Increased Psychiatric Morbidity Before and After the Diagnosis of Hypothyroidism: A Nationwide Register Study

Abstract

Background:

Thyroid hormones are necessary for fetal brain development, and hypothyroidism in adults has been associated with mood symptoms and reduced quality of life. Nevertheless, our knowledge regarding the association and temporal relation between hypothyroidism and mental disorders is ambiguous. Our objective was to investigate, at a nationwide level, whether a diagnosis of hypothyroidism is associated with psychiatric morbidity.

Methods:

This is an observational cohort study. On the basis of record linkage between different Danish health registers, 2822 hypothyroid singletons each matched with 4 nonhypothyroid controls were identified and followed over a mean period of 6 years (range 1–13). Additionally, we included 385 same-sex twin pairs discordant for hypothyroidism. Diagnoses of psychiatric disorders as well as treatment with antidepressants, antipsychotics, and anxiolytics were recorded. Logistic and cox regression models were used to assess the risk of psychiatric morbidity before and after the diagnosis of hypothyroidism, respectively.

Results:

Before the diagnosis of hypothyroidism, such individuals had an increased prevalence of diagnoses with psychiatric disorders (odds ratio, OR, 1.51; 95% confidence interval [CI 1.12–2.04]) and increased prevalence of treatment with antipsychotics (OR 1.49 [CI 1.29–1.73]), antidepressants (OR 1.50 [CI 1.35–1.67]), and anxiolytics (OR 1.28 [CI 1.16–1.41]). After the diagnosis of hypothyroidism, patients had a higher risk of being diagnosed with a psychiatric disorder (hazard ratio, HR, 2.40 [CI 1.81–3.18]), and an increased risk of being treated with antidepressants (HR 1.30 [CI 1.15–1.47]) and anxiolytics (HR 1.27 [CI 1.10–1.47]), but not antipsychotics (HR 1.13 [CI 0.91–1.41]). On the basis of the twin data, we could not demonstrate genetic confounding.

Conclusions:

Subjects with hypothyroidism have an increased risk of being diagnosed with a psychiatric disorder as well as being treated with antidepressants, antipsychotics, and anxiolytics both before and after the diagnosis of hypothyroidism.

Introduction

The lifetime risk of acquiring hypothyroidism is as high as around 5% (1), and our recent investigations based on nationwide data have documented that thyroid dysfunction, whether hypothyroidism or hyperthyroidism, is associated with excess mortality (2 –6). Recently, we dissected the type of somatic morbidity and the temporal relation between these morbidities and thyroid dysfunction, and have shown that the risk of being diagnosed with a number of somatic diseases is increased before and after the diagnosis of thyroid dysfunction (7,8).

An apparent association between thyroid dysfunction and psychiatric conditions has been suggested since the earliest descriptions of hypo- as well as hyperthyroidism (9). In adults, the effect of thyroid hormones—or more correctly, lacking such hormones—on brain and mental function is less clear-cut. Clearly, some studies are in favor of an association between hypothyroidism (whether of autoimmune origin or not) and psychiatric morbidity (10 –14), whereas others do not report such a link (15 –19). Unfortunately, the previous literature is hampered by weaknesses in terms of design. This is best illustrated by the relatively small study populations in most studies (n<100) (11,12,15,19,20), as well as the high heterogeneity in the definition of thyroid dysfunction and psychiatric disorders among studies. In addition, some studies are cohort studies (13,14,18 –20), whereas others are cross sectional (10 –12,15,17). As a consequence, no two studies are comparable. Even when considering the two types of studies based on adequate sample size, they reach dissimilar conclusions regarding the relationship between hypothyroidism and psychiatric disease (10,13). Unfortunately, none of the available studies were designed to evaluate the temporal relation between hypothyroidism and psychiatric morbidity.

It is unknown whether an association between hypothyroidism and psychiatric disorders, as we have shown for the association between hypothyroidism and mortality (5) and somatic morbidity (8), is influenced by genetic confounding. However, since hypothyroidism (21) as well as various psychiatric disorders (22,23) show familial aggregation, these conditions could very well be co-inherited. If present, the risk of psychiatric morbidity attributed to hypothyroidism would be overestimated.

