Introduction

Thyroid dysfunction is a frequent health problem worldwide. Hypothyroidism can be found in up to 10% of the general population1,2,3, and, without treatment, it has been shown to be associated with several health disorders including dyslipidemia, hypertension, endothelial dysfunction, impaired cognitive function, infertility, and adverse pregnancy outcomes4. Hyperthyroidism is a less frequent disorder, whose prevalence ranged between 0.2 and 1.3% of the general population in iodine sufficient areas1, and it has been shown to be associated with an increased risk of osteoporosis, atrial fibrillation, stroke, and cardiovascular mortality5. Due to the high frequency of thyroid dysfunction and the consequences of the delay in its diagnosis and/or treatment, it is advisable to increase the knowledge about the epidemiology and factors associated with the development of this disorder.

However, data about the incidence of thyroid dysfunction are scarce and limited to certain geographical areas or cities3,6,7. Specifically, in the Spanish population, there is only one published study that evaluated the incidence of thyroid dysfunction, but its results are limited by the sample being from a particular region and the short follow-up period8.

The di@bet.es study is a population-based national cohort study designed to determine the epidemiology of metabolic and endocrine diseases9,10,11. The prevalence of thyroid dysfunction and the iodine status of the adult Spanish population were investigated in the cross-sectional study9,12.

Therefore, the present study aims to evaluate the incidence of thyroid dysfunction in Spain using the representative sample of the Spanish adult population from the di@bet.es study. Additionally, several potential exposure factors associated with the onset of hypothyroidism were investigated.

Materials and methods

Study design

The di@bet.es study is a nationwide population-based cohort study. The initial cross-sectional study was undertaken between 2008 and 2010. A cluster sampling design was used to select participants to form a representative random sample of the Spanish population. A total of 100 health centers or their equivalent from all around the country were selected at random, with a probability for selection proportional to their target population size, after which 100 individuals aged ≥ 18 years were randomly selected from each health center (Fig. 1). Thyroid function studies at baseline were performed in 4554 individuals11.

Fig. 1
Fig. 1The alternative text for this image may have been generated using AI.
Full size image

Map showing the 100 clusters included in the Di@bet.es study. [Base map source: MapSVG “Spain provinces” (licensed under Creative Commons Attribution 4.0 International, CC BY 4.0), https://mapsvg.com/maps/spain-provinces. Modified by the authors.].

The cohort was re-evaluated in 2016–17 (follow-up time was 7.5 ± 0.6 years).

For the present analysis, we included 1629 individuals who were euthyroid at baseline (not receiving levothyroxine/thionamides, and with TSH levels ≥ 0.2 and ≤ 5.0 mIU/mL), and completed the follow up study (Fig. 2).

Fig. 2
Fig. 2The alternative text for this image may have been generated using AI.
Full size image

Flowchart of recruitment process for participants in the study.

The research was carried out in accordance with the Declaration of Helsinki (2008) of the Word Medical Association. Both the cross-sectional study and the follow-up study were approved by the Ethics and Clinical Research Committee of the Hospital Regional Universitario de Málaga (Málaga, Spain). All the participants were informed about the nature of the study, and provided written informed consent in the two phases of the study.

Variables and procedures

In both phases of the study, participants were invited to attend a single examination visit at their health center with a nurse specially trained for this project. Information was collected using an interviewer-administered structured questionnaire, followed by a physical examination and blood sampling.

A structured questionnaire including closed questions was used to collect information on demographic characteristics, smoking status, medical history, and medications.

Weight and height were measured by standardized methods. Body mass index (BMI) was calculated (weight/height2).

Blood samples were obtained in fasting conditions. Participants with a baseline capillary blood glucose level lower than 7.8mmol/L (measured by OneTouch system, Lifescan, Johnson & Johnson, S.A., Madrid), and who were not receiving treatment for diabetes, underwent a standard oral glucose tolerance test (OGTT) with 75 g of glucose dissolved in 200mL of water. In this way, fasting and 2-hour venous samples were obtained (8 to 10 h fasting samples were obtained between 8:30am and 10:00am). All samples were analyzed in the same central laboratory. Glucose was determined by the hexokinase enzymatic method. Ferritin levels were analyzed by immunochemiluminiscence in the Architect I2000 analyzer (Abbott Laboratories SA, Madrid, Spain).

