Research

JAMA Psychiatry | Original Investigation

Association of Hypothyroidism and Clinical Depression

A Systematic Review and Meta-analysis
Henry Bode; Beatrice Ivens; Tom Bschor, MD; Guido Schwarzer, PhD; Jonathan Henssler, MD; Christopher Baethge, MD

Supplemental content
IMPORTANCE Hypothyroidism is considered a cause of or a strong risk factor for depression,
but recent studies provide conflicting evidence regarding the existence and the extent of the
association. It is also unclear whether the link is largely due to subsyndromal depression or
holds true for clinical depression.

OBJECTIVE To estimate the association of hypothyroidism and clinical depression in the

general population.

DATA SOURCES PubMed, PsycINFO, and Embase databases were searched from inception
until May 2020 for studies on the association of hypothyroidism and clinical depression.

STUDY SELECTION Two reviewers independently selected epidemiologic and

population-based studies that provided laboratory or International Statistical Classification of
Diseases and Related Health Problems diagnoses of hypothyroidism and diagnoses of
depression according to operationalized criteria (eg, Diagnostic and Statistical Manual of
Mental Disorders or International Statistical Classification of Diseases and Related Health
Problems) or cutoffs in established rating scales.

DATA EXTRACTION AND SYNTHESIS Two reviewers independently extracted data and
evaluated studies based on the Newcastle-Ottawa Scale. Summary odds ratios (OR) were
calculated in random-effects meta-analyses.

MAIN OUTCOMES AND MEASURES Prespecified coprimary outcomes were the association of
clinical depression with either hypothyroidism or autoimmunity.

RESULTS Of 4350 articles screened, 25 studies were selected for meta-analysis, including
348 014 participants. Hypothyroidism and clinical depression were associated (OR, 1.30
[95% CI, 1.08-1.57]), while the OR for autoimmunity was inconclusive (1.24 [95% CI,
0.89-1.74]). Subgroup analyses revealed a stronger association with overt than with
subclinical hypothyroidism, with ORs of 1.77 (95% CI, 1.13-2.77) and 1.13 (95% CI, 1.01-1.28),
respectively. Sensitivity analyses resulted in more conservative estimates. In a post hoc
analysis, the association was confirmed in female individuals (OR, 1.48 [95% CI, 1.18-1.85]) but
not in male individuals (OR, 0.71 [95% CI, 0.40-1.25]).

CONCLUSIONS AND RELEVANCE In this systematic review and meta-analysis, the effect size for Author Affiliations: Department of
the association between hypothyroidism and clinical depression was considerably lower than Psychiatry and Psychotherapy,
previously assumed, and the modest association was possibly restricted to overt Faculty of Medicine, University of
Cologne, Cologne, Germany (Bode,
hypothyroidism and female individuals. Autoimmunity alone may not be the driving factor in Ivens, Henssler, Baethge);
this comorbidity. Department of Psychiatry and
Psychotherapy, Faculty of Medicine,
Technical University of Dresden,
Dresden, Germany (Bschor); Institute
of Medical Biometry and Statistics,
Faculty of Medicine and Medical
Center, University of Freiburg,
Freiburg, Germany (Schwarzer);
Department of Psychiatry and
Psychotherapy, Charité
Universitätsmedizin Berlin, Berlin,
Germany (Henssler).
Corresponding Author: Christopher
Baethge, MD, Department of
Psychiatry and Psychotherapy,
Faculty of Medicine, University of
Cologne, Kerpener St 62, 50931
JAMA Psychiatry. 2021;78(12):1375-1383. doi:10.1001/jamapsychiatry.2021.2506 Cologne, Germany (cbaethge@
Published online September 15, 2021. uni-koeln.de).

(Reprinted) 1375
© 2021 American Medical Association. All rights reserved.
Downloaded from jamanetwork.com by Bryan González on 08/12/2025
 Research Original Investigation Hypothyroidism and Clinical Depression

T
he symptoms of hypothyroidism and depression partly
overlap, but for decades, a more specific link between Key Points
both disorders has been discussed.1 Neurobiological
Question Is there an association of hypothyroidism and thyroid
research has uncovered some mechanisms of thyroid hor- autoimmunity with depression?
mones in the brain, providing possible explanations for an in-
Findings In this systematic review and meta-analysis of 25 studies
teraction with mood.2,3 Also, immunologic processes may pro-
including 348 014 participants, there was a moderate association
vide a link between autoimmune thyroiditis and depression.4,5
of overt, and less so of subclinical, hypothyroidism with clinical
A 2018 meta-analysis6 reported a substantial association depression; this association is stronger in female than in male
of subclinical and clinical depression with hypothyroid auto- individuals. A statistically significant association of verified thyroid
immunity. With an odds ratio (OR) of 3.31, Siegmann et al7 es- peroxidase antibodies positivity and clinical depression was not
timated that each year, more than 20% of patients with auto- found.
immune thyroiditis experience depression. This meta- Meaning A strong connection between hypothyroidism and
analysis has been criticized, for example, for its combination depression was not evident in this analysis; however, a possible
of population-based studies with results from outpatient clin- dose-effect relationship, especially in female individuals, should be
ics, with their bias toward more severely affected patients.8,9 investigated further.
Since the authors associated thyroid status with any change
of depression scores, including and especially changes below
cutoffs for clinically relevant depression, the practical signifi- servants) not suggestive of bias were eligible. Thyroid disor-
cance of the results is uncertain. In contrast, another ders leading to or representing hypothyroidism, either sub-
meta-analysis10 reported only a weak, nonsignificant associa- clinical or overt, autoimmune disorders (eg, Hashimoto thy-
tion of hypothyroidism and depression (OR, 1.24). However, roiditis), as diagnosed by established laboratory methods or
while this study is an individual patient data meta-analysis, drawn from registers including hospital data if reliability of di-
it was based on only 6 studies and was restricted to subclini- agnoses was documented. Laboratory criteria had to be speci-
cal hypothyroidism. fied by the authors. For overt hypothyroidism, criteria needed
As a result, the existence and the extent of an association to consist of at least 1 elevated thyrotropin and 1 lowered free
between hypothyroidism and clinical depression remains thyroxine measurement. Subclinical hypothyroidism is de-
unclear. In addition, if there were such an association, it is fined by increased thyrotropin, without evidence of lowered
unknown whether hypothyroidism or autoimmunity is the free thyroxine. Thyroid peroxidase (TPO) antibody positivity
driving force. Therefore, we conducted a systematic review and had to be assessed by at least 1 measurement of TPO antibod-
meta-analysis of studies presenting data on hypothyroidism ies above a threshold prespecified by the authors. Outcomes
(subclinical or overt) and clinical depression. To reduce selec- included clinically significant depression, either defined as a
tion bias, we restricted the meta-analysis to epidemiologic and major depressive disorder diagnosis according to established
population-based studies. diagnostic systems, eg, International Statistical Classification
of Diseases and Related Health Problems, Tenth Revision (ICD-
10) or Diagnostic and Statistical Manual of Mental Disorders
(Fourth Edition) (DSM-IV), or an above-threshold score in es-
Methods tablished psychopathology rating scales for depression,12 with
This is a systematic review and meta-analysis registered in thresholds prespecified by the authors. Diagnoses could origi-
PROSPERO (CRD42020164791). Its reporting is based on the nate with assessment rating scales, standardized interviews
2020 Preferred Reporting Items for Systematic Reviews and (eg, World Health Organization Composite International Di-
Meta-analyses (PRISMA) reporting guideline.11 agnostic Interview), or from registers including hospital data
if reliability of diagnoses was documented.
Literature Search and Data Extraction Case-control studies were excluded. Two authors (H.B. and
We conducted a systematic search in MEDLINE and PubMed B.I.) independently screened titles and abstracts retrieved in
Central via PubMed, in PsycINFO via EBSCOhost, and in the literature search. We did not exclude gray literature and
Embase to identify epidemiologic and population-based stud- applied no language or date restrictions. Bibliographies of all
ies on the association of hypothyroidism with the occurrence articles eventually included were hand searched. Two raters
of depression from inception to May 4, 2020. We combined (H.B. and B.I.) independently read full texts of all articles po-
generic terms for depression, hypothyroidism, and population- tentially eligible. Data from included studies were extracted
based study settings. Search terms and history are specified independently by 2 authors (H.B. and B.I.) using an Excel-
in the eMethods in the Supplement. based standardized data extraction form (Microsoft) in accor-
dance with the Cochrane Collaboration Handbook. All dis-
Selection Criteria agreements were solved by consensus or discussion with the
Cohort and cross-sectional studies were included. The study senior author (C.B.).
population was representative of the general population. Stud- All studies included were rated independently by 2 au-
ies were population based and not primarily conducted with thors (H.B. and B.I.) using the Newcastle-Ottawa Scale for as-
patients with thyroid or mood disorders in a medical setting. sessing risk of bias, using the adaptations for cohort13 and cross-
Studies conducted in broad and diverse populations (eg, civil sectional studies.14 Studies were rated as carrying an overall

