Prevalence of Thyroid Dysfunction in Autoimmune and Type 2 Diabetes: The Population-Based HUNT Study in Norway

Abstract

Context:

Associations between autoimmune diabetes and autoimmune thyroid disease are known but insufficiently characterized. Some evidence suggests that type 2 diabetes may also be associated with hypothyroidism.

Objective:

The objective of the study was to investigate associations of autoimmune and type 2 diabetes with the prevalence of hypo- and hyperthyroidism.

Design and Setting:

This was a cross-sectional, population-based study of adults in two surveys of the Nord-Trøndelag Health (HUNT) Study.

Participants:

A total of 34 235 participants of HUNT2 (1995–1997) and 48 809 participants of HUNT3 (2006–2008) participated in the study.

Main Outcome Measures:

Prevalence of hypo- and hyperthyroidism was estimated, assessed by self-report, serum measurements, and linkage with the Norwegian Prescription Database.

Results:

In HUNT2, autoimmune diabetes was associated with a higher age-adjusted prevalence of hypothyroidism among both women (prevalence ratio 1.79, 95% confidence interval [CI] 1.30–2.47) and men (prevalence ratio 2.71, 95% CI 1.76–4.19), compared with having no diabetes. For hyperthyroidism, the corresponding cumulative prevalence ratios were 2.12 (95% CI 1.36–3.32) in women and 2.54 (95% CI 1.24–5.18) in men with autoimmune diabetes. The age-adjusted excess prevalence of hypothyroidism (∼6 percentage points) and the presence of thyroid peroxidase antibodies (8–10 percentage points) associated with autoimmune diabetes was similar in women and men. Type 2 diabetes was not associated with the prevalence of hypothyroidism. In HUNT3, associations were broadly similar to those in HUNT2.

Conclusions:

Autoimmune diabetes, but not type 2 diabetes, was strongly and gender neutrally associated with an increased prevalence of hypo- and hyperthyroidism and the presence of thyroid peroxidase antibodies. Increased surveillance for hypothyroidism appears not necessary in patients with type 2 diabetes.

Autoimmune diabetes and autoimmune thyroid disease share some susceptibility genotypes within human leukocyte antigen and other genes involved in immune regulation (1, 2), and people with autoimmune diabetes are at increased risk of autoimmune thyroiditis, hypothyroidism, and Graves' disease (3, 4). Therefore, regular TSH measurements may be warranted in patients with autoimmune diabetes (5–7). The evidence quantifying the association between autoimmune diabetes and thyroid disease mostly comes from hospital-based studies of children or adolescents with type 1 diabetes (3, 4, 8). Some studies have reported a high frequency of thyroid disease in adults with autoimmune diabetes (9–13). Two of these studies included a nondiabetic comparison group and observed an increased prevalence of autoimmune thyroid disease among adults with type 1 diabetes (13) or latent autoimmune diabetes in adults (9).

Associations between type 2 diabetes and thyroid dysfunction are less clear, but one cross-sectional study observed increased prevalence of hypothyroidism in patients with type 2 diabetes and suggested that thyroid screening may be warranted in these patients (14). In support of that observation, a recent prospective study reported increased risk of diabetes among patients with hypothyroidism (15), and some evidence suggests an association between hypothyroidism and insulin resistance (16, 17). However, other cross-sectional studies found no association between type 2 diabetes and hypothyroidism (18, 19), and one small prospective study showed no convincing evidence that people with type 2 diabetes were at increased risk of hypothyroidism (20). A recent systematic review, largely based on Chinese study populations, reported an increased prevalence of subclinical hypothyroidism among patients with type 2 diabetes (21).

Population-based studies of the frequency and timing of hypo- and hyperthyroidism in adults with different forms of diabetes are lacking. We therefore examined associations of autoimmune and type 2 diabetes with the prevalence of thyroid dysfunction in two large population-based surveys within the Nord-Trøndelag Health (HUNT) Study in Norway.

Subjects and Methods

The HUNT Study

The HUNT Study includes three surveys (HUNT1–3) of the population of Nord-Trøndelag County in Norway. All inhabitants aged 20 years or older were invited to participate in the surveys, which included extensive questionnaires, clinical examinations, and, for HUNT2 (1995–1997) and HUNT3 (2006–2008), nonfasting blood sampling of all individuals. Self-reported information on previously known diabetes was collected in all surveys. Measurements of thyroid function were included in HUNT2 and HUNT3, which therefore form the basis of this study. The number of participants was 65 215 (69%) of 93 898 invited people in HUNT2 and 50 807 (54%) of 93 860 invited people in HUNT3 (22).