A random urine sample was also collected. Serum and urinary samples were frozen until analysis. The samples were managed by the Hospital Regional Universitario de Málaga-IBIMA Biobank that also belongs to the Andalusian Public Health System Biobank, and the biorepository CIBERDEM (Instituto de Salud Carlos III).

Thyroid function tests

TSH, FT4, FT3 and TPO Ab concentrations were analyzed using an electrochemiluminescence immunoassay (Modular Analytics E170, cobas e 602, Roche Diagnostics, Basel, Switzerland). The functional sensitivity of the TSH assay was 0.014mIU/L. The intra-assay coefficients of variation (CV) were: TSH, 1.5–1.2%; FT4 1.8–1.6%; FT3 1.3–2.0.3.0%; and TPO Ab 4.8–2.8%. The inter-assay CV for the low and high levels of serum TSH, FT4, FT3 and TPO Ab quality control materials were 3.5 and 2.7%, 4.17 and 2.64%, 3.78 and 2.21%, 8.5 and 5.2%, respectively. All samples were analyzed at the laboratory of Biochemistry of the Hospital Regional Universitario de Málaga, which attends a population of 487,857 people from the city of Málaga. The reference values with correction according to own studies11,13 were as follows: TSH 0.20–5.20.00mIU/L, FT4 11.0–22.0.0pmol/L, FT3 3.1–6.1.8pmol/L, and TPO Ab < 50IU/L. The same assay technique and reference values were used in both phases of the study.

Urinary iodine (UI) was analyzed using the modified method of Benotti and Benotti14. The intra- and inter-assay CV of the UI assay were 2.01% and 4.53%, respectively. The UI assay was subjected to a program of external quality assessment for the determination of iodine in urine of the Spanish Association of Neonatal Screening (AECNE). All samples were analyzed in the Research Laboratory of the Hospital Regional Universitario de Málaga (Malaga, Spain). UI levels were classified in categories according to current WHO, UNICEF and the Iodine Global Network (IGN; formerly International Council for the Control of Iodine Deficiency Disorders (ICCIDD)) recommendations: <50 µg/L moderate-severe deficiency, 50–99 µg/L mild deficiency, 100–199 µg/L adequate, 200–299 µg/L above requirements, and ≥ 300 µg/L excessive15. UI levels in this population (Di@bet.es) have been previously reported and considered adequate12.

Outcomes

Incident hypothyroidism was defined as having TSH levels above the reference range and/or on treatment with levotiroxine at follow-up examination. Two different TSH cut-off points were considered: (1) TSH ≥ 5mIU/L; and (2) TSH ≥ 10mIU/L. Incident hyperthyroidism was defined as having TSH levels below the reference range (0.2mIU/L) and/or on treatment with thionamides at follow-up examination. TPO Ab were established as positive if above the reference range (≥ 50IU/mL).

Potential exposure factors

The following variables from the initial cross-sectional study were considered: age (< 30, 30–40, 40–50, 50–60, 60–70, ≥ 70 y/o), sex (male, female), smoking status (current smokers, former smokers, never smokers), BMI (< 25, 25–30, ≥ 30 kg/m2), diabetes mellitus (no, yes known diabetes mellitus and unknown diabetes mellitus diagnosed by OGTT), abnormal glucose metabolism according to 1998 World Health Organization criteria16 (normal, prediabetes impaired fasting glucose, impaired glucose tolerance or both diabetes known diabetes mellitus and unknown diabetes mellitus diagnosed by OGTT), TPO Ab (< 20, 20–50, 50–150, ≥ 150IU/mL), TSH (0.20–1.50, 1.50–2.80, 2.80-5mIU/L), UI (50, 50–100, 100–200, 200–300, ≥ 300 µg/L), ferritin (> 100, 100 − 30, 30 − 15, ≤ 15 µg/L), and area of residence (South/East/North East/North/Center).

Statistical analysis

Data were presented as means ± SD, or proportions. Differences in baseline variables according to participation in follow-up were determined by the t-test, the Mann- Whitney test, or chi2 test when appropriate. The sample incidence rates (IRs) were calculated as the number of events/person-time at risk of hypothyroidism TSH ≥ 5mIU/L and TSH ≥ 10mIU/L, or hyperthyroidism respectively, assuming a constant incidence over time. These incidence rates were adjusted by the direct method, using the 2009 Spanish population age and sex structure as a standard (https://www.ine.es/, accessed June 2009). For IRs, 95% confidence intervals were computed.

