Prevalence of Undiagnosed Hypothyroidism in Europe: A Systematic Review and Meta-Analysis

Hypothyroidism is a condition of thyroid hormone deficiency which is essentially defined based on biochemical parameters. Overt hypothyroidism is the combination of an elevated level of serum thyroid-stimulating hormone (TSH) with a decreased level of serum free thyroxin (fT₄) as compared to the reference ranges in the general population. Subclinical hypothyroidism is defined as an elevated serum TSH level in combination with a normal serum fT₄ level [1, 2]. The reference ranges for TSH and fT₄ currently used to define thyroid dysfunction are subject of discussion because of the arbitrary nature of the cutoffs. This issue is of clinical importance for diagnosis and treatment decision purposes [1, 3].

The prevalence of hypothyroidism varies considerably across the general population. There is a number of factors that can influence the prevalence of this condition. For example, the occurrence of hypothyroidism is affected by differences in the iodine status between populations, with higher prevalence among those with high iodine intake and in severely iodine-deficient populations [4]. The prevalence of hypothyroidism is also higher in women, in senior populations, and in Caucasian individuals [5-9]. Furthermore, there is an increased risk of hypothyroidism in patients with autoimmune diseases, including diabetes mellitus, rheumatoid arthritis, or systemic lupus erythematosus, and also in patients with other conditions, such as HIV infection [10-14].

Several studies have been conducted to estimate the prevalence of hypothyroidism. The estimates can vary between 0.1 and 12.5%, depending on the definition used [15, 16]. For example, some studies evaluated the prevalence of undiagnosed and untreated subclinical and/or overt hypothyroidism [17-19], while others addressed previously diagnosed and treated hypothyroidism [20, 21]. Patients with undiagnosed hypothyroidism are not treated for the disease and therefore might be at higher risk of developing long-term complications, such as serious and even fatal cardiovascular diseases, diabetes, or others [22-25]. This situation may result in important implications for public health. Although a systematic review with a meta-analysis on the European prevalence and incidence of (undiagnosed and diagnosed) thyroid dysfunction has been published, that work considered only studies that reported simultaneously on both hypo- and hyperthyroidism. This approach led to the exclusion of studies that exclusively addressed hypothyroidism. Furthermore, several studies were concluded after the publication of that meta-analysis [26]. Therefore, there is rationale to perform a systematic review and meta-analysis aiming to determine the prevalence of undiagnosed hypothyroidism in Europe.

This study conforms to standard guidelines and was written according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement [27]. Online supplementary Table 1 (for all online suppl. material, see www.karger.com/doi/10.1159/000499751) presents the PRISMA checklist.

The Medline, EMBASE, and Cochrane Central databases were searched for articles published between January 2008 and April 2018. Bibliographic reference lists of all relevant studies, meta-analyses, and systematic reviews were hand searched to identify additional eligible articles. The electronic database search strategy is available in online supplementary Tables 2 and 3.

The titles and abstracts of all retrieved citations were screened by two independent reviewers (D.M. and C.A.) to identify potentially relevant publications. Full texts were retrieved for relevant citations. Discrepancies were resolved by majority decision (two of three) involving a third investigator (F.B.M.).

Studies were included if they met the following criteria: (1) based on a European population, (2) included participants from the general population (without age or sex restrictions) without previously known thyroid disease, (3) provided epidemiological data on the prevalence of subclinical, overt, and/or total hypothyroidism, and (4) provided definitions for subclinical and overt hypothyroidism according to laboratory measurements, namely TSH and fT₄ levels. Studies involving only participants with underlying diseases (e.g., diabetes) or conditions (e.g., pregnancy) were excluded. Reviews, case reports, abstracts, and conference proceedings were excluded. No further exclusion criteria were applied.

Data were extracted by two independent reviewers (D.M. and C.A.). The data retrieved from each study included first author’s name, bibliographic reference, year of publication, country/region, study design, demographic characteristics (mean age, proportion of females, number of subjects included), prevalence of hypothyroidism, reference ranges for thyroid function tests and thyroid disorder categories, as well as assays used to measure thyroid function in each study.

The Newcastle-Ottawa Scale was used to assess the methodological quality of the included studies. This scale considers the following: selection of the study groups, comparability of the groups, and ascertainment of either the exposure (for case-control studies) or outcome of interest (for cohort studies) [28]. A maximum of 1 point for each item within the “selection” and “exposure/outcome” categories could be awarded. For the “comparability” category, a maximum of 2 points could be awarded. The summary score equals the number of points earned by each study, totaling a maximum of 9 (maximum of 8 points for cross-sectional analysis). An adapted form of the Newcastle-Ottawa Scale was used to assess cross-sectional studies [29]. Studies scoring ≥7 points were considered to be of good quality, those scoring <7 and ≥5 points were considered to be of moderate quality, and those scoring <5 points were considered to be of poor quality.