1376 JAMA Psychiatry December 2021 Volume 78, Number 12 (Reprinted) jamapsychiatry.com

© 2021 American Medical Association. All rights reserved.

Downloaded from jamanetwork.com by Bryan González on 08/12/2025
 Hypothyroidism and Clinical Depression Original Investigation Research

Figure 1. PRISMA Flow Diagram

Identification of studies via databases and registers Identification of studies via other methods

4350 Records identified 978 Records removed before screening 5 Records identified
1831 PubMed 978 Duplicate 0 Websites
1876 Embase 0 Marked as ineligible by 0 Organizations
643 PsycINFO automation tools 0 Citation searching
0 Registers 0 Other reasons

3372 Records screened 3314 Records excluded

58 Reports sought for retrieval 1 Report not retrieved 5 Reports sought for retrieval 0 Reports not retrieved

57 Reports assessed for eligibility 34 Reports excluded 5 Reports assessed for eligibility 3 Reports excluded
12 No or insufficient data on depression 1 Not population based
as a dichotomous outcome 2 No or insufficient data
14 No or insufficient data on thyroid on thyroid disease
disease
8 Not population based

25 Studies included in review

and meta-analysis

low risk of bias when falling into the highest Newcastle- ently. We also stratified our analyses by age, comparing stud-
Ottawa Scale category, ie, receiving all or all but 1 star in the ies on older populations (minimum age ≥60 years) with stud-
rating system. ies on individuals of all ages.

Data Analysis Data Synthesis

Primary Outcome and Analysis Owing to differences in study design and settings, we used ran-
The primary analysis consists of a 2-part investigation: we mea- dom-effects analyses (DerSimonian & Laird), which assume
sured the association of depression with (1) hypothyroidism that effects vary according to the specifics of a study rather than
and (2) thyroid autoimmunity, expressed as ORs with 95% CIs. 1 true effect underlying all studies.15 For primary outcomes,
We compared the occurrence of clinical depression in people we also report prediction intervals to account for the hetero-
with vs without hypothyroidism/autoimmunity. If studies re- geneity between studies.16 Statistical heterogeneity, which de-
ported the observed effect as either a risk ratio or hazard ra- scribes the variation in results between studies, is reported as
tio, we transformed these effect sizes into ORs. If studies re- I2 statistic and as tau, the standard deviation of the effect es-
ported adjusted effect sizes, we included those with the least timate. We assessed publication bias in funnel plots and
comprehensive adjustment to be as coherent as possible with Egger test 17 and estimated the role of missing studies in
unadjusted effect sizes. The first part of the primary analysis trim-and-fill-analyses. 18 A leave-1-out analysis was con-
included all studies reporting results for overt or subclinical ducted if forest plots indicated a disproportionate influence
hypothyroidism. The other part of the analysis included all of single studies in calculating summary ORs. We used the
studies reporting associations of TPO antibody positivity. software package Comprehensive Meta-Analysis version 3
Calculations and formulae are listed in the eMethods in the (Biostat) as well as R software (R Foundation), including the
Supplement. R packages meta19 and metafor.20 Two-sided P values were
significant at .05.
Subgroup Analyses
We subdivided hypothyroidism into subclinical and overt hy-
pothyroidism, as defined above, and stratified our primary
analyses by sex, risk of bias, intake of thyroid medication, a
Results
core group of strictly population-based studies, and assess- After screening 4350 articles and excluding duplicates, we re-
ment of depression. viewed 62 in full text. Of those, 25 articles21-45 were included
in this study (Figure 1). Nine studies provided data on indi-
Post hoc Analyses viduals with overt hypothyroidism, 17 on individuals with sub-
Hypothyroidism and depression affect more female than male clinical hypothyroidism, and 9 on individuals with thyroid
individuals, but it has not been established so far whether de- autoimmunity.
pression also occurs more often in female individuals among Table 1 displays characteristics of all studies: 6 cohort
patients diagnosed with hypothyroidism. Therefore, we fur- (236 646 of 348 014 participants [68%]) and 19 cross-
ther explored sex-specific results found in subgroup analyses sectional studies (111 368 [32%]), with 348 014 participants,
by analyzing studies reporting results on both sexes differ- ranging from 10039 to 131 041.44 The study size weighted mean

jamapsychiatry.com (Reprinted) JAMA Psychiatry December 2021 Volume 78, Number 12 1377

© 2021 American Medical Association. All rights reserved.

