This study demonstrated a 14,7% prevalence of TD in the diabetic patients studied. Subclinical hypothyroidism was the most frequent dysfunction found corresponding to 11,8% of the patients which was similar to studies already described in the literature [2, 7] but higher than that reported in studies with non-diabetics [8–10].
Analyzing data from T1DM patients, the frequency of TD was 13% of SC -Hypo in patients without prior TD and 14,6% anti-TPO positive. In the study by Ramos et al., where 126 patients with T1DM were evaluated, the prevalence of TD was 20.6%. However, they considered a positive test for autoimmunity as TD. In the same study, the prevalence of SC-Hypo was 13.5% and anti-TPO antibodies was 18.2%, but patients were younger compared to our study. In the study by Souza et al., evaluating 101 patients with T1DM, the prevalence of Sc-Hypo was 6.5% but FT4 was not used for diagnosis. A prevalence of 30.7% of at least one antibody was found positive in the same study, where 3 antibodies were assayed: anti-thyroglobulin, anti-microssomal, and anti-TPO. The variability of the laboratorial methods on screening of thyroid autoimmunity could influence the results described above because the latest studies use anti-TPO as preferred test .
According to various studies, the prevalence of autoimmunity in T1DM patients may vary between 3 and 50% [13, 14]. One factor that may account for this variability could be the different laboratory methods used for the evaluation of autoimmunity.
Nine new cases of DT were diagnosed in patients with T1DM, whom ignored or denied prior TD. Patients whom already known to have TD, only half (50%) presented hormone levels in normal range. This data are consistent with the NHANES study , where only 67% of individuals who self-reported having TD were adequately treated. Non-adherence to therapy can be explained by factors such as: the excess of medications used due to comorbidities, adverse effects of medications, high cost, education of patients that limits access to information, the asymptomatic nature of the disease(in some cases) in which the importance of drug treatment is not recognized . Data from the study of “Adherence to Medication for Chronic Diseases (diabetes and hypertension) in the City of Teresina” demonstrated that the degree of compliance percentage was lower than the “recommended”  level of 80%. Other Brazilian studies demonstrated adherence rates of 11% in Bahia and 66.6% in São Paulo .
In the analysis of T2DM patients, we observed a frequency of all TD of 13,1% and SC-Hypo of 12%. This frequency found is higher than those described in the study of Fremantle (8.6%)  and in the study Chu et cols. (8.4%) . Previous studies have shown that the risk of thyroid dysfunction increases with age [11, 15]. It is important to emphasize that our patients with SC-Hypo are older than the two earlier studies afore mentioned. These findings may be more favorable to subclinical abnormalities that could explain the higher frequencies found in our study. Subclinical TD are characterized with biochemical changes without inclusion of clinical signs or symptoms .
We observed a overall frequency of 10,8% positive anti-TPO, 14,6% in T1DM and 9,9% in T2DM. The epidemiological study of Whickam , showed that 55% of patients with elevated TSH and positive anti-thyroid antibodies progressed to clinical hypothyroidism in contrast with 33% of those who had elevated TSH and negative antibodies.
Some limitations of our study must be discussed. This study was a cross- sectional with internal validity and we used a convenience sample of diabetic patients already treated in an university hospital. Some types of selection bias may have occurred because these patients are already under medical care. Some patients whom already had prior knowledge of their thyroid function could have some nonspecific symptoms of TD but not yet with laboratorial confirmation. They might be more interested on research. However, all patients were submitted to the same study protocol.
Other limitation of the study is that only one sample of TSH and FT4 was collected, which may have contributed to a high frequency of TD, since the phenomenon of regression to the mean was not minimized by a second sample. However, the dosage of FT4 is not always included in other prevalence studies like NHANES . The level of FT4 add a greater specificity in the detection of TD.
The strength of this study is the great number of patients (n = 386), more than described in other Brazilian studies with diabetic patients mentioned previously and the dosage of TSH, FT4 and anti-TPO using ultra-sensitive laboratory tests performed in all patients.
This study joins previous studies to determine the prevalence of thyroid dysfunction in diabetic patients in Brazil and also includes a greater number of type 2 diabetic patients whose prevalence is poorly described in previous studies.