In the present study, we have used Danish nationwide registers to evaluate the relation between hypothyroidism and a number of psychiatric outcomes. On the basis of these registers, we are able to address some of the shortcomings of previous studies. This includes a large sample size of incident hypothyroid cases analyzed prospectively; a wide spectrum of psychiatric morbidity; and evaluation of psychiatric morbidity both before and after the diagnosis of hypothyroidism, and thereby, for the first time, we illuminate the temporal relation between hypothyroidism and psychiatric morbidity. Additionally, by evaluating the risk of psychiatric disease in twin pairs discordant for hypothyroidism, we obtain insight into whether there is genetic confounding.

Materials and Methods

Data sources

The Danish Civil Registration System (24), the Danish Twin Registry (25), the Danish National Patient Registry (DNPR) (24), and the Danish National Prescription Registry (DNPrR) (24) are nationwide registers covering information such as demographic data, hospital treatments, and prescriptions of drugs on an individual person level. All registers are hosted at Statistics Denmark (24) and have previously been described in detail (5). The 10-digit personal identification number (CPR number), assigned to all Danish citizens, enables record linkage between all the mentioned databases on an individual level.

Diagnosis of hypothyroidism

Information on thyroid status was drawn from DNPrR and DNPR. Individuals should be recorded in at least one of these two registers in order to be classified as having hypothyroidism. Hypothyroidism was defined by at least two dispensed prescriptions of thyroid hormone, L-thyroxine (Anatomical Therapeutic Chemical classification system [ATC]=H03A), in DNPrR. In DNPR, hypothyroidism was defined by the relevant International Classification of Diseases 10th Revision (ICD-10) codes (E03.2–E03.9). Only incident subjects diagnosed with primary hypothyroidism between December 31, 1995, and December 31, 2008, and older than 18 years were eligible for the present study. The validity of DNPR is considered high, and misclassification of thyroid dysfunction has been shown to occur in less than 2% of cases (26). Individuals diagnosed with malignant thyroid diseases, congenital hypothyroidism, or pituitary hypothyroidism, represented by the ICD-8 codes 193.99, 243.99, 253.00–253.02, 253.09–253.11, 253.15, 253.18, 253.19, 253.90, and 253.99 and the ICD-10 codes C7.39, E00.0–E00.9, E03.0–E03.1, E22.0–E22.9, E23.0–E23.7, and E24.0, were excluded. Additionally, subjects diagnosed with hyperthyroidism (defined by the ICD-8 codes 242.00–242.09 and the ICD-10 codes E05.0–E05.9 in DNPR, or by at least two dispensed prescriptions of antithyroid drugs [ATC=H03B] in DNPrR before the diagnosis of hypothyroidism) were excluded.

Study population

Singleton cases were selected from a random 5% sample of all Danes, identified from the Danish Civil Registration System between 1996 and 2008 (24). In addition, twin cases were selected from the Danish Twin Registry (25). In all, 2822 hypothyroid singleton individuals were matched for age and sex with 11,288 nonhypothyroid control individuals (1:4 match), after the principles of density sampling (27). From the Danish Twin Registry, 385 same-sex twin pairs, discordant for hypothyroidism, were identified. Date of debut with hypothyroidism (index date) was the first date of registration in either DNPR or DNPrR. Cases and controls were followed until death, migration, or December 31, 2008, whichever came first.

Psychiatric morbidity

The overall outcome was psychiatric disease, as identified from DNPR and DNPrR between 1995 and 2008. In Denmark, most cases with mild to moderate mental disorders are diagnosed and treated by the general practitioners and are therefore not registered in DNPR (28). Consequently, psychiatric morbidity was defined either based on being diagnosed with a psychiatric disorder, including that of hospital outpatient clinics (based on registrations of ICD codes), or by the use of psychiatric medication (antipsychotics, antidepressants, or anxiolytics) with at least one prescription. All relevant ATC codes as well as ICD-10 codes are shown in Table 1. Both DNPR and DNPrR have been used in other Danish register-based surveys studying psychiatric morbidity, for example, based on the use of benzodiazepines, antipsychotics, and antidepressants (13,29,30). For each individual, the first date of possible registration in each of these outcome groups was identified. Accordingly, we stratified for the first registration in each group before and after the date of the diagnosis of hyperthyroidism.