Downloaded from jamanetwork.com by Bryan González on 08/12/2025
 Research Original Investigation Hypothyroidism and Clinical Depression

Table 1. Characteristics of Studies Included

No. of NOS Age range, Assessment of Assessment of
Source Study type patients score Thyroid disorder Sex y thyroid disorder depression
Almeida et al,21 2011 Cohort 3901 9a Subclinical Male 69-87 FT4, TSH Clinical diagnosisb
22
Benseñor et al, 2016 Cross-sectional 13 221 7 Subclinical Mixed 35-74 FT4, TSH Clinical diagnosisb
23
Bould et al, 2012 Cross-sectional 325 6 Subclinical Mixed 17-74 FT4, TSH PHQ-9
Carta et al,24 2004 Cross-sectional 222 7 Autoimmunity Mixed, female, ≥18 TPO-abs Clinical diagnosisb
male
de Jongh et al,25 2011 Cross-sectional 1185 7 Subclinical Mixed ≥65 FT4, TSH CES-D
Delitala et al,26 2016 Cross-sectional 3138 8 Autoimmunity Mixed NA TPO-abs CES-D
Engum et al,27 2002 Cross-sectional 30 062 9a Overt, Mixed 40-89 FT4, TSH HADS-D
subclinical
Engum et al,28 2005 Cross-sectional 30 175 9a Autoimmunity Mixed 40-84 TPO-abs HADS-D
29
Fjaellegaard et al, Cross-sectional 8214 7 Autoimmunity Mixed, female, ≥20 FT4, TSH, Clinical diagnosisb
2015 male TPO-abs
Subclinical Mixed
Guimarães et al,30 Cross-sectional 1249 9a Overt, Female 35-91 FT4, TSH PRIME-MD
2009 subclinical
Hong et al,31 2018 Cross-sectional 1717 8 Subclinical Mixed 19-76 FT4, TSH PHQ-9
Ittermann et al,32 2015 Cohort 1895 8a Overt, Mixed 20-79 TSH, TPO-abs Clinical diagnosisb
autoimmunity
Kim et al,33 2010 Cross-sectional 481 8 Subclinical Mixed ≥65 TSH GMS-B3
Kim et al,34 2018 Cohort 92 206 8a Subclinical Mixed NA FT4, TSH CES-D
Kvetny et al,35 2015 Cross-sectional 14 502 8 Subclinical Mixed, female, ≥20 TSH Clinical diagnosisb
male
Lee et al,36 2019 Cross-sectional 1651 8 Autoimmunity Mixed, female, ≥20 TPO-abs PHQ-9
male
Lin et al,37 2016 Cohort 6100 9a Overt Mixed, female, ≥20 Register Clinical diagnosisb
male
Manciet et al,38 1995 Cross-sectional 407 7 Overt, Mixed ≥65 FT4, TSH CES-D
subclinical
Maugeri et al,39 1998 Cross-sectional 100 6 Overt, Mixed ≥70 T4, TSH GDS-30
subclinical
Medici et al,40 2014 Cohort 1503 9a Autoimmunity Mixed ≥55 TPO-abs Clinical diagnosisb
41
Park et al, 2010 Cross-sectional 918 7 Subclinical Mixed, female, ≥65 FT4, TSH Clinical diagnosisb
male
Pop et al,42 1998 Cross-sectional 583 8 Overt, Female 47-54 FT4, TSH, EDS
subclinical, TPO-abs
autoimmunity
Shinkov et al,43 2014 Cross-sectional 2312 7 Subclinical Mixed, female, 20-84 TSH Zung SDS
male
Thomsen et al,44 2005 Cohort 131 041 8a Overt Mixed ≥15 Register Clinical diagnosisb
van de Ven et al,45 Cross-sectional 906 8 Overt, Mixed 50-70 FT4, TSH BDI-Ia
2012 subclinical,
autoimmunity
Abbreviations: BDI-Ia, Beck Depression Inventory Ia; CES-D, Center for Self-Rating Depression Scale.
Epidemiologic Studies Depression Scale; EDS, Edinburgh Depression Scale; a
Included in the risk of bias analysis.
FT4, free thyroxine; GDS-30, Geriatric Depression Scale 30; GMS-B3, Geriatric b
Diagnostic and Statistical Manual of Mental Disorders– or International
Mental State Diagnostic Schedule; HADS-D, Hospital Anxiety and Depression
Statistical Classification of Diseases and Related Health Problems–conforming
Scale; NA, not applicable; NOS, Newcastle-Ottawa Scale; PHQ-9, Patient Health
diagnosis of depression.
Questionnaire 9; PRIME-MD, Primary Care Evaluation of Mental Disorders;
TPO-abs, thyroid peroxidase antibodies; TSH, thyrotropin; Zung SDS, Zung

age of participants was 44.9 years. The overall proportion of Primary Analysis
female individuals was 53.6%. The analysis resulted in an OR of 1.30 (95% CI, 1.08-1.57) for all
Fifteen studies assessed depressive symptoms using a types of hypothyroidism (Figure 2). Separated in subclinical and
score. Ten studies reported DSM- and/or ICD-conforming di- overt hypothyroidism ORs were 1.13 (95% CI, 1.01-1.28) and 1.77
agnoses of major depressive disorder. Twenty-three studies re- (95% CI, 1.13-2.77), respectively (Table 2). In combined analysis,
ported diagnoses of thyroid disorders based on established there was a difference between female individuals (OR, 1.62 [95%
laboratory methods, and 2 used register data on ICD-based di- CI, 1.20-2.19]) and male individuals (OR, 0.69 [95% CI, 0.44-1.11]).
agnoses, which included laboratory assessments as well. In- Strictly population-based studies yielded a moderately stronger
dividuals taking thyroid medication were included in 12 stud- association. Cohort design, inclusion of individuals taking thy-
ies. Six cohort and 3 cross-sectional studies had a low risk of roid medication, and a DSM- or ICD-conforming diagnosis of de-
bias. Additional data on the included studies can be obtained pression also resulted in moderately higher associations (eTable 2
from eTable 1 in the Supplement. in the Supplement).

1378 JAMA Psychiatry December 2021 Volume 78, Number 12 (Reprinted) jamapsychiatry.com

© 2021 American Medical Association. All rights reserved.

