How should the TSH reference range be determined?

Issue: June 2008

An age-specific range may be useful

Some people believe that the upper limit should be lowered to 2.5 mIU/L. They believe that if you take groups of patients and exclude all of the people with any form of thyroid disease by any measurement and you measure the antibodies that may be present, anyone who has them should be excluded. Then, you perform an ultrasound on the thyroid and if there are any abnormalities, you exclude those patients as well. In the end, you’re left with a super-pure population and it’s stated that those individuals will have an upper limit of 2.5 mIU/L; however, there are no data to support that. By contrast, a study published by Hamilton et al in the Journal of Clinical Endocrinology & Metabolism did exactly that, and they found an upper limit of 4.1 mIU/L.

Martin Surks

If you consider an absolute upper limit in the 4 mIU/L-range, a percentage of individuals will be considered abnormal across all ages. If the range is lowered to 2.5 mIU/L, the percentage quadruples to about 12%, which represents about 25 million people in North America. This is not an argument over a number here or a number there, but rather has very important public health consequences. When you consider older people, 40% of them have a TSH level above 2.5 mIU/L, which is a considerable amount of people involved if they would be considered to have thyroid trouble. Hollowell and I found what seems to be a shift in the whole population of older people; it’s not just the same curve. This is a population effect that implies that there might be justification for the consideration of an age-specific reference range. However, the information needs to be confirmed and discussion surrounding this needs to be increased before a limit should be changed.

Martin Surks, MD, MACP, is the Program Director for the Division of Endocrinology and Metabolism at Montefiore Medical Center in Bronx, NY.

It is impossible to establish a range using population data

The TSH upper reference limit is really a moving target. It depends upon the population being studied, the underlying pathology and iodine intake of that population, as well as the specificity of the assay for detecting the various TSH isoforms present in sera. Its essential to ensure that individuals with any degree of thyroid dysfunction are excluded from reference range calculations. However, there are some individuals that are TPOAb-negative and yet have thyroid dysfunction. The inclusion of such individuals has the potential to skew the TSH reference limits to some extent.

Carole Spencer

Surks and Hollowell reported that the TSH reference range increased with age and concluded that perhaps a higher TSH is a natural consequence of aging. We interpret this finding differently. It is well known that in the United States the prevalence of autoimmune thyroid disease increases with age and not all affected individuals display TPOAb. In the SHIP study, there was a trend for a decrease in TSH with age, despite excluding individuals with nodules visible on ultrasound. Both the NHANES and SHIP studies did their best to screen out occult thyroid dysfunction yet opposite trends in TSH were seen with aging. These trends are less likely to be the physiological effect of aging, but more likely the influence of occult thyroid pathology specific to that population contaminating the data and causing these age-related shifts in TSH.

It is not possible to establish a universal TSH upper limit from population data. An appropriate compromise would be to adopt an empiric TSH reference range approximating 0.3 to 3 mIU/L, as suggested by AACE. It is important to recognize that the upper TSH reference limit is not the therapeutic threshold for initiating levothyroxine replacement therapy. Instead, the clinical response to a confirmed mildly elevated TSH should be determined on an individual basis.

Carole Spencer, MD, PhD, is a Professor of Research at the Keck School of Medicine at the University of Southern California.