The co-primary outcomes were the proportion of people with undiagnosed (1) subclinical, (2) overt, and (3) total hypothyroidism (i.e., the sum of subclinical and overt cases). The classification of cases as subclinical or overt hypothyroidism was the one provided by the authors of the studies included in this systematic review.

Pooled estimates of proportions with corresponding 95% confidence intervals (CIs) were calculated using the “exact” method within a random-effects model [30, 31]. Between-study heterogeneity was assessed using the χ² test and the I² measure of inconsistency [32]. An I² estimate >50% was considered indicative of substantial heterogeneity. Publication bias was evaluated by Egger’s regression asymmetry test and visually examined by a funnel plot [33]. All reported p values are two-sided with significance being set as <0.05. A sensitivity analysis was carried out to test the influence of sex (male vs. female), age (mean age <65 years vs. ≥65 years), TSH reference values for determining hypothyroidism (<4.5 mIU/L vs. ≥4.5 mIU/L), the sample size of the studies (number of patients <1,000 vs. 1,000–10,000 vs. >10,000), and geographic region (Northern, Southern, Western, and Eastern Europe) on the estimates of prevalence. All statistical analyses were performed in STATA version 13 (StataCorp, 2013).

Figure 1 presents the search strategy flowchart. The search returned 15,565 citations. After excluding 4,526 duplicates and 11,003 studies based on revision of titles and abstracts, 36 full-text articles were assessed for eligibility. Of those, 20 articles were selected. One article reported results for two populations [15].

Table 1 presents the main characteristics of the studies selected for systematic review. The sample size of selected studies varied between 307 [34] and 80,490 [35]. The studies included sample populations from Austria [35], Belgium [36], Bulgaria [37], Denmark [38], Germany [39, 40], Italy [16, 18, 41], Norway [15, 42], Spain [17, 19, 34, 43], The Netherlands [44, 45], the UK [46, 47], and Turkey [48].

The report of selected articles results from cross-sectional analysis of 8 population-based surveys [15-18, 40-42, 49], 7 prospective cohort studies [34, 38, 39, 44-47], 2 retrospective cohort studies [35, 36], and 3 case-control studies [37, 43, 48].

The conceptual definition of subclinical hypothyroidism (elevated TSH and normal free thyroid hormone) and overt hypothyroidism (elevated TSH and low free thyroid hormone) was the same across the selected studies, but the reference ranges for TSH and free thyroid hormones (i.e., fT₄ and/or fT₃) were not. The highest limit of the reference range for TSH varied between 3.4 mIU/L [40] and 5.5 mIU/L [48]. The lowest limits of the reference ranges varied between 8.0 pmol/L [42] and 12.8 pmol/L [40] for fT₄ and between 2.96 pmol/L [48] and 3.92 pmol/L [40] for fT₃.

The studies’ methodological quality scores are available in online supplementary Tables 4–6. All cohort studies (n = 9), 2 case-control studies, and 3 cross-sectional studies were considered to have good methodological quality; 1 case-control study and 5 cross-sectional studies were considered to have moderate quality.

The prevalence of undiagnosed hypothyroidism in European countries is presented in Table 2.

With the exception of one study [15], all studies reported the prevalence of undiagnosed subclinical hypothyroidism, which varied between 0.5% in Germany [39, 50] and 12.5% in Italy [16].

The prevalence of undiagnosed overt hypothyroidism was reported in 11 studies [15-19, 40-43, 46, 48] and varied between 0.1% in Norway [15] and 3.2% in Germany [46].

The prevalence of undiagnosed total (subclinical plus overt) hypothyroidism was ascertained in 11 studies [15-19, 40-43, 46, 48], with estimates ranging between 2.7% [41] and 12.8% [16]. The prevalence of hypothyroidism was lower in men than in women for subclinical (men: min. 0.9%, max. 8.5%; women: min. 2.8%, max. 16.9%), overt (men: min. 0.0%, max. 0.3%; women: min. 0.5%, max. 0.9%), and total hypothyroidism (men: min. 1.2%, max. 8.5%; women: min. 3.8%, max. 17.5%).

The estimated prevalence of undiagnosed subclinical hypothyroidism was 4.11% (95% CI 3.05–5.31%, p < 0.001, I² = 99.32%) (Fig. 2). The prevalence of undiagnosed overt hypothyroidism was 0.65% (95% CI 0.38–0.99%, p < 0.001, I² = 96.67%) (Fig. 3). Undiagnosed total hypothyroidism was estimated with a prevalence of 4.70% (95% CI 2.98–6.79%, p < 0.001, I² = 99.53%) (Fig. 4).