Downloaded from jamanetwork.com by Bryan González on 08/12/2025
 Hypothyroidism and Clinical Depression Original Investigation Research

Figure 2. Association of Hypothyroidism and Depression

Source OR (95% CI) Decreased Increased

Almeida et al,21 2011 0.67 (0.30-1.50)
Benseñor et al,22 2016 1.29 (0.93-1.79)
Bould et al,23 2012 1.01 (0.92-1.11)
De Jongh et al,27 2011 1.14 (0.58-2.26)
Engum et al,27 2002 0.53 (0.31-0.91)
Guimarães et al,30 2009 4.57 (1.61-12.99)
Hong et al,31 2018 3.35 (0.71-15.80)
Ittermann et al,32 2015 2.94 (0.75-11.46)
Kim et al,33 2010 1.27 (0.54-3.00)
Kim et al,34 2018 1.05 (0.94-1.17)
Kvetny et al,35 2015 1.06 (0.73-1.54)
Lin et al,37 2016 1.59 (1.18-2.16)
Manciet et al,38 1995 1.07 (0.13-8.87)
Maugeri et al,39 1998 4.00 (0.97-16.55)
Park et al,41 2010 0.99 (0.57-1.72)
Pop et al,42 1998 3.62 (0.80-16.39)
Shinkov et al,43 2014 1.85 (1.26-2.71)
Thomsen et al,44 2005 2.00 (1.60-2.49)
van de Ven et al,45 2012 1.09 (0.05-22.86)
Total 1.30 (1.08-1.57)
Prediction interval (0.71-2.38)
Heterogeneity: χ218 = 70.39 (P < .001), I2 = 74% 0.04 0.1 1 10 13
Test for overall effect: z = 2.83 (P = .005) OR (95% CI)
OR indicates odds ratio.

Small study associations are possible (P = .09) and adding CI, 0.63-1.27]), as did the analysis restricted to studies carry-
5 studies in trim and fill lowered the OR to 1.17 (95% CI, 0.97- ing a low risk of bias (Table 2). In addition, a leave-1-out-
1.41). Egger test (P = .012) was also positive in studies on sub- analysis revealed that Carta et al 2 4 disproportionately
clinical hypothyroidism, and after adding 6 studies in trim and increased the effect size (exclusion led to an OR of 1.11 [95%
fill, the OR was 1.04 (95% CI, 0.9-1.20). In studies on overt hy- CI, 0.82-1.51]).
pothyroidism, no small study associations were detected We calculated an I2 of 65% in the analysis on TPO anti-
(Table 2). bodies positivity, which was reduced when we restricted the
In primary leave-1-out analysis, omitting the study of calculation to studies with a low risk of bias. Eliminating
Thomsen et al44 reduced the association to 1.22 (95% CI, 1.03- studies by Carta et al24 or Engum et al28 each reduced I2 by
1.43). Similarly, removing the study of Shinkov et al43 low- 10%. In the analysis on hypothyroidism, I 2 amounted to
ered the association of subclinical hypothyroidism to 1.07 (95% 74% and did not decrease with limiting the calculation to
CI, 0.97-1.17). In overt hypothyroidism, removal of studies by studies with low risk of bias. However, it was brought down
Engum et al27 and Thomsen et al44 increased ORs to 1.93 (95% to 60% when leaving out Thomsen et al.44 Among investi-
CI, 1.63-2.30) and 1.84 (95% CI, 0.97-3.47), respectively. Con- gations on subclinical hypothyroidism, heterogeneity was
versely, removal of Guimaraes et al30 decreased the associa- lower (45%). Here, removal of the study by Shinkov et al43
tion to 1.58 (95% CI, 1.00-2.50). further decreased I 2 to 26%. In overt hypothyroidism,
Risk of bias analyses showed decreased associations heterogeneity (70%) was reduced to 0% after omitting the
throughout (Table 2), and subclinical hypothyroidism was no study by Engum et al.27
longer associated with clinical depression in risk of bias analy- None of these analyses substantially changed the sum-
ses. As an exception, our primary analysis on hypothyroid- mary ORs (Table 2; eTable 2 in the Supplement). Tau, another
ism yielded slightly increased associations with studies with measurement of heterogeneity, was much lower than the ef-
a low risk of bias (OR, 1.33 [95% CI, 0.90-1.97]), but statistical fect estimate in almost all analyses.
significance was lost.
Individuals with TPO antibodies had a nominally Post hoc Analyses
increased OR of 1.24 (95% CI, 0.89-1.74) (Table 2 and To reduce bias, we restricted the analysis to studies compar-
Figure 3). Subgroup analyses revealed a nonsignificant ing men and women. For hypothyroidism, women (OR, 1.48
difference in associations between male and female [95% CI, 1.18-1.85]) showed a higher OR than men (OR, 0.71
individuals. Stratification for intake of thyroid medication [95% CI, 0.40-1.25]). There was no such contrast in autoim-
and DSM- or ICD-conforming diagnoses of depression munity studies (Table 2).
yielded slightly stronger associations (eTable 2 in the With regard to hypothyroidism, studies on older popula-
Supplement). tions reported smaller associations than studies on all ages.
Adjustment for small study effects (Egger test = 0.022) There were no such studies investigating TPO antibodies posi-
by 4 added studies reversed the association (OR, 0.89 [95% tivity (eTable 2 in the Supplement).

jamapsychiatry.com (Reprinted) JAMA Psychiatry December 2021 Volume 78, Number 12 1379

© 2021 American Medical Association. All rights reserved.

Downloaded from jamanetwork.com by Bryan González on 08/12/2025
 Research Original Investigation Hypothyroidism and Clinical Depression

P value
Discussion

Egger
test
.02

.11
.21
.30
for

NA

NA
Our analysis yielded 3 main results. (1) There is a moderate as-

0.89 (0.63-1.27)

0.79 (0.56-1.11)
sociation of overt, and less so of subclinical, hypothyroidism

OR (95% CI)
with clinical depression. (2) There is no statistically signifi-
Trim-and-fill analysis

cant association of verified TPO antibodies positivity with clini-

cal depression. (3) We found a stronger association of hypo-
NA

NA

NA
NA
thyroidism and clinical depression in female individuals than
studies

in male individuals.
No. of

NA

NA

NA
NA
4

2
Hypothyroidism and Depression
Heterogeneity

0.390

0.133
0.596
0.821
Hypothyroidism and clinical depression are associated with
NA

NA

an OR of 1.3 and even smaller when only studies with a low

τ

risk of bias are included or if possible reporting bias is taken

I2, %

NA

NA
65

12
72
56

into account. However, there is evidence for a dose-effect re-

Associations are reported as ORs and 95% CIs. Egger test P value is reported as 2-sided; less than .10 indicates funnel plot asymmetry and the possibility of reporting bias. lationship, as indicated by an OR of about 1.1 for subclinical
P value

and 1.8 for overt hypothyroidism.