Classification of diabetes

At each survey, participants were asked about their history of diabetes and their age at diagnosis. Participants with self-reported diabetes were invited to an additional, fasting examination, which included measurement of autoantibodies against glutamic acid decarboxylase (GADA) (23). For diabetic participants in HUNT2 who did not attend this additional examination, GADA measurements were available from the serum sample drawn at the main HUNT2 examination. Information on GADA was available for 1869 of 1972 diabetic participants in HUNT2 (95%) and for 1400 of 2189 in HUNT3 (64%).

We classified participants as having autoimmune diabetes if they reported having diabetes and either tested positive for GADA in HUNT2 or HUNT3 and/or were diagnosed with diabetes 30 years of age or younger. We classified participants as having type 2 diabetes if they reported being diagnosed with diabetes older than 30 years, were GADA negative and, if they currently used insulin, did not start insulin treatment within 12 months after the diabetes diagnosis. Participants who lacked information on GADA or insulin treatment, but who otherwise fulfilled the criteria for type 2 diabetes, were classified as type 2 diabetes in the main analyses. In supplemental analyses, these participants were excluded to reduce potential misclassification of autoimmune diabetes as type 2 diabetes.

Classification of thyroid dysfunction

In HUNT2, subsamples were selected for measurement of serum TSH concentration, including all women aged 40 years or older, a 50% random sample of men 40 years or older, 5% random samples of women and men younger than 40 years, and all participants with self-reported diabetes. In total, TSH was measured in 34 484 individuals from these samples. In HUNT3, TSH measurement was planned for all participants and was measured in 49 181 individuals. At both surveys, participants self-reported current or previous thyroid diseases, including hypothyroidism, hyperthyroidism, and use of thionamides. At HUNT2, participants were also asked about current or previous use of levothyroxine (24). To obtain information on the use of thyroid medication at HUNT3, we used the unique 11-digit personal identification numbers for Norwegian citizens to link the study data to individual-level information on dispensing of levothyroxine and thionamides from the Norwegian Prescription Database (NorPD). NorPD includes information on all prescription drugs dispensed by pharmacies to noninstitutionalized inhabitants in Norway since January 2004. In this study we used information on dates of dispensing of levothyroxine and thionamides between January 2004 and each individual's date of participation at HUNT3 between 2006 and 2008.

We classified hypothyroid participants as having treated or untreated hypothyroidism. Treated hypothyroidism was defined as use of levothyroxine without use of thionamides, as indicated by self-report for HUNT2 and by information from NorPD for HUNT3. Untreated hypothyroidism was defined as serum TSH greater than 4.5 mU/L in participants who did not use levothyroxine or thionamides. Hyperthyroid participants were classified as having previously diagnosed or undiagnosed hyperthyroidism. Diagnosed hyperthyroidism was defined as self-reported hyperthyroidism or use of thionamides in HUNT2 and, for HUNT3, as self-reported hyperthyroidism or prescription of thionamides recorded in NorPD. We defined undiagnosed hyperthyroidism as serum TSH less than 0.45 mU/L in participants without diagnosed hyperthyroidism and who did not use levothyroxine. Hyperthyroid disease is often transient or definitely cured by surgery or radioiodine. Because the questionnaire did not separate between current and previous thyroid disease, our results for hyperthyroidism represent the participants' cumulative prevalence of hyperthyroidism. In sensitivity analyses, mildly elevated serum TSH (4.6–9.9 mU/L) without serum-free T₄ below the reference range did not qualify as untreated hypothyroidism, and mildly reduced serum TSH (0.10–0.44 mU/L) without free T₄ or total T₃ concentration above the reference range did not qualify as undiagnosed hyperthyroidism.

Laboratory measurements

GADA was analyzed at the Aker Hormone Laboratory, Oslo University Hospital, as described previously (23). Antibody levels were expressed as an antibody index relative to a standard serum. An index of 0.08 or greater was considered positive (23).