Table 1.

Identification of Psychiatric Morbidity in the Danish National Patient Registry and the Danish National Prescription Registry

Outcome category	ICD-10	ATC
Use of antidepressants	—	N06A
Use of antipsychotics	—	N05A
Use of anxiolytics	—	N05BA
Diagnoses of psychiatric disorders	DF20–DF25, DF28–DF29, DF30–DF34, DF38–DF39, DF41.0–DF41.9

ATC, Anatomical Therapeutic Chemical classification system; ICD-10, International Classification of Diseases 10th Revision.

Data analyses

Group frequencies were compared with the Pearson χ²-test, whereas group means were compared by the t-test.

In the singleton population, the odds ratio (OR) for the prevalence of psychiatric morbidity before the diagnosis of hypothyroidism was evaluated in a conditional logistic regression analysis adjusted for age and sex. The relationship between first registration of hypothyroidism and subsequent first registration of psychiatric morbidity was evaluated by a Cox regression model where age was chosen as the underlying time variable. Person-years of follow-up, for cases as well as controls, were accumulated from the date of diagnosis of hypothyroidism of the case and terminated on the date of diagnosis of morbidity, migration, death, or end of follow-up (December 31, 2008), whichever came first.

In all Cox analyses of the singletons, the variable “pair” (sets of 1 case and 4 controls) was used as a stratum variable, fixing the baseline hazard within a matched pair, while at the same time allowing this baseline hazard to vary freely between pairs. Afterward, all Cox regression analyses were adjusted for the degree of comorbidity preceding the diagnosis of hypothyroidism, using the Charlson score (CS) (31), which is a score accounting for 19 disease groups (myocardial infarction, heart failure, vascular disease, cerebrovascular disease, dementia, chronic lung disease, rheumatic disease, gastric ulcer, liver disease, diabetes mellitus without complications, diabetes mellitus with complications, hemiplegia, kidney disease, cancer, cancer with metastases, lymphoma, leukemia, liver failure, and AIDS) by creating a weighted score on an individual level, to optimize the prediction of the one-year mortality risk within each disease category. The CS was originally constructed to estimate one-year mortality in patients with breast cancer, but has subsequently been validated and used in different phenotypes, including nonmalignant diseases (32).

As hypothyroidism is associated with increased mortality (5), cases have a shorter observation period than the controls and, at least theoretically, a lower risk of being diagnosed with psychiatric morbidity. To account for this influence of competing risk, the risk of psychiatric morbidity was reevaluated by the method of Fine and Gray (33).

We also performed intrapair analyses of the twin population. In these analyses, the hypothyroid twin was matched with the corresponding euthyroid co-twin. After pooling all psychiatric disease groups, the differences in overall frequency of psychiatric morbidity among the twin pairs discordant for hypothyroidism were evaluated by conditional logistic regression analyses adjusted for CS. Significant differences were defined as a p-value below 0.05, using two-tailed tests. All analyses were conducted using STATA version 11.0 (2009; Stata Corporation, College Station, TX).

Results

Baseline characteristics of the study populations

The baseline characteristics of the 2822 singletons diagnosed with hypothyroidism, of which the majority, 2116, were from DNPrR and assumed primary care, as well as the 385 twin pairs discordant for hypothyroidism are presented in Table 2. There were more females than males with hypothyroidism, and the mean age at diagnosis for singletons and twins was 58 and 54 years, respectively. Cases as well as controls were observed over a mean period of 6 years (range 1–13 years). In general, cases had a higher CS than their respective controls.

Table 2.

Baseline Characteristics of the Study Populations

	Singletons			Twins ^a
	Cases	Controls	p	Cases	Controls	p
Number	2822	11,288		385	385
Mean age at diagnosis	58	58	1.000	54	54	1.000
Females (%)	84	84	1.000	82	82	1.000
Mean follow-up (years)	6	6	0.218	6	6	0.672
CS^b=0 (%)	60	67	<0.001	60	69	0.016
CS^b=1 (%)	21	19	0.027	22	18	0.178
CS^b>1 (%)	19	14	<0.001	18	13	0.090

Twins from twin pairs discordant for hypothyroidism.