.20

.41
.98
.85
NA

NA

The extent of the association is weaker than what some-

1.24 (0.89-1.74)

0.87 (0.64-1.20)
0.99 (0.44-2.20)
0.88 (0.26-3.06)

times seems to be assumed in clinical practice and, in part, in

TPO antibodies positivity

psychiatric research. It is also at odds with work by Loh et al,46

OR (95% CI)

who estimated the association of subclinical hypothyroidism

and depression to be 2.35 in their meta-analysis. However, they
NA

NA

analyzed a lower number of studies (n = 15) and mixed case-

control and cohort studies. In our opinion, the divide be-
studies
No. of

tween case-control and cohort studies needs to be strongly

NA

NA
9

3
3
3

emphasized.8 Case-control studies are frequently conducted

in tertiary care centers with a preponderance of severe cases,
P value

Egger
test

not representative of the general population. To avoid the bias

.09
.89

.01

.57
.35
.66
for

inherent to case-control studies, Wildisen et al10 carried out a

1.17 (0.97-1.41)

1.04 (0.90-1.20)

meta-analysis of individual patient data ascertained in 6 popu-

OR (95% CI)

lation-based studies. In contrasting subclinically hypothy-

Trim-and-fill analysis

roid and euthyroid probands, they estimated that the former

scored a mean of 0.29 (95% CI, −0.17 to 0.76) points higher on
NA

NA
NA
NA

the Beck Depression Inventory, less than 0.5% of the scale

width,47 irrelevant in the view of the authors.
studies
No. of

In comparison with the continuous measurement of de-

NA

NA
NA
NA
5

pression by Wildisen et al,10 we focused on a dichotomous out-

Heterogeneity

0.269
0.464

0.125

0.432
0.085

come: clinically relevant depression as defined by study au-

0
τ

Abbreviations: NA, not applicable; OR, odds ratio; TPO, thyroid peroxidase.

thors. Statistically, it is advisable to use continuous outcomes.

I2, %

However, with continuous measures, such as the Hamilton

74
70

45

88
13
0

Rating Scale for Depression, statistically significant differ-

P value

ences between individuals with and without hypothyroid-

.005

.001
.01

.04

.15

.23

ism can be clinically unimportant as long as differences re-

main below pathological limits. In this sense, using continuous
1.30 (1.08-1.57)
1.77 (1.13-2.77)

1.13 (1.01-1.28)

1.33 (0.90-1.97)
1.48 (1.18-1.85)
0.71 (0.40-1.25)

end points runs the risk of creating false-positive results.

OR (95% CI)

Nevertheless, our results are in line with Wildisen et al.10

Zhao et al48 estimated a higher association of subclinical hy-
Hypothyroidism

pothyroidism and depression (OR, 1.75 [95% CI, 0.97-3.17]), al-

though of borderline significance. Of note, they included fewer
studies

and smaller studies and measured higher heterogeneity, in-

No. of

dicating the preliminary nature of their result.

19

16
9

7
4
4
Table 2. Main Resultsa

Autoimmunity and Depression

Primary analysis

Post hoc female

hypothyroidism

hypothyroidism
Low risk of bias

Post hoc male

We did not find a statistically significant association of auto-

Subclinical

immunity and depression. To our knowledge, this is the first

Analysis

Overt

meta-analysis focusing on general population samples with

documented TPO antibody status. Our findings are at
a

1380 JAMA Psychiatry December 2021 Volume 78, Number 12 (Reprinted) jamapsychiatry.com

© 2021 American Medical Association. All rights reserved.

Downloaded from jamanetwork.com by Bryan González on 08/12/2025
 Hypothyroidism and Clinical Depression Original Investigation Research

Figure 3. Association of Thyroid Peroxidase Antibodies Positivity and Depression

Source OR (95% CI) Decreased Increased

Carta et al,24 2004 2.91 (1.33-6.37)
Delitala et al,36 2016 1.40 (0.75-2.61)
Engum et al,28 2005 0.79 (0.62-1.01)
Fjaellegaard et al,29 2015 0.97 (0.54-1.76)
Ittermann et al,32 2015 1.85 (0.53-6.40)
Lee et al,36 2019 0.68 (0.41-1.13)
Medici et al,40 2014 1.12 (0.49-2.56)
Pop et al,42 1998 3.00 (1.31-6.86)
van de Ven et al,45 2012 1.31 (0.79-2.17)
Total 1.24 (0.89-1.74)
Prediction interval (0.45-3.40)
Heterogeneity: χ28 = 23.03 (P = .003), I2 = 65% 0.1 1 10
Test for overall effect: z = 1.28 (P = .20) OR (95% CI)
OR indicates odds ratio.

variance with a recent meta-analysis7 publishing an OR of 3.3. thyroxine compared with placebo.54 In any case, this finding
This would be a very strong association, as indicated by the may be false positive because results were reported by sex in
projection by Siegmann et al6 that, annually, more than 20% only 4 studies and sex has been included in adjustments in
of patients with autoimmune hypothyroidism experience de- several studies.
pression. With our results, the figures are in the 7% to 9% range,
barely higher than the population prevalence.49,50 It is worth Limitations
pointing out the differences between the 2 approaches; in re- The inclusion of a multitude of studies led to variations in study
stricting our study to epidemiologic studies, we hope our re- design and methods of assessment. For example, the study by
sults are less vulnerable to biases arising from the use of Benseñor et al22 was conducted with civil servants. However,
samples from endocrinology or psychiatry clinics. We re- we consider it unlikely that such a selection introduces bias.
stricted our analyses to verified TPO antibody positivity, We did not investigate absolute risks, and a differential bias
whereas Siegmann et al6 considered hypothyroidism in gen- seems unlikely in the studies included. Nevertheless, in a sen-
eral a proxy for autoimmunity and included 35 168 individu- sitivity analysis, we restricted our summary estimate to stud-
als as opposed to 47 707 in the present investigation. ies that are strictly population based, and results did not sub-
In our sample of studies, TPO antibody status was mea- stantially change (eTable 2 in the Supplement). Another
sured in individuals with euthyroidism except for the inves- limitation arises from varying recruitment processes of stud-
tigations by Engum et al28 and Pop et al.42 Hence, we ex- ies. Most samples consisted of random samples or complete
cluded both studies in a sensitivity analysis and found that the registers of the population, and others, like Bould et al,23 re-
results still hold; the OR went slightly down to 1.23 (95% CI, cruited participants from a primary or ambulatory care set-
0.87-1.73) without reaching statistical significance. ting, possibly introducing biases. Reassuringly, leaving out such
Our result may in part reflect the preponderance of sub- studies showed no substantially different results.
clinical hypothyroidism in individuals with TPO antibody posi- Several investigations included patients taking thyroid
tivity, but it may also have bearing on pathophysiological con- medication, which may have blurred an association of under-
siderations, in particular in view of the negative analyses lying hypothyroidism with depression. However, when we con-
considering reporting bias, low risk of bias studies, and popu- trasted studies with vs those without intake of thyroid medi-
lation-based studies in the strict sense. Possibly, it is not the cation, we found a stronger association with medication
disturbance of the immune system that explains the comor- (eTable 2 in the Supplement). Therefore, in these studies, thy-
bidity. Hypothyroidism may work differently. More specific roid medication may be an indicator of severe thyroid disor-
pathways aside, studies conducted by Patten et al51,52 show der rather than a successful treatment of depression.
that, in an unspecific way, many chronic disorders increase the Further, several studies used cutoff values for cases, but
risk of having depression. in a strict sense, cutoffs represent a range of symptoms rather
than diagnostic entities. However, they serve their purpose as
Sex Differential reasonably good approximations in large epidemiologic stud-
A post hoc analysis confirmed the association of hypothyroid- ies, as discussed, for example, in the article by Engum et al28
ism and depression in female individuals (OR, 1.5) but not in from Norway. Nevertheless, we have carried out a sensitivity
male individuals (OR, 0.7). This possible gradient may be analysis restricted to those studies and found only marginal
caused by physiological differences. In a randomized clinical differences from the main analysis (eTable 2 in the Supple-
trial, supraphysiological add-on thyroxine in patients with de- ment).
pression and bipolar disorder was effective in female but not Between-study heterogeneity, as measured by I2, was sub-
in male individuals.53 On the other hand, female individuals stantial in various analyses. Single studies24,27,43,44 exerted a
with subclinical hypothyroidism did not benefit regarding de- strong influence on I2, but their elimination from the analysis
pressive symptoms when they had been given antenatal did not significantly change the main results. It is important