TSH measurements were carried out at the Aker Hormone Laboratory (for HUNT2) and at Levanger Hospital, Nord-Trøndelag Hospital Trust (for HUNT3). In both surveys, the laboratory's reference range for TSH was 0.2–4.5 mU/L, but we used TSH concentrations of 0.45 and 4.5 mU/L as cutoff values to indicate biochemical hyperthyroidism and hypothyroidism, respectively, as in recent studies by the Thyroid Studies Collaboration (25, 26). In HUNT2, serum-free T₄ (reference range 8.0–20.0 pmol/L) was measured if TSH was less than 0.20 or greater than 4.0 mU/L, and serum total T₃ (reference range 1.2–2.7 nmol/L) was measured if TSH was less than 0.20 mU/L. In HUNT3, serum-free T₄ (reference range 9.0–19.0 pmol/L) was measured if TSH was less than 0.10 or greater than 3.0 mU/L. The methods for TSH and thyroid hormone measurements have been described in detail previously (24, 27).

In HUNT2, subgroups were selected for serum measurement of thyroid peroxidase antibodies (TPOAbs), including all participants with serum TSH greater than 4.0 mU/L, a 3% random sample of participants with serum TSH of 4.0 mU/L or less, and all participants with diabetes who attended the additional diabetes examination. TPOAb was measured at the Aker Hormone Laboratory using a luminoimmunoassay from B.R.A.H.M.S. Diagnostica GmbH. The laboratory's upper reference limit was 200 U/mL. During the measurements of serum samples from the additional diabetes examination, the laboratory switched to another TPOAb assay, the DYNOtest anti-TPOn competitive RIA, also from B.R.A.H.M.S. Diagnostica GmbH. For that assay, we used the upper reference limit of 60 U/mL that was provided by the manufacturer. No measurements among people without diabetes and 88% of measurements among people with diabetes were performed using the latter assay. Positive TPOAb status, indicating the presence of autoimmune thyroid disease, was defined as serum TPOAb concentration higher than the upper reference limit, and strongly positive TPOAb status was defined as serum TPOAb concentration higher than 5 times the upper reference limit. TPOAb measurements were available for 2633 participants without diabetes, 252 with autoimmune diabetes, and 1067 with type 2 diabetes.

Statistical analysis

Among 34 484 participants with TSH measurement in HUNT2, we excluded 105 participants who did not reply to the questions on previously diagnosed diabetes and thyroid disease and 144 participants whose diabetes could not be classified as either autoimmune or type 2, leaving 34 235 participants for analysis. Among 49 181 participants with TSH measurement in HUNT3, we excluded 11 individuals who did not reply to the question on previously diagnosed diabetes and 361 participants whose diabetes could not be classified, leaving 48 809 participants for analysis. The analysis of hyperthyroidism in HUNT3 was further restricted to 39 940 participants who returned (by mail in a prepaid envelope) the questionnaire that included items on previously diagnosed thyroid diseases.

For each survey, we calculated the prevalence (with 95% confidence interval [CI]) of each type of hypo- and hyperthyroidism among participants with autoimmune diabetes, type 2 diabetes, and no diabetes. We used log-binomial regression analysis to estimate age-adjusted (using 10 y age categories) prevalence ratios (PR; with 95% CI) of each type of thyroid dysfunction among participants with autoimmune diabetes and type 2 diabetes, compared with nondiabetic participants. We analyzed women and men separately because both hypo- and hyperthyroidism are more common in women than in men (28). We additionally adjusted for smoking (never, former, and current daily smoking) and body mass index (BMI; weight in kilograms divided by the squared value of height in meters [<25.0, 25.0–29.9, 30.0–34.9, and ≥35.0 kg/m²]), which are associated with the risk of autoimmune (29–31) and type 2 diabetes (30, 31), and with the risk of thyroid dysfunction or abnormal TSH concentrations (32–34). From the model, we estimated the age-adjusted excess prevalence associated with autoimmune diabetes.

In analogous analyses, we estimated the prevalence of positive TPOAb status and the age- and smoking-adjusted (35) PRs of positive TPOAb status among participants with autoimmune and type 2 diabetes, compared with nondiabetic participants. This analysis was restricted to participants 40 years of age or older because due to the sampling criteria, very few younger nondiabetic participants with normal TSH concentration had their TPOAb measured. Among people with TPOAb measurement, we also estimated the association of diabetes with autoimmune hypothyroidism, defined as hypothyroidism combined with positive TPOAb status. We additionally estimated the PR of positive TPOAb status among participants with autoimmune compared with type 2 diabetes.