First, a logistic regression analysis was performed to assess the association between each potential exposure factor with the onset of hypothyroidism. The factors significantly associated with new cases of hypothyroidism in the individual model, were included in a multivariable analysis. Odds ratios (ORs) and 95% confidence intervals (CI) were calculated.

Receiver-operator characteristic (ROC) analysis was calculated to determine the area under the curve (AUC), 95% CI and optimum cut-off points for TPO Ab and TSH at the cross-sectional study as predictors of hypothyroidism.

We decided to limit the logistic regression and ROC curve analysis to TSH ≥ 5 mIU/L as an outcome, because of the limited number of events of hyperthyroidism or hypothyroidism using the TSH cut-off point 10mIU/L available for analyses.

Reported p-values were based on two-sided tests, with statistical significance set at 0.05.

Results

The main characteristics in the cross-sectional study of the subjects included and not included in the incidence study are summarized in Table 1. Both groups had similar age, sex, thyroid function, urinary iodine, and ferritin levels. There was, however, a small imbalance between groups in the proportion of smokers and in the areas of residence.

Table 1 Main characteristics of the population in the cross-sectional study. Comparison between participants and non-participants in the follow-up.
Full size table

TSH levels in the follow-up evaluation were within the normal range (0.20–5.00.20.00 mIU/L) in 95.9% of subjects, below the normal range in 0.8%, and over the normal range in 3.4% of subjects. TPO Ab were positive in 5.7% of the sample. The median IU was 121.63 µg/L (79.15–174.27.15.27).

In total, eighty-three people developed hypothyroidism using the cut-off point TSH ≥ 5mIU/L during 7.5 years of follow-up, with an incidence adjusted for the age and sex structure of the Spanish population was 13.38 cases/1,000 persons-year (IC95% 10.52–16.25). Among these eighty-three subjects, thirty-three were on treatment with levotiroxine at follow-up examination, and three had FT4 levels below 11 pmol/L and were not on thyroid hormone replacement therapy (TSH levels were respectively: 8.11, 6.23 and 14.39 mIU/L; FT4: 10.24, 10.85 and 10.80 pmol/L; FT3: 3.92, 4.44 and 2.96 pmol/L; and TPO Ab: 21.0, 182.5 and 216.8 IU/L).

Forty-two people developed hypothyroidism using the cut-off point of TSH ≥ 10mIU/L during 7.5 years of follow-up, with an incidence of 6.78 cases/1,000 person-year (IC95% 4.73–8.83). Fourteen subjects developed hyperthyroidism during 7.5 years of follow-up. The incidence of hyperthyroidism adjusted for the structure of age and sex of the Spanish population was 2.65 cases/1,000 persons-year (IC95% 1.46–3.83). Table 2; Fig. 3 show the incidence of the three thyroid dysfunction outcomes, according to sex and age groups. The incidence of hypothyroidism TSH ≥ 5mIU/L in men remains low and stable in young and middle-aged subjects and has a small increase in the group aged over 65 years; in contrast the incidence in women seems to be highest at 35–50 years with a small decline in further ages (Table 2; Fig. 3A). The incidence of hypothyroidism TSH ≥ 10mIU/L is very low in men. In women, the highest incidence rates were at age 35–50 years (Table 2; Fig. 3B).

Table 2 Incidence of hypothyroidism and hyperthyroidism according to sex and age groups.
Full size table
Fig. 3
Fig. 3The alternative text for this image may have been generated using AI.
Full size image

Incidence of the different thyroid dysfunction phenotypes in both sexes according to age. (A) Hypothyroidism TSH > 5 mIU/L. (B) Hypothyroidism TSH > 10 mIU/L. (C) Hyperthyroidism.

The incidence of hyperthyroidism by sex and age groups is displayed in Table 2; Fig. 3C. The highest incidence in men was observed in the age group over 65 years, whereas the highest rates in women were at age 18–35 years.