jamapsychiatry.com (Reprinted) JAMA Psychiatry December 2021 Volume 78, Number 12 1381

© 2021 American Medical Association. All rights reserved.

Downloaded from jamanetwork.com by Bryan González on 08/12/2025
 Research Original Investigation Hypothyroidism and Clinical Depression

to bear in mind that with large sample sizes, as in our study individuals who recently gave birth, health care personnel need
with a combined N close to 350 000 individuals, I2 is ex- to be aware of incident depression.
pected to be large. An indicator of heterogeneity indepen- Because it was necessary to calculate associations for many
dent of sample size is tau,55 and the fact that tau is low sup- of the included samples, effect sizes differ in degrees of ad-
ports the robustness of our findings (Table 2; eTable 2 in the justment. We tried to compensate for this by including mini-
Supplement). In sum, while the confidence intervals and even mally adjusted effect sizes. This, in turn, may have led to an
more so the prediction intervals show that larger or smaller ef- overestimation of associations.
fects remain a possibility, the present evidence suggests a mod-
erate association of hypothyroidism and clinical depression.
We cannot draw conclusions regarding hypothyroidism in
pregnancy because in our sample of studies, pregnant indi-
Conclusions
viduals were often excluded based on the assumption that hy- It may be time to reconsider the paradigm of a strong connec-
pothyroidism in pregnancy differs from that seen in the gen- tion between hypothyroidism and depression. The results of
eral population. Recently, however, Minaldi et al56 published other groups and our own findings indicate the contribution
a meta-analysis specifically on pregnancy and the postpar- of hypothyroidism to the pandemic of depression is probably
tum period and found an association similar to our result. In small. This is good news for patients with hypothyroidism or,
the 5 studies summarized, the risk ratio of developing post- in particular, with thyroid autoimmunity. In counseling, we
partum depression among individuals who were positive for may not be able to rule out depression as a comorbidity, but it
TPO antibodies compared with those unaffected was 1.49 (95% is not looming large as a very likely threat. Regarding re-
CI, 1.11-2.0). Assuming a causal relationship, by the numbers search, it appears autoimmunity is not a forceful driver of af-
of this study, 1 in 21 female individuals with TPO antibodies fective symptoms. A more promising link seems to be the level
will experience postpartum depression because of their thy- of thyroid hormones and disturbances of the hypothalamic pi-
roid condition. Against the backdrop of the generally as- tuitary adrenal/hypothalamic pituitary thyroid axis. Finally,
sumed 10% to 15% prevalence of postpartum depression,57 the our results point to a possible effect of sex on the interaction
finding does support current clinical practice because in of hypothyroidism and depression.

ARTICLE INFORMATION REFERENCES 10. Wildisen L, Del Giovane C, Moutzouri E, et al.

Accepted for Publication: July 19, 2021. 1. Joffe RT, Sokolov ST. Thyroid hormones, the An individual participant data analysis of
brain, and affective disorders. Crit Rev Neurobiol. prospective cohort studies on the association
Published Online: September 15, 2021. between subclinical thyroid dysfunction and
1994;8(1-2):45-63.
doi:10.1001/jamapsychiatry.2021.2506 depressive symptoms. Sci Rep. 2020;10(1):19111.
2. Bauer M, Goetz T, Glenn T, Whybrow PC. The doi:10.1038/s41598-020-75776-1
Author Contributions: Mr Bode and Dr Baethge thyroid-brain interaction in thyroid disorders and
had full access to all of the data in the study and mood disorders. J Neuroendocrinol. 2008;20(10): 11. Page MJ, McKenzie JE, Bossuyt PM, et al. The
take responsibility for the integrity of the data and 1101-1114. doi:10.1111/j.1365-2826.2008.01774.x PRISMA 2020 statement: an updated guideline for
the accuracy of the data analysis. reporting systematic reviews. BMJ. 2021;372(n71):n71.
3. Hage MP, Azar ST. The link between thyroid doi:10.1136/bmj.n71
Concept and design: Bode, Bschor, Baethge.
function and depression. J Thyroid Res. 2012;2012:
Acquisition, analysis, or interpretation of data: 12. Smarr KL, Keefer AL. Measures of depression
590648. doi:10.1155/2012/590648
All authors. and depressive symptoms: Beck Depression
Drafting of the manuscript: Bode, Baethge. 4. Jucevičiūtė N, Žilaitienė B, Aniulienė R, Inventory-II (BDI-II), Center for Epidemiologic
Critical revision of the manuscript for important Vanagienė V. The link between thyroid Studies Depression Scale (CES-D), Geriatric
intellectual content: Bode, Ivens, Bschor, Schwarzer, autoimmunity, depression and bipolar disorder. Depression Scale (GDS), Hospital Anxiety and
Open Med (Wars). 2019;14:52-58. doi:10.1515/med- Depression Scale (HADS), and Patient Health
Henssler.
2019-0008 Questionnaire-9 (PHQ-9). Arthritis Care Res
Statistical analysis: Bode, Ivens, Bschor, Schwarzer,
Baethge. 5. Raison CL, Capuron L, Miller AH. Cytokines sing (Hoboken). 2011;63(suppl 11):S454-S466.
Obtained funding: Bode, Baethge. the blues: inflammation and the pathogenesis of doi:10.1002/acr.20556
Administrative, technical, or material support: depression. Trends Immunol. 2006;27(1):24-31. 13. Wells GA, Shea B, O'Connell D, et al. The
Bschor, Henssler. doi:10.1016/j.it.2005.11.006 Newcastle-Ottawa Scale (NOS) for assessing the
Supervision: Bschor, Henssler, Baethge. 6. Siegmann EM, Müller HHO, Luecke C, Philipsen quality of nonrandomised studies in meta-analyses.
A, Kornhuber J, Grömer TW. Association of The Ottawa Hospital. Accessed April 28, 2021.
Conflict of Interest Disclosures: Dr Schwarzer
depression and anxiety disorders with autoimmune http://www.ohri.ca/programs/clinical_
reported personal fees from Roche Pharma as epidemiology/oxford.asp
thyroiditis: a systematic review and meta-analysis.
external statistical consultant outside the JAMA Psychiatry. 2018;75(6):577-584. doi:10.1001/
submitted work. No other disclosures were 14. Modesti PA, Reboldi G, Cappuccio FP, et al; ESH
jamapsychiatry.2018.0190 Working Group on CV Risk in Low Resource
reported.
7. Siegmann EM, Grömer TW. Additional data from Settings. Panethnic differences in blood pressure in
Funding/Support: This work was supported by the omitted study in a meta-analysis of the association Europe: a systematic review and meta-analysis.
Koeln Fortune Program/Faculty of Medicine, of depression and anxiety with autoimmune PLoS One. 2016;11(1):e0147601. doi:10.1371/journal.
University of Cologne (grant 389/2020). thyroiditis. JAMA Psychiatry. 2019;76(8):871. pone.0147601
Role of the Funder/Sponsor: The funder had doi:10.1001/jamapsychiatry.2019.1503 15. DerSimonian R, Laird N. Meta-analysis in clinical
no role in the design and conduct of the study; 8. Baethge C. Autoimmune thyroiditis and trials. Control Clin Trials. 1986;7(3):177-188.
collection, management, analysis, and depression. JAMA Psychiatry. 2018;75(11):1204-1204. doi:10.1016/0197-2456(86)90046-2
interpretation of the data; preparation, review, or doi:10.1001/jamapsychiatry.2018.2411 16. Riley RD, Higgins JP, Deeks JJ. Interpretation of
approval of the manuscript; and decision to submit 9. Hennessey JV. Autoimmune thyroiditis and random effects meta-analyses. BMJ. 2011;342:d549.
the manuscript for publication. depression. JAMA Psychiatry. 2018;75(11):1204-1205. doi:10.1136/bmj.d549
doi:10.1001/jamapsychiatry.2018.2414 17. Egger M, Davey Smith G, Schneider M, Minder
C. Bias in meta-analysis detected by a simple,