The funnel plots (for studies studying subclinical, overt, and total hypothyroidism) appear asymmetric, with smaller studies tending to report larger proportions of hypothyroidism, which may suggest publication bias (online suppl. Fig. 1–3). Egger’s regression asymmetry test detected publication bias among the studies included to assess the proportion of overt hypothyroidism, but not among the subclinical or total hypothyroidism analysis (subclinical hypothyroidism: coefficient 6.06, standard error 3.56, p = 0.105; overt hypothyroidism: coefficient 4.00, standard error 1.46, p = 0.021; total hypothyroidism: coefficient 9.43, standard error 4.45, p = 0.06). The test provides weak evidence for the presence of small-study effects. However, there is evidence of substantial heterogeneity among the results of the three proportion estimates, which makes the publication bias analysis difficult. Moreover, there are studies reporting a reduced number of events, particularly in the estimation of overt hypothyroidism, which can lead to false-positive publication bias test results.

Table 3 presents the results of the sensitivity analysis. In general, the prevalence of hypothyroidism tends to be higher in female patients, those aged ≥65 years, among studies with lower sample sizes, in case of TSH reference levels <4.5 mIU/L, and in Eastern and Southern Europe.

This systematic review and meta-analysis aimed to identify the prevalence of undiagnosed hypothyroidism (subclinical, overt, and total) in the general population of Europe. As we were interested in studying current figures, the time horizon of the search strategy was limited to articles published over the last 10 years. Several studies were dedicated to characterizing the epidemiology of thyroid dysfunctions, but only few addressed the problem of underdiagnosing such diseases [26]. For example, some studies applied drug utilization approaches or electronic health record analysis based on prescriptions data, pharmacy claims, or surveys to identify consumption of antithyroid medication and estimate the prevalence of treated hypothyroidism [21, 51-54]. Furthermore, many studies investigated the prevalence of this disorder or its role as a risk factor for other diseases in specific populations, for example in pregnant women, patients with diabetes, patients with cardiovascular diseases, or immunocompromised individuals [12, 24, 55-59]. Therefore, there is room and need for systematically review of the evidence on the prevalence of undiagnosed hypothyroidism in the general population. As such, patients with known thyroid dysfunction and/or using antithyroid medication at baseline were not considered in this review, as defined in the inclusion criteria.

The present results point out that the current prevalence of undiagnosed subclinical hypothyroidism (4.11%) is higher than the prevalence of undiagnosed overt hypothyroidism (0.65%). This is an unsurprising finding taking into account the conclusions of other studies. A previous meta-analysis on thyroid dysfunctions estimated the mean prevalence of undiagnosed hypothyroidism in Europe at 4.94%, with a clear predominance of the subclinical form of the disease: subclinical hypothyroidism 4.59%, overt hypothyroidism 0.62% [26]. The higher prevalence of subclinical hypothyroidism may be explained by the fact that many patients are asymptomatic or report fewer and milder symptoms than patients with overt hypothyroidism [2]. Furthermore, 75% of patients with subclinical hypothyroidism have a serum TSH level <10 mIU/L (i.e., mild subclinical hypothyroidism) [60], being less prone to hypothyroid symptoms and, for example, cardiovascular events than patients with severe subclinical hypothyroidism [2, 22-24]. Moreover, the TSH concentration usually normalizes within 2 years for 46% of patients with subclinical hypothyroidism if the TSH level is <7 mIU/L in a single measurement [61-63]. Yet, the risk of progression to overt hypothyroidism among patients with severe forms of subclinical hypothyroidism is estimated at 2–6% per year [51, 61, 62].

The results of this meta-analysis should be interpreted with caution given its high heterogeneity (I² > 96%). There are considerable differences between the selected studies, including study design (e.g., prospective and retrospective cohort studies, population-based surveys), studied population (e.g., inclusion/exclusion criteria, sex, age), sample size (i.e., range from hundreds to tens of thousands), laboratory tests performed (e.g., TSH, fT₄, fT₃, antibodies), laboratory techniques and material, and reference ranges of circulating thyroid hormones used to diagnose the several forms of hypothyroidism. Despite the limitations of the present meta-analysis, a quantitative synthesis that matches the research question is possibly preferred over qualitative interpretations of results or unclear quasi-quantitative analyses [64]. In addition to the production of overall estimates, meta-analysis provides the advantage of assessing the consistency of findings and improving the understanding of moderator variables, boundary conditions, and generalizability [64-66].

