The association between subclinical hypothyroidism and hyperthyroidism and mortality in the elderly is poorly defined. This study was designed to evaluate the association between subclinical hypothyroidism and subclinical hyperthyroidism and mortality in the elderly and to define the thyroid-stimulating hormone values associated with excess mortality in the elderly.

We performed a retrospective cohort study with a review of a computerized database of a large health care organization. Patients aged more than 65 years evaluated in the years 2002 to 2012 with documented normal free T4 values were included in the analysis. All cases of known thyroid disease or cases in which thyroid medications were dispensed were excluded. Analysis was performed only on individuals who were not treated for hyperthyroidism or hypothyroidism during the follow-up period. Subjects were divided into 3 groups based on thyroid-stimulating hormone values: normal (normal thyroid-stimulating hormone), subclinical hypothyroidism (thyroid-stimulating hormone >4.2 mIU/L), and subclinical hyperthyroidism (thyroid-stimulating hormone <0.35 mIU/L). All-cause mortality hazard ratio (HR) was compared among the 3 groups, and a subanalysis according to thyroid-stimulating hormone values was performed in those with subclinical hypothyroidism and subclinical hyperthyroidism.

A final analysis was performed on 17,440 individuals with subclinical thyroid disease (538 with subclinical hyperthyroidism [3.1%], 1956 with subclinical hypothyroidism [11.2%], 14,946 normal cases [85.7%], average age of 83 years, 10,289 were women) who were followed up for 10 years. Both subclinical hypothyroidism (HR, 1.75; confidence interval [CI], 1.63-1.88) and subclinical hyperthyroidism (HR, 2.33; CI, 2.08-2.63) were associated with significantly increased mortality, and this association persisted on multivariate analysis (subclinical hypothyroidism HR, 1.68; CI, 1.56-1.8, subclinical hyperthyroidism HR, 1.93; CI, 1.7-2.17). Crude mortality was elevated at 1, 2, and 5 years, but this association seemed to decrease as time from initial analysis increased (most significant association at 1 year). Thyroid-stimulating hormone values greater than 6.38 mIU/L were associated with the highest mortality in those with subclinical hypothyroidism after multivariate adjustment (HR, 1.708; CI, 1.38-2.12), whereas in subclinical hyperthyroidism, no threshold for increased mortality was identified. Mortality was higher.

Both subclinical hypothyroidism and subclinical hyperthyroidism are associated with increased mortality in the elderly. A threshold thyroid-stimulating hormone value (>6.35 mIU/L) exists for increased mortality in subclinical hypothyroidism, but not in subclinical hyperthyroidism.