The Charlson score before the diagnosis of hypothyroidism of the case.

Overall associations between hypothyroidism and psychiatric morbidity

The results from the overall analyses of hypothyroidism and psychiatric morbidity are evident from Table 3. Hypothyroidism was positively associated with the diagnosis of a psychiatric disorder, which did not change significantly when only evaluating on the primary cause of hospitalization (data not shown). When analyzing the use of psychiatric medication, there was an increased frequency of users of antidepressants (p<0.001), antipsychotics (p<0.001), and anxiolytics (p<0.001) among the hypothyroid individuals.

Table 3.

Disease Prevalences, Odds Ratios Before Diagnosis of Hypothyroidism, and Hazard Ratios After Diagnosis

	Disease prevalence (%)
Disease	Cases ^a (n)	Controls ^a (n)	p	Odds ratio [CI] ^b	Hazard ratio [CI] ^c
Diagnoses of psychiatric disorders	5.1 (144)	2.7 (305)	<0.001	1.51 [1.12–2.04]	2.40 [1.81–3.18]
				1.23 [0.87–1.73]^d	2.18 [1.58–3.01]^e
					2.11 [1.55–2.87]^f
Use of antidepressants	34 (960)	25 (2872)	<0.001	1.50 [1.35–1.67]	1.30 [1.15–1.47]
				1.45 [1.30–1.62]^d	1.18 [1.03–1.36]^e
					1.15 [1.00–1.32]^f
Use of antipsychotics	13 (380)	10 (1140)	<0.001	1.49 [1.29–1.73]	1.13 [0.91–1.41]
				1.40 [1.20–1.64]^d	1.02 [0.79–1.33]^e
					1.01 [0.78–1.30]^f
Use of anxiolytics	36 (1029)	30 (3424)	<0.001	1.28 [1.16–1.41]	1.27 [1.10–1.47]
				1.22 [1.11–1.35]^d	1.11 [1.00–1.39]^e
					1.15 [0.99–1.35]^f

Percentage of cases and controls with a first-time hit of psychiatric morbidity.

Odds ratios before the diagnosis of hypothyroidism.

Hazard ratios after the diagnosis of hypothyroidism.

Odds ratios before the diagnosis of hypothyroidism with a 365-day censoring window.

Hazard ratios after the diagnosis of hypothyroidism with a 365-day censoring window.

Competing risk analyses: hazard ratios after the diagnosis of hypothyroidism with a 365-day censoring window.

CI, 95% confidence interval.

Psychiatric morbidity preceding the diagnosis of hypothyroidism

Overall, subjects with hypothyroidism had an increased risk of being diagnosed with a psychiatric disorder (OR 1.51; 95% confidence interval [CI 1.12–2.04]) before the diagnosis of hypothyroidism (Table 3). Although not significant, this was also seen when evaluating only based on the primary causes of hospitalization (OR 1.38 [CI 0.93–2.06]) and when analyzing the prevalence of users of psychiatric medication: antidepressants (OR 1.50 [CI 1.35–1.67]), antipsychotics (OR 1.49 [CI 1.29–1.73]), and anxiolytics (OR 1.28 [CI 1.16–1.41]). Evaluating the same categories, but censoring diagnoses made within 365 days before the diagnosis of hypothyroidism, in order to minimize detection bias, such as Berkson's bias (34), yielded essentially similar results for the use of antidepressants (OR 1.45 [CI 1.30–1.62]), antipsychotics (OR 1.40 [CI 1.20–1.64]), and anxiolytics (OR 1.22 [CI 1.11–1.35]), whereas the risk of being diagnosed with a psychiatric disorder was not statistically significant (OR 1.23 [CI 0.87–1.73]).