The prevalence of hypothyroidism (both subclinical and overt) was higher in women than in men, which is also in line with previous findings [26]. Monitoring the thyroid status is particularly important in the group of pregnant women or women of childbearing potential since correction of overt hypothyroidism reduces the risk of fetal loss and preterm birth [67, 68]. Treatment of overt hypothyroidism is therefore recommended during pregnancy [69]. Pregnant women with subclinical hypothyroidism before 20 weeks of gestation are at a higher risk of miscarriage [70]. Treatment may reduce miscarriage in thyroid autoantibodies (TPOAb)-positive pregnant women, and therefore women with TSH concentrations >2.5 mIU/L should be evaluated for TPOAb status. Furthermore, TPOAb-positive women with TSH levels greater than the pregnancy-specific ranges as well as women with TPOAb-negative status and TSH levels >10 mIU/L should be treated with levothyroxine. Other subgroups of pregnant women may be considered for treatment depending on TSH levels [69]. Nevertheless, universal thyroid screening in pregnancy is still a matter of debate [71]. The results of randomized controlled trials pointed out that treating pregnant women with subclinical hypothyroidism with levothyroxine provided no benefit on the IQ of the offspring or obstetric outcomes [72, 73]. However, levothyroxine therapy was initiated after the critical phase of fetal brain development, i.e., from the end of the first trimester of pregnancy.

The results of our sensitivity analysis also indicated that the prevalence of any form of hypothyroidism is higher among patients aged ≥65 years (subclinical 5.12%, overt 1.10%) as compared to younger ones (subclinical 3.58%, overt 0.54%). The study by Lucas et al. [17] also illustrates the influence of age on estimates of prevalence: when the authors restricted the analysis to subjects >60 years, the prevalence of subclinical and overt hypothyroidism was estimated at 6.2 and 0.44%, respectively (vs. 3.5 and 0.2% in the general population).

With regards to the reference ranges of serum thyroid hormones, the variation of the highest limit of the reference range for TSH concentration in the studies included in this meta-analysis is considerable (between 3.4 and 5.5 mIU/L) [40, 48]. According to the sensitivity analysis, a higher prevalence of subclinical and total hypothyroidism was found for studies using lower TSH reference values. The most commonly cited reference range for TSH concentration in the clinical literature set the highest limit at 4.0 mIU/L (and the lowest limit at 0.4 mIU/L), while the reference range for fT₄ depends on the type of assay and the population in question [1]. Nevertheless, the reference level cutoffs used to determine overt and subclinical hypothyroidism have changed over the years [46]. Thus, in the context of the present meta-analysis, the lack of consensus on reference ranges for levels of thyroid hormones used to establish diagnoses may result in inaccurate comparisons because of nonoverlapping definitions of disease across the included studies.

Moreover, the sensitivity analysis also detected differences in the prevalence of hypothyroidism depending on the sample sizes of included studies – higher estimates of prevalence among those with lower number of patients – and on the European region – the prevalence of hypothyroidism tended to be higher in Eastern and Southern Europe.

There is great variation in the clinical manifestations of hypothyroidism, ranging from the absence of signs or symptoms to life-threatening conditions, such as myxedema coma [1]. The most common symptoms are nonspecific and vary depending on different individual factors, such as the age or sex of patients. Furthermore, autoimmune hypothyroidism is asymptomatic or associated with only one symptom in nearly 15% of patients with the condition, while 70% of euthyroid individuals have at least one symptom that is usually related with hypothyroidism [74]. These issues may contribute to preclude or at least delay the diagnosis of hypothyroidism. In addition, with the exception of some populations, such as patients with type 1 diabetes mellitus [75, 76], there is no consensus on TSH screening in the general population [1].

The long-term consequences of hypothyroidism are clinically important and adversely affect the health of patients. Those consequences have been mainly studied in patients with subclinical hypothyroidism, given that overt hypothyroidism is usually treated. The body of evidence is particularly robust with regards to the association between hypothyroidism and adverse cardiovascular outcomes. A meta-analysis of individual data on >2,500 elderly participants found that the risk of coronary heart disease and mortality is increased in subjects with higher serum TSH levels [22]. Conversely, patients with TSH concentrations >10 mIU/L are at higher risk of heart failure [22, 23]. Hence, timely diagnosis and treatment initiation are potentially important for preventing adverse consequences of hypothyroidism in patients with the condition. However, the results of a double-blind, randomized, placebo-controlled, parallel-group trial pointed out that giving levothyroxine to senior patients (>65 years) with subclinical hypothyroidism (TSH level 4.60–19.99 mIU/L, fT₄ level within the reference range) provided no significant beneficial effects on thyroid-related quality of life symptoms or cardiovascular events [77]. Other illnesses potentially associated with hypothyroidism include nonalcoholic fatty liver disease, cancer mortality, arthritis, renal failure, and diabetes [25, 78-80].

In conclusion, the current evidence suggests that a considerable proportion of the European population has hypothyroidism, particularly subclinical hypothyroidism, which is not diagnosed. This issue deserves further investigation because of possible deleterious consequences for public health.

The authors have no ethical conflicts to disclose.

This work was financially supported by Merck S.A., Portugal.