OR19-2 Big Data Strategy Used to Draw TSH Reference Values for Elderly Population

Introduction: Determination of reference ranges for thyroid function tests, such as TSH, are not a simple task. There are several biological and individual aspects to be considered, one of the most important is age, particularly for the extremes, children and the elderly population. Several studies demonstrated a progressive increase in age-related TSH in older adults. Therefore, the aim of this study is to establish specific reference values for Brazilian population over 60 years old using Big Data strategy. Methods: Data from 10.43 million TSH tests performed in the same method were retrospectively analyzed. All the samples are from the same clinical laboratory, with samples from several regions of Brazil. The samples were selected using the following filters: free T4 concentration within the reference values (RV), absence of anti-thyroid antibodies and of medication use. Groups were divided by age from 18 to 59 years old (under-60), and over 60 up to 107 years old (over-60). Statistical analyzes were performed using R software and EP Evaluator. Both R package ‘boot’ and EP Evaluator nonparametric (CLSI C28-A) were used to estimate the 97.5% confidence interval (IC). Results: After filtering the samples, 306,289 were selected. Mean TSH was similar in both sex. However, there were differences between the two age-groups, in under-60 group, the lower RV was 0.59 mUI/L (95% IC = 0.57, 0.59) and the upper RV was 6.00 mUI/L (95% IC = 5.94, 6.08). Interestingly, in over-60 group the lower RV was 0.36 mUI/L (95% IC = 0.34, 0.38) and the upper RV was 9.38 mUI/L (95% IC = 8.86, 9.44). Furthermore, this upper limit of reference range increases within the older group with TSH of 7.4 mUI/L from 60 to 69, 9.60 mUI/L from 70 to 79 and 12.30 mUI/L above 80 years old. Conclusion: Big data strategy could be a useful tool to reach agreeable population reference values. In particular, for TSH, the higher upper reference limits for adults over 60 years old would be important to avoid overdiagnosis and over-treatment of subclinical hypothyroidism in this population.