1382 JAMA Psychiatry December 2021 Volume 78, Number 12 (Reprinted) jamapsychiatry.com

© 2021 American Medical Association. All rights reserved.

Downloaded from jamanetwork.com by Bryan González on 08/12/2025
 Hypothyroidism and Clinical Depression Original Investigation Research

graphical test. BMJ. 1997;315(7109):629-634. Korea National Health and Nutrition Examination study. Thyroid. 2005;15(7):700-707. doi:10.1089/
doi:10.1136/bmj.315.7109.629 Survey. PLoS One. 2018;13(8):e0202258. thy.2005.15.700
18. Duval S, Tweedie R. Trim and fill: a simple doi:10.1371/journal.pone.0202258 45. van de Ven AC, Muntjewerff JW, Netea-Maier
funnel-plot-based method of testing and adjusting 32. Ittermann T, Völzke H, Baumeister SE, Appel K, RT, et al. Association between thyroid function,
for publication bias in meta-analysis. Biometrics. Grabe HJ. Diagnosed thyroid disorders are thyroid autoimmunity, and state and trait factors of
2000;56(2):455-463. doi:10.1111/j.0006-341X.2000. associated with depression and anxiety. Soc depression. Acta Psychiatr Scand. 2012;126(5):377-
00455.x Psychiatry Psychiatr Epidemiol. 2015;50(9):1417-1425. 384. doi:10.1111/j.1600-0447.2012.01870.x
19. Balduzzi S, Rücker G, Schwarzer G. How to doi:10.1007/s00127-015-1043-0 46. Loh HH, Lim LL, Yee A, Loh HS. Association
perform a meta-analysis with R: a practical tutorial. 33. Kim JM, Stewart R, Kim SY, et al. Thyroid between subclinical hypothyroidism and
Evid Based Ment Health. 2019;22(4):153-160. stimulating hormone, cognitive impairment and depression: an updated systematic review and
doi:10.1136/ebmental-2019-300117 depression in an older Korean population. meta-analysis. BMC Psychiatry. 2019;19(1):12.
20. Viechtbauer W. Conducting meta-analyses in R Psychiatry Investig. 2010;7(4):264-269. doi:10.1186/s12888-018-2006-2
with the metafor package. J Stat Software. 2010; doi:10.4306/pi.2010.7.4.264 47. Beck AT, Steer RA, Brown GK. Manual for the
36(3):48. 34. Kim JS, Zhang Y, Chang Y, et al. Subclinical Beck Depression Inventory-II. Psychological
21. Almeida OP, Alfonso H, Flicker L, Hankey G, hypothyroidism and incident depression in young Corporation; 1996.
Chubb SA, Yeap BB. Thyroid hormones and and middle-age adults. J Clin Endocrinol Metab. 48. Zhao T, Chen BM, Zhao XM, Shan ZY.
depression: the Health in Men study. Am J Geriatr 2018;103(5):1827-1833. doi:10.1210/jc.2017-01247 Subclinical hypothyroidism and depression:
Psychiatry. 2011;19(9):763-770. doi:10.1097/JGP. 35. Kvetny J, Ellervik C, Bech P. Is suppressed a meta-analysis. Transl Psychiatry. 2018;8(1):239.
0b013e31820dcad5 thyroid-stimulating hormone (TSH) associated with doi:10.1038/s41398-018-0283-7
22. Benseñor IM, Nunes MA, Sander Diniz MF, subclinical depression in the Danish General 49. Kessler RC, Berglund P, Demler O, et al;
Santos IS, Brunoni AR, Lotufo PA. Subclinical Suburban Population Study? Nord J Psychiatry. National Comorbidity Survey Replication. The
thyroid dysfunction and psychiatric disorders: 2015;69(4):282-286. doi:10.3109/08039488.2014. epidemiology of major depressive disorder: results
cross-sectional results from the Brazilian Study of 972454 from the National Comorbidity Survey Replication
Adult Health (ELSA-Brasil). Clin Endocrinol (Oxf). 36. Lee S, Oh SS, Park EC, Jang SI. Sex differences (NCS-R). JAMA. 2003;289(23):3095-3105.
2016;84(2):250-256. doi:10.1111/cen.12719 in the association between thyroid-stimulating doi:10.1001/jama.289.23.3095
23. Bould H, Panicker V, Kessler D, et al. hormone levels and depressive symptoms among 50. Kessler RC, Chiu WT, Demler O, Merikangas KR,
Investigation of thyroid dysfunction is more likely in the general population with normal free T4 levels. Walters EE. Prevalence, severity, and comorbidity
patients with high psychological morbidity. Fam Pract. J Affect Disord. 2019;249:151-158. doi:10.1016/j.jad. of 12-month DSM-IV disorders in the National
2012;29(2):163-167. doi:10.1093/fampra/cmr059 2019.02.027 Comorbidity Survey Replication. Arch Gen Psychiatry.
24. Carta MG, Loviselli A, Hardoy MC, et al. The link 37. Lin IC, Chen HH, Yeh SY, Lin CL, Kao CH. Risk of 2005;62(6):617-627. doi:10.1001/archpsyc.62.6.617
between thyroid autoimmunity (antithyroid depression, chronic morbidities, and l-thyroxine 51. Patten SB, Williams JV, Esposito E, Beck CA.
peroxidase autoantibodies) with anxiety and mood treatment in Hashimoto thyroiditis in Taiwan: Self-reported thyroid disease and mental disorder
disorders in the community: a field of interest for a nationwide cohort study. Medicine (Baltimore). prevalence in the general population. Gen Hosp
public health in the future. BMC Psychiatry. 2004; 2016;95(6):e2842. doi:10.1097/MD. Psychiatry. 2006;28(6):503-508. doi:10.1016/
