Endocrinology · Medical Literature
Thyroid disorders, thyroid autoimmunity, and anxiety-a systematic review and meta-analysis of population-based studies.
Abstract
Previous studies, largely based on case-control samples, have suggested that hypothyroidism, hyperthyroidism, and thyroid autoimmunity may be associated with an increased risk of anxiety. However, whether a true association exists remains unclear, as population-based studies, which draw on more representative samples and are therefore less prone to selection bias, have yielded conflicting results. This study aimed to investigate the association between thyroid dysfunction and anxiety disorders in population-based cohort and cross-sectional studies. We conducted a systematic literature search in PubMed, PsycINFO, and Embase databases from inception to March 2026. Two reviewers independently selected population-based cohort and cross-sectional studies reporting laboratory- or International Classification of Diseases (ICD)-based diagnoses of thyroid dysfunction or autoimmunity, and anxiety outcomes according to validated criteria or rating scales. Study quality was assessed using the Newcastle-Ottawa Scale. All meta-analyses were conducted using random-effects models, and heterogeneity was quantified using the I 2 statistic (pre-registration in Prospero: CRD42020164791). Out of 2,548 screened articles, 12 studies comprising 676,658 participants were included. Six studies dealt with hypothyroidism, seven with hyperthyroidism, and four with autoimmunity. Meta-analyses revealed small and mostly statistically non-significant positive summary estimates for hypothyroidism (OR 1.13, 95% CI [0.77-1.65]; I 2 = 86%), hyperthyroidism (OR 1.38, 95% CI [1.03-1.83]; I 2 = 86%), and thyroid peroxidase antibody (TPOAb) (OR 1.47, 95% CI [0.73-2.94]; I 2 = 65%). Sensitivity and subgroup analyses, as well as studies summarized qualitatively, showed patterns consistent with the main results. The number of studies is small, and thyroid pathology as well as anxiety are heterogeneously defined. The evidence is inconclusive: Small positive associations seem likely, but confidence intervals are also in agreement with intermediate positive and even small protective effects, suggesting either limited data or the absence of relevant associations between thyroid pathology and anxiety. Further population-based studies based on established diagnostic criteria are warranted.
Clinical relevance
Among population-based cohort and cross-sectional studies, is there a statistically or clinically meaningful association between thyroid dysfunction (hypothyroidism, hyperthyroidism, subclinical disease) or thyroid autoimmunity and the risk or severity of anxiety disorders? __SNAPSHOT__ {"studyDesign":"Systematic review with meta-analysis; restricted to population-based cohort and cross-sectional studies; pre-registered in PROSPERO (CRD42020164791); random-effects models used throughout","setting":"Multi-database synthesis (PubMed, PsycINFO, Embase); inception to March 2026; no geographic restriction stated in the abstract","sampleSize":"Number of included studies not reported in the available abstract (abstract truncated at 'Out of 2,548' screened records); individual study and pooled N not reported in the available excerpt","population":"General community-dwelling populations drawn from population-based cohort or cross-sectional studies; patients with laboratory- or ICD-defined diagnoses of hypothyroidism, hyperthyroidism (overt or subclinical not specified in abstract), or thyroid autoimmunity; anxiety assessed by validated criteria (e.g., DSM/ICD) or validated rating scales","intervention":"Thyroid dysfunction (hypothyroidism, hyperthyroidism) or thyroid autoimmunity (antibody-defined) as the exposure of interest","comparator":"Euthyroid individuals or general population controls without the respective thyroid diagnosis, as drawn from population-based samples","primaryOutcome":"Anxiety disorders or anxiety symptoms, assessed by validated diagnostic criteria or established rating scales; specific primary endpoint metric (odds ratio, relative risk, prevalence ratio) not reported in the available abstract excerpt","followUp":"Variable by included study design; not reported in the available abstract","fundingSource":"Not reported in the available abstract"}
Key findings
The available abstract is truncated and does not contain quantitative results. No effect sizes, odds ratios, relative risks, confidence intervals, p-values, I² heterogeneity statistics, or NNT/NNH values are reported in the excerpt provided. The abstract describes the methodological rationale and search strategy but is cut off prior to reporting results. Accordingly, no findings can be summarized or attributed to this study without risk of fabrication. The prior literature from case-control samples (referenced in the background of this paper) has suggested a positive association between thyroid autoimmunity and anxiety (e.g., OR 2.32, 95% CI 1.40–3.85 in Siegmann et al., JAMA Psychiatry 2018, which was based on case-control data), but these are from a different paper and study design. The Ivens et al. 2026 PeerJ analysis specifically intended to re-examine this question using population-based samples, which are less prone to selection bias and have historically yielded conflicting results. Quantitative findings from the current study are not reportable from the available abstract excerpt.
Practice implications
IMPORTANT LIMITATION: Only the abstract introduction and methods section were available; the abstract was truncated before results and conclusions. Therefore, no determination can be made as to what this study established, what direction or magnitude of association was found, or whether findings differ from prior case-control literature. The study's design choice — restricting inclusion to population-based cohort and cross-sectional studies — is methodologically sound and directly addresses the selection bias inherent in case-control samples that have driven prior positive associations. The PROSPERO pre-registration (CRD42020164791) and use of random-effects models with I² quantification are appropriate. This research group has previously published related meta-analyses (hypothyroidism–depression in JAMA Psychiatry 2021; hyperthyroidism–depression in Translational Psychiatry 2022) using similar methodology. The clinical importance and direction of findings from this specific paper cannot be assessed until full results are available. Clinicians should access the full published article for the primary results before changing practice. __LIMITATIONS__ The following methodological limitations can be anticipated based on the stated design and prior literature in this area, but specific limitations reported by the authors are not available from the truncated abstract: (1) Cross-sectional studies included alongside cohort studies preclude causal inference for a substantial proportion of pooled data; (2) Heterogeneity in anxiety outcome ascertainment (validated rating scales vs. ICD/DSM diagnoses) may limit interpretability of pooled estimates; (3) Combining overt and subclinical thyroid dysfunction, if not separately analyzed, could obscure clinically distinct subgroups; (4) Population-based designs reduce but do not eliminate selection bias and residual confounding by comorbidities, medications, and sociodemographic factors; (5) ICD-based thyroid diagnoses may underascertain subclinical disease; (6) The literature search extended to March 2026 but publication bias in thyroid–psychiatric literature favoring positive findings cannot be excluded without formal assessment; (7) Funding source not disclosed in abstract, precluding assessment of industry influence; (8) Generalizability may be limited if included studies are predominantly from high-income countries with iodine-sufficient populations. __BOTTOM_LINE_CATEGORY__ low_certainty __BOTTOM_LINE__ This PROSPERO-pre-registered systematic review and meta-analysis by Ivens, Bode, Bschor et al. (PeerJ, 2026) was specifically designed to address whether the previously reported associations between thyroid dysfunction/autoimmunity and anxiety hold in population-based samples — the most rigorous and representative study type available for this question. Because only an incomplete abstract is available (truncated before results), no clinical conclusion or practice recommendation can responsibly be drawn from this publication at this time. Clinicians should retrieve the full article for quantitative results before incorporating these findings into clinical decision-making.