Among participants with previously diagnosed hypothyroidism, we assessed whether autoimmune diabetes was associated with an earlier onset of hypothyroidism. For that purpose, we used linear regression analysis to examine whether the mean self-reported age at diagnosis of hypothyroidism differed between participants with autoimmune and no diabetes. We adjusted for age at participation (as a continuous variable) and sex. We had too few participants with autoimmune diabetes and thyroid dysfunction to perform sex-specific analyses and analyses of age at onset of hyperthyroidism. To evaluate whether people with diabetes might have benefited from screening for hypothyroidism, we calculated the proportion of hypothyroid cases that were diagnosed the same year as diabetes or later.

In analyses of HUNT2, participants were weighted to account for differences in the probability of being sampled for TSH or TPOAb measurement. The statistical analyses were performed using Stata for Windows, version 13.1 (StataCorp).

All participants gave informed consent. The study was approved by the Regional Committee for Medical and Health Research Ethics and by the Norwegian Data Inspectorate.

Results

Characteristics of the participants are shown in Table 1.

Autoimmune diabetes and the prevalence of hypo- and hyperthyroidism

In HUNT2, autoimmune diabetes was associated with higher age-adjusted prevalence of hypothyroidism among women (PR 1.79, 95% CI 1.30–2.47) and men (PR 2.71, 95% CI 1.76–4.19), compared with having no diabetes. Autoimmune diabetes was associated with higher prevalence of treated hypothyroidism but not untreated hypothyroidism (Table 2). In HUNT3 (2006–2008), the association between autoimmune diabetes and hypothyroidism was similar but slightly stronger than in HUNT2 (Table 3). The age-adjusted excess prevalence of hypothyroidism associated with autoimmune diabetes was similar in women (HUNT2: 6.6 percentage points [pp], 95% CI 1.8–11.4; HUNT3: 11.2 pp, 95% CI 4.2–18.2) and men (HUNT2: 5.4 pp, 95% CI 1.8–9.0; HUNT3: 9.4 pp, 95% CI 3.6–15.2).

In HUNT2, autoimmune diabetes was associated with higher age-adjusted cumulative prevalence of hyperthyroidism among women (PR 2.12, 95% CI 1.36–3.32) and men (PR 2.54, 95% CI 1.24–5.18), compared with having no diabetes. The PRs did not convincingly differ between diagnosed and undiagnosed hyperthyroidism (Table 2). In HUNT3, the association of autoimmune diabetes with hyperthyroidism in men was similar to that observed in HUNT2. We observed no association between autoimmune diabetes and hyperthyroidism among women in HUNT3 but a low number of cases prevented precise estimates (Table 3). Additional adjustment for smoking and BMI did not substantially change the associations between autoimmune diabetes and thyroid dysfunction (Tables 2 and 3). Sensitivity analyses, in which mildly abnormal TSH with normal thyroid hormone concentration did not qualify as thyroid dysfunction, yielded broadly similar results to the main analyses (Supplemental Tables 1 and 2).

Type 2 diabetes and the prevalence of hypo- and hyperthyroidism

In HUNT2, type 2 diabetes was not associated with the overall prevalence of hypothyroidism. Among women, type 2 diabetes was associated with a slightly higher age-adjusted prevalence of treated hypothyroidism that was counterbalanced by a lower prevalence of untreated hypothyroidism (Table 2). In HUNT3, type 2 diabetes was associated with a slightly higher prevalence of hypothyroidism among women (PR 1.24, 95% CI 1.04–1.47), with similar but less precise estimates among men (PR 1.25, 95% CI 0.93–1.67). However, this association was attenuated after adjustment for BMI and smoking (Table 3). After exclusion of participants with type 2 diabetes who lacked information on GADA or insulin treatment, the age-adjusted PR of hypothyroidism comparing type 2 with no diabetes in HUNT3 was attenuated among women (PR 1.10, 95% CI 0.86–1.41), although not among men (PR 1.29, 95% CI 0.88–1.89).

In HUNT2, type 2 diabetes was associated with higher age-adjusted prevalence of any (PR 1.75, 95% CI 1.12–2.73) and undiagnosed hyperthyroidism (PR 2.16, 95% CI 1.30–3.60) in men, and the association did not attenuate after adjustment for smoking and BMI (Table 2). There was no similar association among women (Table 2) and no association between type 2 diabetes and hyperthyroidism in HUNT3 (Table 3). Sensitivity analyses (Supplemental Tables 1 and 2), in which mildly reduced TSH without elevated thyroid hormone concentration did not qualify as undiagnosed hyperthyroidism, showed no association between type 2 diabetes and hyperthyroidism.