4:25. doi:10.1186/1471-244X-4-25 0000000000002842 j.genhosppsych.2006.09.001
25. de Jongh RT, Lips P, van Schoor NM, et al. 38. Manciet G, Dartigues JF, Decamps A, et al. The 52. Patten SB, Williams JVA, Lavorato DH, et al.
Endogenous subclinical thyroid disorders, physical PAQUID survey and correlates of subclinical Patterns of association of chronic medical
and cognitive function, depression, and mortality in hypothyroidism in elderly community residents in conditions and major depression. Epidemiol
older individuals. Eur J Endocrinol. 2011;165(4):545- the southwest of France. Age Ageing. 1995;24(3): Psychiatr Sci. 2018;27(1):42-50. doi:10.1017/
554. doi:10.1530/EJE-11-0430 235-241. doi:10.1093/ageing/24.3.235 S204579601600072X
26. Delitala AP, Terracciano A, Fiorillo E, Orrù V, 39. Maugeri D, Motta M, Salerno G, et al Cognitive 53. Stamm TJ, Lewitzka U, Sauer C, et al.
Schlessinger D, Cucca F. Depressive symptoms, and affective disorders in hyper- and hypothyreotic Supraphysiologic doses of levothyroxine as
thyroid hormone and autoimmunity in a elderly patients. Arch Gerontol Geriatr. 1998;26: adjunctive therapy in bipolar depression:
population-based cohort from Sardinia. J Affect 305-312. doi:10.1016/S0167-4943(98)80043-5 a randomized, double-blind, placebo-controlled
Disord. 2016;191:82-87. doi:10.1016/j.jad.2015.11.019 40. Medici M, Direk N, Visser WE, et al. Thyroid study. J Clin Psychiatry. 2014;75(2):162-168.
27. Engum A, Bjøro T, Mykletun A, Dahl AA. An function within the normal range and the risk of doi:10.4088/JCP.12m08305
association between depression, anxiety and depression: a population-based cohort study. J Clin 54. Costantine MM, Smith K, Thom EA, et al;
thyroid function: a clinical fact or an artefact? Acta Endocrinol Metab. 2014;99(4):1213-1219. Eunice Kennedy Shriver National Institute of Child
Psychiatr Scand. 2002;106(1):27-34. doi:10.1034/ doi:10.1210/jc.2013-3589 Health and Human Development (NICHD)
j.1600-0447.2002.01250.x 41. Park YJ, Lee EJ, Lee YJ, et al. Subclinical Maternal-Fetal Medicine Units (MFMU) Network,
28. Engum A, Bjøro T, Mykletun A, Dahl AA. hypothyroidism (SCH) is not associated with Bethesda, MD. Effect of thyroxine therapy on
Thyroid autoimmunity, depression and anxiety; are metabolic derangement, cognitive impairment, depressive symptoms among women with
there any connections? an epidemiological study of depression or poor quality of life (QoL) in elderly subclinical hypothyroidism. Obstet Gynecol. 2020;
a large population. J Psychosom Res. 2005;59(5): subjects. Arch Gerontol Geriatr. 2010;50(3):e68-e73. 135(4):812-820. doi:10.1097/AOG.
263-268. doi:10.1016/j.jpsychores.2005.04.002 doi:10.1016/j.archger.2009.05.015 0000000000003724

29. Fjaellegaard K, Kvetny J, Allerup PN, Bech P, 42. Pop VJ, Maartens LH, Leusink G, et al. Are 55. Rücker G, Schwarzer G, Carpenter JR,
Ellervik C. Well-being and depression in individuals autoimmune thyroid dysfunction and depression Schumacher M. Undue reliance on I(2) in assessing
with subclinical hypothyroidism and thyroid related? J Clin Endocrinol Metab. 1998;83(9):3194- heterogeneity may mislead. BMC Med Res Methodol.
autoimmunity: a general population study. Nord J 3197. 2008;8:79. doi:10.1186/1471-2288-8-79
Psychiatry. 2015;69(1):73-78. doi:10.3109/ 43. Shinkov AD, Borisova AM, Kovacheva RD, et al. 56. Minaldi E, D’Andrea S, Castellini C, et al. Thyroid
08039488.2014.929741 Influence of serum levels of thyroid-stimulating autoimmunity and risk of post-partum depression:
30. Guimarães JM, de Souza Lopes C, Baima J, hormone and anti-thyroid peroxidase antibodies, a systematic review and meta-analysis of
Sichieri R. Depression symptoms and age and gender on depression as measured by the longitudinal studies. J Endocrinol Invest. 2020;43
hypothyroidism in a population-based study of Zung Self-Rating Depression Scale. Folia Med (3):271-277. doi:10.1007/s40618-019-01120-8
middle-aged Brazilian women. J Affect Disord. (Plovdiv). 2014;56(1):24-31. doi:10.2478/folmed- 57. Anokye R, Acheampong E, Budu-Ainooson A,
2009;117(1-2):120-123. doi:10.1016/j.jad.2008.12.012 2014-0004 Obeng EI, Akwasi AG. Prevalence of postpartum
31. Hong JW, Noh JH, Kim DJ. Association between 44. Thomsen AF, Kvist TK, Andersen PK, Kessing depression and interventions utilized for its
subclinical thyroid dysfunction and depressive LV. Increased risk of developing affective disorder in management. Ann Gen Psychiatry. 2018;17:18.
symptoms in the Korean adult population: the 2014 patients with hypothyroidism: a register-based doi:10.1186/s12991-018-0188-0

jamapsychiatry.com (Reprinted) JAMA Psychiatry December 2021 Volume 78, Number 12 1383

© 2021 American Medical Association. All rights reserved.

Downloaded from jamanetwork.com by Bryan González on 08/12/2025