[22,26] In addition, between-assay differences would affect the prevalence of SCH depending on the analytical method used and may have contributed to the reported differences in the prevalence of SCH.[1,3]

In summary, clinicians and laboratorians should be aware that between-assay differences and variations in reference ranges will directly impact on the diagnosis and management of subclinical hypothyroidism.