Psychiatric morbidity after the diagnosis of hypothyroidism

As evident from Table 3, the risk of a first registration of a diagnosis of a psychiatric disorder was significantly increased (hazard ratio, HR, 2.40 [CI 1.81–3.18]), and remained so when evaluating only based on the primary causes of hospitalization (HR 1.79 [CI 1.16–2.75]). Focusing on the risk of a first registration of using psychiatric medication after the diagnosis of hypothyroidism, a significantly increased risk was found for antidepressants (HR 1.30 [CI 1.15–1.47]) and anxiolytics (HR 1.27 [CI 1.10–1.47]), which, however, attenuated for antipsychotics (HR 1.13 [CI 0.90–1.41]). When applying censoring of diagnoses made within 365 days after the diagnosis of hypothyroidism, the results changed significantly neither for the use of antidepressants (HR 1.18 [CI 1.03–1.36]), nor for the use of anxiolytics (HR 1.11 [CI 1.00–1.39]) or antipsychotics (HR 1.02 [CI 0.79–1.33]). In the competing risk analyses, the results did not change significantly (Table 3).

Overall associations between hypothyroidism and morbidity in twins

In a combined analysis, based on being diagnosed with a psychiatric disorder and/or receiving antipsychotics, antidepressant, or anxiolytics, a significantly increased risk of psychiatric morbidity was found in the hypothyroid twins compared with their euthyroid co-twins (OR 1.64 [CI 1.17–2.31]). Even when evaluating only monozygotic twin pairs, we found a significantly increased risk of psychiatric morbidity (OR 2.14 [CI 1.14–4.04]). Because of the lack of power, further subdivision, according to whether treated in a hospital setting and type of drug used, was not meaningful.

Discussion

The current knowledge regarding the impact of psychiatric morbidity—especially, the temporal relation between psychiatric morbidity and hypothyroidism—is limited. Some studies (10 –14), but not all (15 –19), have demonstrated an increased risk of psychiatric morbidity associated with hypothyroidism. This inconsistency may partly be explained by inhomogeneous definitions of psychiatric outcome as well as thyroid phenotypes, the cross-sectional designs, and inadequate sample size in most of the studies. In the present study, we limited these shortcomings by following and investigating a large population-based sample of individuals with accepted definitions of thyroid dysfunction and psychiatric diseases. Using this design, we found a positive overall association between hypothyroidism on the one hand and being diagnosed with a psychiatric disorder as well as treatment with antipsychotics, antidepressants, or anxiolytics, on the other hand.

Consistent with another register-based survey (13), we demonstrated an increased risk of being diagnosed with a psychiatric disorder associated with hypothyroidism. In addition to this, our design allowed evaluation and demonstration of an increased risk of being treated with antipsychotics, antidepressants, and anxiolytics, both before and after the diagnosis of hypothyroidism registered from hospitals or primary care. Although we cannot directly distinguish cause and effect, these findings could be compatible with a causal relation between hypothyroidism and psychiatric morbidity, or vice versa. Our findings are biologically plausible as, for example, thyroid hormones interact with the equilibrium of neurotransmitters, such as serotonin and norepinephrine (35), both involved in the pathway of developing psychiatric diseases (36,37), so the hypothyroid state could be the cause of the psychiatric diagnosis. Conversely, since, for example, lithium is known to inhibit thyroid hormone secretion directly (38), it is possible that the psychiatric condition, or its treatment, could cause the hypothyroid state. Although our study population is relative large, it does not allow investigation of the potential independent effect of the numerous drugs that have been shown to affect thyroid hormone levels or thyroid autoimmunity. Also, shared genetic factors could potentially explain the increased morbidity before and after the diagnosis of hypothyroidism. Since there is a considerable genetic component in the etiology of hypothyroidism—at least the autoimmune and most common subtype—and various psychiatric diseases, they could potentially be co-inherited (21 –23,39). However, when controlling for genetic confounding, by analyzing twin pairs discordant for hypothyroidism, the increased risk of being diagnosed with psychiatric morbidity persisted, also after stratifying for zygosity. This indicates that the results from the overall association of the singleton population were not significantly influenced by shared genetic factors.