Diabetes and TPOAb status in HUNT2

Compared with having no diabetes, autoimmune diabetes was associated with higher age-adjusted prevalence of positive TPOAb status among men (PR 3.00, 95% CI 1.54–5.85), but the association was less convincing among women (PR 1.41, 95% CI 0.96–2.06). The prevalence of strongly positive TPOAb status was higher among both women (PR 1.77, 95% CI 1.10–2.83) and men (PR 3.58, 95% CI 1.72–7.49) with autoimmune diabetes. The associations were nearly unchanged after additional adjustment for smoking. There was no association between type 2 diabetes and TPOAb status (Table 4). The age-adjusted excess prevalence of positive TPOAb status associated with autoimmune diabetes was similar in women (7.9 pp, 95% CI −2.0 to 17.8) and men (10.1 pp, 95% CI 2.2–17.9). The analogous estimates for strongly positive TPOAb status were 8.5 pp (95% CI −0.1 to 17.2) in women and 9.1 pp (95% CI 2.2–16.0) in men. In the subsample with TPOAb measurement, we observed a strong positive association between autoimmune diabetes and autoimmune hypothyroidism in men, but not in women, and no association between type 2 diabetes and autoimmune hypothyroidism (Supplemental Table 3).

The difference in TPOAb assay could influence the comparison of TPOAb status between participants with and without diabetes but not the comparison between autoimmune and type 2 diabetes. Among women, autoimmune diabetes was associated with higher age- and smoking-adjusted prevalence of both positive (PR 1.75, 95% CI 1.19–2.58) and strongly positive (PR 2.23, 95% CI 1.35–3.70) TPOAb status, compared with type 2 diabetes. Among men, the analogous PRs were 2.12 (95% CI 1.25–3.59) for positive and 3.06 (95% CI 1.66–5.63) for strongly positive TPOAb status.

Age at diagnosis of hypothyroidism in autoimmune vs no diabetes

Compared with nondiabetic participants, mean age at diagnosis of hypothyroidism among participants with autoimmune diabetes was slightly younger both in HUNT2 (mean difference 4.0 y, 95% CI −0.1 to 8.1) and in HUNT3 (mean difference 2.4 y, 95% CI −1.1 to 6.0) (Table 5).

Order of diagnosis of hypothyroidism and diagnosis of diabetes

Approximately two-thirds of participants with autoimmune diabetes and previously diagnosed hypothyroidism had hypothyroidism diagnosed the same year as diabetes or later. Conversely, half to two-thirds of participants with type 2 diabetes and hypothyroidism had hypothyroidism diagnosed before diabetes (Table 6).

Discussion

In this large population-based study, we found strong associations between autoimmune diabetes and the prevalence of hypothyroidism, hyperthyroidism, and positive TPOAb status in adulthood. These findings confirm and extend the evidence that autoimmune diabetes is associated with thyroid dysfunction (3, 4, 8–13). In particular, we demonstrated that the excess prevalence of hypothyroidism and presence of TPOAb was similar in women and men, in contrast to the strong female preponderance of hypothyroidism in the nondiabetic population. Another major finding is the absence of association between type 2 diabetes and hypothyroidism, which contrasts with the results of some studies (14, 15, 21), but agrees with the results of several others (18–20).

Major strengths of our study are the large sample size, which enabled more precise estimates, and the population-based setting, which carries less risk of selection bias compared with hospital-based studies that have previously examined associations between diabetes and thyroid dysfunction. A limitation is that for participants with a history of thyroid dysfunction, we did not know their thyroid function at the time of diagnosis. The two surveys in our study differed somewhat in strengths and weaknesses. One strength of HUNT3 is that the information on thyroid medication was obtained from the national prescription database, which may be more accurate than the self-report in HUNT2. On the other hand, HUNT3 had a lower participation rate and a higher proportion of diabetic participants lacking information on GADA. Reassuringly, the two surveys yielded broadly similar results. In a study comparing nonparticipants with participants of HUNT3, previously diagnosed diabetes was more common among nonparticipants, but the prevalence of previously diagnosed thyroid dysfunction did not substantially differ between participants and nonparticipants (36).