In univariate analysis, female sex, middle age and older age subjects, former smokers, TPO Ab > 20IU/L, TSH > 2.8mIU/L and ferritin < 15 µg/L were significantly associated with an increased risk of developing hypothyroidism (Table 3). When these variables were included together in a multivariable logistic regression analysis, only female sex, age 30–40 years or over 60 years, TPO Ab > 20 IU/L and TSH > 2.8mIU/L were independently associated with incident hypothyroidism (Table 3).

Table 3 Univariate and multivariate analysis to assess potential exposure factors associated with incident hypothyroidism.
Full size table

The baseline TPO Ab cut-off that best predicted hypothyroidism was 9.1IU/L (sensitivity of 67% and specificity of 60%). The AUC was 0.654 (IC95% 0.583–0.725; p < 0.001) (Fig. 4A). For baseline TSH, a cut-off point of 2.03mIU/L was obtained (sensitivity of 78% and specificity of 61%). AUC was 0.702 (IC95% 0.639–0.766; p < 0.001) (Fig. 4B).

Fig. 4
Fig. 4The alternative text for this image may have been generated using AI.
Full size image

ROC curve. (A) Baseline TPOAb. (B) Baseline TSH.

Discussion

The results of the present study are the first published data about the incidence of thyroid dysfunction in the Spanish adult population.

There were previous reports aimed at evaluating the incidence of thyroid dysfunction in other European populations, but unlike the current study, they were conducted in specific geographical areas. The Wickham cohort carried out in the United Kingdom, showed an incidence of hypothyroidism of 3.5–5.0 per 1,000 persons-year in women and 0.6–1.0 per 1,000 persons-year in men; and an incidence of hyperthyroidism of 80 per 100,000 persons-year in women.6 In a metanalysis about the incidence of thyroid dysfunction in Europe, the authors described a rate of 2.26 cases of hypothyroidism per 1,000 persons-year; and a rate of 0.51 cases of hyperthyroidism per 1,000 persons-year3. In other populations, a recent Iranian cohort study reported an incidence of overt hypothyroidism (TSH > 10.0 mIU/L) of 2.7 per 1,000 persons-year, an incidence of subclinical hypothyroidism (TSH 3.6–10.0 mIU/L) of 20.6 per 1,000 person-year, an incidence of over hyperthyroidism of 1.9 per 1,000 persons-year, and an incidence of subclinical hyperthyroidism of 2.7 per 1,000 persons-year7.

The wide variation in the incidence rates of thyroid dysfunction in previous studies may be explained by different diagnostic criteria for hypothyroidism/hyperthyroidism, differences in race/ethnicity, environmental factors and iodine nutritional status between populations. The various laboratory techniques and reference ranges in the measurements of thyroid hormones that were used in those studies1,11 can also influence the variability of the results. For these reasons, it would be advisable for each country to have its own data on the incidence of these disorders.

In addition, it is of great interest to investigate the factors that might promote the onset of hypothyroidism in any population to design strategies for the early detection of this disease. We investigated factors associated with the thyroid function such as age, sex, TPO Ab, TSH, but also new or controversial factors such as ferritin, iodine or the presence of diabetes and obesity.

The relationship between the TPO Ab levels and TSH levels within the normal range with incident hypothyroidism has been previously described in other longitudinal studies7,17,18,19,20, and it is confirmed with our results. Specifically, TSH > 2.03 mIU/L and TPO Ab > 9.1 IU/L were the cut-offs which best predicted new cases of hypothyroidism in our population. TSH levels in the range of > 2–2.5 mIU/L have been previously proposed to be associated with incident hypothyroidism in other cohort studies.17,18 However, our cut-off point of TPO Ab is lower than that published in other studies, such as the Australian longitudinal study (29 IU/L)18 or a recent study from Iran (38 IU/L)7,17. In any case, these calculated cut-off points for TPO Ab in longitudinal studies are lower than the normal ranges included in most commercial kits. This highlights that the risk of hypothyroidism may already be relevant within “high normal” TSH and TPO levels, and that a monitoring strategy could be needed even below these “normal” thresholds. More studies are needed to assess the impact of incorporating this strategy to improve early detection and management of thyroid dysfunction in clinical practice.