A possible bias in the present study is the risk of Berkson's bias (34), which, in this case, refers to an increased awareness of other diseases because of diagnosis and treatment of hypothyroidism, or by investigating thyroid status in patients with depression. In order to minimize this potential detection bias, we censored our morbidity data according to a time frame of 365 days before and after the diagnosis of hypothyroidism, and reanalyzed our findings. We cannot be certain that this procedure completely eliminates such bias. However, the fact that our results did not change significantly, whether analyzed one way or the other, in addition to the consistency of our results, independent of whether analyses were restricted to psychiatric diagnoses as the primary diagnoses or included these as secondary diagnoses from patients treated in a hospital setting, suggests that detection bias is not a major concern. Importantly, despite these indications of potential temporality in the relation between hypothyroidism and psychiatric morbidity, it is not possible to determine the exact time of onset of either thyroid disorders or psychiatric morbidity, and therefore this remains speculative.

The strengths of our study include ascertainment of participants from nationwide population-based registers, high power because of a large sample size, and a relatively long observation period. The fact that patients from both a hospital setting, whether in- or outpatient, and primary care are accounted for minimizes the risk of selection bias in the identification of the hypothyroid cases. In contrast, although no systematic bias was introduced, the lack of information regarding the cause of hypothyroidism (whether autoimmune or not), as well as any effect of treatment on thyroid dysfunction, in line with most other surveys, are weaknesses. Also, because of the lack of biochemical data, the impact of thyroid hormone therapy on the psychiatric disorders remains unknown. Our definition of psychiatric morbidity was based on diagnoses from hospitals (40), as well as records of prescriptions of antidepressants, antipsychotics, and anxiolytics as a proxy for the risk of depression, psychoses, and anxiety disorders, respectively, which for our purpose is a strength. Several studies, investigating, for example, schizophrenia and single-episode depression, have shown high validity of these diagnoses in DNPR (41,42). In contrast, the use of psychiatric medication as a proxy for psychiatric diagnoses has a low specificity, since some drugs used in the treatment of psychiatric diseases are also used in the treatment of other conditions such as pain, headache, insomnia, and irritable bowel. Therefore, when based on prescriptions, we undoubtedly overestimate the total number of individuals with psychiatric diseases in the present study. This overestimation was to some degree controlled for by adjusting for the CS. Still, since this is most likely true for cases as well as for controls, the relative difference is most likely not affected by this potential bias. Moreover, we find it reassuring that the results are consistent when analyzing psychiatric morbidity based on the more specific hospital diagnoses (through DNPR) and when solely based on the use of the pertinent medication (from DNPrR). The lack of significance of the OR when evaluating diagnoses with psychiatric disorders 365 days before diagnosis of hypothyroidism is most likely because of lack of power in these analyses.

In summary, hypothyroidism is significantly associated with psychiatric morbidity. Before the diagnosis of hypothyroidism, there is a significantly increased prevalence of being diagnosed with a psychiatric disorder or being treated with antidepressants, antipsychotics, or anxiolytics. After the diagnosis of hypothyroidism, there is a significant excess risk of being diagnosed with a psychiatric disorder or being treated with antidepressants or anxiolytics. Our data suggest that neither genetic confounding nor detection bias constitutes major pitfalls in our interpretation.

Footnotes

Acknowledgments

F.B. and M.T. are enrolled as PhD students financially supported by the School of Endocrinology, University of Southern Denmark in Odense. F.B. and M.T. have also received funding from the Danish Thyroid Patient Organization. L.H. is the recipient of an unrestricted research grant from the Novo Nordisk Foundation.

Author Disclosure Statement

M.T., F.B., D.A., K.C., T.H.B., and L.H. have nothing to declare. The authors had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

References

Carle

, Laurberg

, Pedersen

, Knudsen

, Perrild

, Ovesen

, Rasmussen

, Jørgensen

. 2006. Epidemiology of subtypes of hypothyroidism in Denmark. Eur J Endocrinol, 154:21–28.

Brandt

, Green

, Hegedüs

, Brix

. 2011. A critical review and meta-analysis of the association between overt hyperthyroidism and mortality. Eur J Endocrinol, 165:491–497.

Brandt

, Almind

, Christensen

, Brix

, Hegedüs

. 2012. Excess mortality in hyperthyroidism: the influence of pre-existing comorbidity and genetic confounding: a Danish nationwide register-based cohort study of twins and singletons. J Clin Endocrinol Metab, 97:4123–4129.

Thvilum

, Brandt

, Brix

, Hegedüs

. 2012. A review of the evidence for and against increased mortality in hypothyroidism. Nat Rev Endocrinol, 6:417–424.

