Association between serum TSH and fT4 concentrations and occurrence of atrial fibrillation in a large community-based population of elderly subjects. The Birmingham Elderly Thyroid Study (BETS)
M Gammage2, JV Parle1, JA Franklyn2, S Wilson1, L Roberts1, R Holder1, M Sheppard2, FDR Hobbs1, A Roalfe1, C Heath1 & H Pattison
We conducted a prospective study of 5784 subjects aged 65 or over, screened and identified in primary care. The cohort comprised 2934 women, 2850 men, mean age 73.5 years (range 63.5-98.6). All had TSH measured in a sensitive assay (normal range 0.4-5.5 mU/l), as well as free T4 (normal range 9-20 pmol/l), and resting 12-lead ECG. Atrial fibrillation (AF) was present in 4.81% of the cohort. The prevalence of AF showed significant variation with serum TSH concentrations, being highest in the group with low but detectable TSH (AF in 1 of 43 subjects with TSH <0.1 mU/l [2.33%], AF in 12 of 109 subjects with TSH 0.1-0.3 [11.01%], AF in 258 of 5443 with normal TSH [4.74%], AF in 7 of 189 subjects with TSH >5.5 mU/l [3.70%], P<0.02). Only 6 of 109 subjects with low but detectable TSH had free T4 above the normal range, the highest value being 22.6 pmol/l. The association between TSH and AF was similar in men and women; as expected, AF was associated with increasing age (P<0.001) and male sex (P<0.001). Serum free T4 concentration was higher in those with AF than those without (mean serum free T4 14.89, SE 0.13 versus 14.41, SE 0.03 pmol/l, P<0.001). There was an association between AF occurrence and free T4 considered as a continuous variable (P=0.001), both in the whole cohort and in men and women. This large prospective community-based study of the over 60's has demonstrated an association between low serum TSH and occurrence of AF, as well as a positive association between serum free T4 and AF occurrence. Our findings are suggestive of an association between even mild thyroid hormone excess and AF in this age group, a finding with important epidemiological implications for cardiovascular morbidity and mortality.