We also investigated the role of ferritin and trace elements such as iodine, which have been associated with hypothyroidism and hyperthyroidism, and also with the development of thyroid autoimmunity21,22. Our previous work showed a cross sectional link between iron deficiency and low FT4 and FT3 levels in the adult Spanish general population23. In this new analysis of the di@bet.es study cohort, we indeed found that low ferritin levels (< 15 µg/L) were significantly associated with incident hypothyroidism in the univariate analysis. However, this association did not remain statistically significant in the multivariable model. Regarding the role of iodine status, our results did not find any association between iodine urinary excretion and new cases of hypothyroidism. It is important to note that our results may be limited by the relatively small sample size with a low number of events available, so further investigations regarding the relation between ferritin, iodine and other trace elements with thyroid dysfunction, in larger population-based cohort studies and from other countries are needed.

Recent cross-sectional studies have shown an increased prevalence of both autoimmune hypothyroidism24 and subclinical autoimmune hypothyroidism in individuals with obesity25, however results from longitudinal cohorts remain inconclusive. Our results showed indeed that BMI ≥ 30 kg/m2 was not associated with incident hypothyroidism. Similarly, in the Australian cohort, there was no association between BMI and hypothyroidism18. On the contrary, in an older population, a five-year longitudinal study showed that BMI ≥ 30 kg/m2 was significantly associated with the development of hypothyroidism19. A recent cohort study in an Asian population found that both obesity and metabolic abnormalities were associated with a higher risk of hypothyroidism in males, but not in females.

The hypothetical bidirectional relationship between diabetes and hypothyroidism has been previously evaluated. Two recent metanalyses of prospective studies found that hypothyroidism increased the risk of incident diabetes26,27. However, the role of subclinical thyroid dysfunction on the risk of developing type 2 diabetes has not been confirmed in a recent study using the individual participant data of multiple prospective cohort studies, including the di@bet.es study28. Regarding the risk of thyroid dysfunction in subjects with type 2 diabetes, a metanalysis that included data from cross-sectional studies, found that subjects with type 2 diabetes had a higher risk of subclinical hypothyroidism29. Nevertheless, data from longitudinal studies are limited. Gopinath et al. found no increased risk of hypothyroidism in people aged over 55 years old with type 2 diabetes30. Results from the Tehran Thyroid Study31 showed no differences in the incidence of thyroid dysfunction in subjects with type 2 diabetes, prediabetes, or normal glucose metabolism in the adult Iranian population. Like previous cohort studies, we did not find an increased risk of developing hypothyroidism in subjects without diabetes or prediabetes at baseline.

As a strength, this is a population-based cohort study with a representative sample of the Spanish adult population with a mean follow-up period of 7.5 years. Data were obtained by interview, blood and urine samples were collected, including an OGTT to assess glucose metabolism, ferritin and trace elements such as iodine were measured. Therefore, we present a longitudinal study that is representative of the general Spanish population in which a comprehensive approach to risk factors for thyroid dysfunction was performed.

However, we acknowledge that the study has some limitations that should be considered when interpreting its results, including a relatively small sample size and an incomplete participation in the follow up evaluation. To rule out possible selection bias, we compared the baseline characteristics of participants and non-participants at the re-evaluation visit. Age, sex and baseline thyroid and thyroid antibody status did not differ significantly between groups, so we can consider our results to be reliable. There was however a small imbalance in the participation in the different geographical areas. We believe that these differences are of little relevance due to their small magnitude.

Patients with severe concomitant diseases were excluded to avoid potential confounding, as non-thyroidal illness can alter thyroid function tests independently of true thyroid dysfunction. This exclusion criterion was applied to ensure that the incidence assessed reflected genuine thyroid-related abnormalities. While we acknowledge that thyroid dysfunction can occasionally contribute to the severity of underlying conditions, the available clinical information did not suggest this was the case for the excluded patients.

Being a cohort study with 2 evaluations, we could not asses TSH values at the time of starting levotiroxine or thionamides, assuming participants under these medications at follow up, as true cases of hypo and hyperthyroidism respectively. Finally, while our 7.5 follow-up period is sufficient to capture the medium-term incidence of thyroid dysfunction, a longer follow up may be necessary to cover the long-term spectrum of the disease.

In conclusion, the present work provides data on the incidence rate of thyroid dysfunction in a nationwide cohort of the adult Spanish population, as well as associated risk factors for hypothyroidism. Further longitudinal studies are needed to expand the knowledge of the incidence of the disease in different populations and to clarify the relationships of ferritin and trace elements (iodine), obesity and diabetes on its development.