Thvilum

, Brandt

, Almind

, Christensen

, Hegedüs

, Brix

. 2013. Excess mortality in patients diagnosed with hypothyroidism. A nationwide cohort study of singletons and twins. J Clin Endocrinol Metab, 98:1069–1075.

Brandt

, Thvilum

, Almind

, Christensen

, Green

, Hegedüs

, Brix

. 2013. Graves' disease and toxic nodular goiter are both associated with increased mortality but differ with respect to the cause of death. A Danish population-based register study. Thyroid, 23:408–413.

Brandt

, Thvilum

, Almind

, Christensen

, Green

, Hegedüs

, Brix

. 2013. Morbidity before and after the diagnosis of hyperthyroidism: a nationwide register-based study. PLoS ONE, 8:e66711.

Thvilum

, Brandt

, Almind

, Christensen

, Brix

, Hegedüs

. 2013. Type and extent of morbidity before and after the diagnosis of hypothyroidism. A nationwide register study. PLoS ONE, 8:e75789.

Whybrow

, Prange

Jr , Treadway

. 1969. Mental changes accompanying thyroid gland dysfunction. A reappraisal using objective psychological measurement. Arch Gen Psychiatry, 20:48–63.

10.

Engum

, Bjorø

, Mykletun

, Dahl

. 2002. An association between depression, anxiety and thyroid function—a clinical fact or an artefact?. Acta Psychiatr Scand, 106:27–34.

11.

Carta

, Loviselli

, Hardoy

, Massa

, Cadeddu

, Sardu

, Carpiniello

, Dell'Osso

, Mariotti

. 2004. The link between thyroid autoimmunity (antithyroid peroxidise autoantibodies) with anxiety and mood disorders in the community: a field of interest for public health in the future. BMC Psychiatry, 4:25.

12.

Carta

, Hardoy

, Carpiniello

, Murru

, Marci

, Carbone

, Deiana

, Cadeddu

, Mariotti

. 2005. A case control study on psychiatric disorders in Hashimoto disease and euthyroid goitre: not only depressive but also anxiety disorders are associated with thyroid autoimmunity. Clin Pract Epidemiol Ment Health, 1:23.

13.

Thomsen

, Kvist

, Andersen

, Kessing

. 2005. Increased risk of developing affective disorder in patients with hypothyroidism: a register-based study. Thyroid, 15:700–707.

14.

Panicker

, Evans

, Bjorø

, Åsvold

, Dayan

, Bjerkeset

. 2009. A paradoxical difference in relationship between anxiety, depression and thyroid function in subjects on and not on T4: findings from the HUNT study. Clin Endocrinol, 71:574–580.

15.

Pop

, Martens

, Leusink

, van Son

, Knottnerus

, Ward

, Metcalfe

, Weetman

. 1998. Are autoimmune thyroid dysfunction and depression related?. J Clin Endocrinol Metab, 83:3194–3197.

16.

Hermann

, Hewer

, Lederbogen

. 2003. Testing the association between thyroid dysfunction and psychiatric diagnostic group in an iodine-deficient area. J Psychiatry Neurosci, 29:444–449.

17.

Engum

, Bjøro

, Mykletun

, Dahl

. 2005. Thyroid autoimmunity, depression and anxiety; are there any connections? An epidemiological study of a large population. J Psychosom Res, 59:263–268.

18.

Williams

, Harris

, Dayan

, Evans

, Gallacher

, Ben-Shlomo

. 2009. Thyroid function and the natural history of depression: findings from the Caerphilly Prospective Study (CaPS) and a meta-analysis. Clin Endocrinol, 70:484–492.

19.

Eaton

, Pedersen

, Nielsen

, Mortensen

. 2010. Autoimmune diseases, bipolar disorder, and non-affective psychosis. Bipolar Disord, 12:638–646.

20.

Gulseren

, Gulseren

, Hekimsoy

, Cetinay

, Ozem

, Toaktlioglu

. 2006. Depression, anxiety, health-related quality of life, and disability in patients with overt and subclinical thyroid dysfunction. Arch Med Res, 37:133–139.