The study was performed in a long-time iodine-replete area (37), and autoimmune thyroid disease is likely to be the most common cause of hypo- and hyperthyroidism. Nonetheless, nonautoimmune causes of thyroid dysfunction or abnormal TSH concentrations may have influenced our results. A possible explanation for the association between type 2 diabetes and slightly increased prevalence of hypothyroidism in HUNT3 may be that adiposity, which is common in type 2 diabetes, may cause increased pituitary TSH secretion that may lead to a false diagnosis of hypothyroidism (33, 34). In support of that explanation, the association of type 2 diabetes with hypothyroidism was attenuated after statistical adjustment for BMI and was apparent only in HUNT3, which was performed in a time period when levothyroxine treatment for mild TSH elevation appeared to be more common than in HUNT2 (27).

Another possible explanation for the association between type 2 diabetes and hypothyroidism in HUNT3 is the lack of GADA measurement in many participants with diabetes in that survey, possibly leading to misclassification of latent autoimmune diabetes in adults as type 2 diabetes. On the other hand, maturity-onset diabetes of the young or early-onset type 2 diabetes may have been misclassified as autoimmune diabetes because we classified all diabetes with onset at 30 years of age or younger as autoimmune diabetes. However, the extent of such misclassification is likely to be low and would, if anything, attenuate the association of autoimmune diabetes with thyroid dysfunction.

The well-known female preponderance of autoimmune hypothyroidism and hyperthyroidism (8, 28) was also evident in this study, among participants with autoimmune or type 2 diabetes as well as among nondiabetic participants. However, the absolute increases in the prevalence of hypothyroidism and positive TPOAb status associated with autoimmune diabetes were similar in women and men. Hence, the impact of diabetes-associated autoimmunity on autoimmune hypothyroidism appears gender neutral, a notion that agrees with the prevalence of type 1 diabetes being similar in women and men (38). This suggests that the autoimmune mechanisms behind the added risk of hypothyroidism in autoimmune diabetes differ from those governing the female preponderance of hypothyroidism.

Our results suggest that people with autoimmune diabetes may be prone to earlier development of hypothyroidism because their age at diagnosis tended to be lower than among people without diabetes. Alternatively, increased surveillance for hypothyroidism among people with autoimmune diabetes may contribute to their earlier diagnosis. The risk of autoimmune thyroid disease may be especially high among people who develop autoimmune diabetes at an early age, but low numbers precluded separate analyses of this subgroup.

The association between type 2 diabetes and hyperthyroidism that we observed among men in HUNT2 may have several explanations. First, hyperthyroidism may increase insulin resistance and thus contribute to the development of type 2 diabetes (17). Second, low TSH concentrations, leading to a false diagnosis of subclinical hyperthyroidism, might be a consequence of hypercortisolism or nonthyroidal illness (39, 40), which may be more common among people with type 2 diabetes. Third, the lack of similar associations among women or in HUNT3 suggests that the association may be a chance finding.

In summary, this large population-based study showed strong and gender-neutral associations of autoimmune diabetes with increased prevalence of hypo- and hyperthyroidism and presence of TPOAbs. Conversely, type 2 diabetes was not convincingly associated with the prevalence of hypo- or hyperthyroidism or presence of TPOAbs. The findings support the recommendation of regular follow-up of thyroid function in people with autoimmune diabetes. On the other hand, our results do not support proposals for increased surveillance of hypothyroidism in people with type 2 diabetes.

Acknowledgments

The HUNT Study is a collaborative effort of the HUNT Research Centre (Faculty of Medicine, The Norwegian University of Science and Technology), Nord-Trøndelag County Council, Central Norway Health Authority, and the Norwegian Institute of Public Health.

The HUNT Research Center and the Norwegian Prescription Database provided the data for this study.

This work was supported by the Research Council of Norway. Financial support was also obtained from Wallac Oy (Turku, Finland) for the TSH measurements in HUNT2 and from GlaxoSmithKline Norway AS (Oslo, Norway), who supported the diabetes part of HUNT2 and HUNT3.

Disclosure Summary: The authors have nothing to disclose.

Abbreviations

 
• BMI

  body mass index

 
• CI

  confidence interval

 
• GADA

  glutamic acid decarboxylase antibodies

 
• HUNT

  Nord-Trøndelag Health Study

 
• NorPD

  Norwegian Prescription Database

 
• pp

  percentage points

 
• PR

  prevalence ratio

 
• TPOAb

  thyroid peroxidase antibody.

References

Author notes