21.

Brix

, Kyvik

, Hegedüs

. 2000. A population-based study of chronic autoimmune hypothyroidism in Danish twins. J Clin Endocrinol Metab, 85:536–539.

22.

Shih

, Belmonte

, Zandi

. 2004. A review of the evidence from family, twin and adoption studies for genetic contribution to adult psychiatric disorders. Int Rev Psychiatry, 16:260–283.

23.

Sullivan

, Kendler

, Neale

. 2003. Schizophrenia as a complex trait: evidence from a meta-analysis of twin studies. Arch Gen Psychiatry, 60:1187–1192.

24.

Thygesen

, Daasnes

, Thaulow

, Brønnum-Hansen

. 2011. Introduction to Danish (nationwide) registers on health and social issues: structure, access, legislation and archiving. Scand J Public Health, 39:12–16.

25.

Thomsen

, Kvist

, Andersen

, Kessing

. 2005. Increased risk of affective disorder following hospitalisation with hyperthyroidism—a register based study. Eur J Endocrinol, 152:535–543.

26.

Skytthe

, Kyvik

, Holm

, Christensen

. 2011. The Danish Twin Registry. Scand J Public Health, 39:75–78.

27.

Rothman

, Greenland

, Lash TL 2008 Modern Epidemiology: Case Control Studies. Third edition. Lippincott Williams & Wilkins, Philadelphia, PA.

28.

Mors

, Perto

, Mortensen

. 2011. The Danish Psychiatric Research Register. Scand J Public Health, 39:54–57.

29.

Ekedahl

, Lidbeck

, Lithman

, Noreen

, Melander

. 1993. Benzodiazepine prescribing patterns in a high-prescribing Scandinavian community. Clin Pharmacol, 44:141–146.

30.

Nakagawa

, Pedersen

, Olsen

, Mortensen

, Sørensen

, Johnsen

. 2006. Antipsychotics and risk of first-time hospitalization for myocardial infarction: a population-population case-control study. J Intern Med, 260:451–458.

31.

Charlson

, Pompei

, Ales

, MacKenzie

. 1987. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis, 40:373–383.

32.

Almargo

, Salvado

, Garcia-Vidal

, Rodríguez-Carballeira

, Cuchi

, Torres

, Heredia

. 2012. Pseudomonas aeruginosa and mortality after hospital admission for chronic obstructive pulmonary disease. Respiration, 84:36–43.

33.

Fine

, Gray

. 1999. A proportional hazards model for the subdistribution of a competing risk. J Am Stat Assoc, 94:496–509.

34.

Berkson

. 1946. Limitations of the application of fourfold table analysis to hospital data. Biometric Bull, 2:126–135.

35.

Bauer

, Goetz

, Glenn

, Whybrow

. 2008. The thyroid-brain interaction in thyroid disorders and mood disorders. J Neuroendocrinol, 20:1101–1114.

36.

Argyropoulus

, Bell

, Nutt

. 2001. Brain function in social anxiety disorder. Psychiatr Clin North Am, 24:707–722.

37.

Krishnan

, Nestler

. 2008. The molecular neurobiology of depression. Nature, 455:894–902.

38.

McKnight

, Adida

, Budge

, Stockton

, Goodwin

, Geddes

. 2012. Lithium toxicity profile: a systematic review and meta-analysis. Lancet, 379:721–728.

39.

Hansen

, Brix

, Iachine

, Kyvik

, Hegedüs

. 2006. The relative importance of genetic and environmental effects for the early stages of thyroid utoimmunity: a study of healthy Danish twins. Eur J Endocrinol, 154:29–38.

40.

Munk-Jørgensen

, Østergaard

. 2011. Register-based studies of mental disorders. Scand J Public Health, 39:170–174.

41.

Jakobsen

, Frederiksen

, Hansen

, Jansson

, Parnas

, Werge

. 2005. Reliability of clinical ICD-10 schizophrenia diagnoses. Nord J Psychiatry, 59:209–212.

42.

Bock

, Bukh

, Vinberg

, Gether

, Kessing

. 2009. Validity of the diagnosis of a single depressive episode in a case register. Clin Pract Epidemol Ment Health, 5:4.