Coeliac disease and risk of mood disorders — A general population-based cohort study

Journal of Affective Disorders 99 (2007) 117 – 126 www.elsevier.com/locate/jad

Research report

Coeliac disease and risk of mood disorders — A general population-based cohort study ☆ Jonas F. Ludvigsson a,b,⁎, Johan Reutfors b , Urban Ösby c , Anders Ekbom b,d , Scott M. Montgomery b,e a

b

Department of Paediatrics, Örebro University Hospital, Sweden Clinical Epidemiology Unit, Department of Medicine, Karolinska University Hospital, Karolinska Institutet, Sweden c Department of Molecular medicine and surgery, Karolinska Institutet, Sweden d Harvard Medical School, Boston, Massachusetts, United States e Clinical Research Centre, Örebro University Hospital, Sweden Received 4 May 2006; received in revised form 14 August 2006; accepted 23 August 2006 Available online 6 October 2006

Abstract Background: Earlier research has indicated a positive association between coeliac disease (CD) and some mental disorders. Studies on CD and depression have inconsistent findings and we know of no study of CD and the risk of bipolar disorder (BD). Methods: We used Cox regression to investigate the risk of subsequent mood disorders (MD); depression and BD in 13,776 individuals with CD and 66,815 age- and sex-matched reference individuals in a general population-based cohort study in Sweden. We also studied the association between prior MD and CD through conditional logistic regression. Results: CD was associated with an increased risk of subsequent depression (Hazard ratio (HR) = 1.8; 95% CI = 1.6–2.2; p b 0.001, based on 181 positive events in individuals with CD and 529 positive events in reference individuals). CD was not associated with subsequent BD (HR = 1.1; 95% CI = 0.7–1.7; p = 0.779, based on 22 and 99 positive events). Individuals with prior depression (OR = 2.3; 95% CI = 2.0–2.8; p b 0.001) or prior BD (OR = 1.7; 95% CI = 1.2–2.3; p = 0.001) were at increased risk of a subsequent diagnosis of CD. Limitations: Study participants with CD and MD may have more severe disease than the average patient with these disorders since they were identified through a hospital-based register. Conclusions: CD is positively associated with subsequent depression. The risk increase for CD in individuals with prior depression and BD may be due to screening for CD among those with MD. © 2006 Elsevier B.V. All rights reserved. Keywords: Autoimmune; Bipolar disorder; Coeliac; Cohort study; Depression; Mood disorder; Unipolar disorder

Abbreviations: BD, Bipolar disorder; CD, Coeliac disease; CI, Confidence interval; HR, Hazard ratio; IPR, Inpatient register; MD, Mood disorders; OR, Odds ratio; SEI, Socioeconomic index. ☆ This project (04-030/1) was approved by the Research Ethics Committee of the Karolinska Institute, Stockholm, Sweden on the 18th March 2004. ⁎ Corresponding author. Present address: Department of Paediatrics, Örebro University Hospital, Sweden. Tel.: +46 19 6023439; fax: +46 19 187915. E-mail address: [email protected] (J.F. Ludvigsson). 0165-0327/$ - see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.jad.2006.08.032

118

J.F. Ludvigsson et al. / Journal of Affective Disorders 99 (2007) 117–126

1. Introduction Coeliac disease (CD) affects up to 1% of all individuals in the Western world (Dube et al., 2005). The disease occurs mainly in individuals with HLA-DQ2 or DQ8 (Sollid, 2000) and is triggered by exposure to gluten (Kagnoff, 2005). CD is characterized by mucosal atrophy in the small bowel, but is also associated with extraintestinal complications (Green and Jabri, 2003), including neuropsychiatric disorders (Bushara, 2005). Depression is a major cause of worldwide disability and is projected to become the second most important cause of disability in the Western world in the year 2020, only surpassed by ischemic heart disease (Murray and Lopez, 1997). Goldberg (Goldberg, 1970) and Hallert et al. (Hallert and Astrom, 1982a,b; Hallert and Derefeldt, 1982) were among the first to describe the association of CD with depression. Their papers have since been followed by a number of reports on CD and depression (Accomando et al., 2005; Addolorato et al., 2001, 1996; Carta et al., 2003; Ciacci et al., 2003, 1998; Cicarelli et al., 2003; Fera et al., 2003; Pynnonen et al., 2002; Siniscalchi et al., 2005) with varying results. Bipolar disorder (BD) is another mood disorder (MD). BD is characterized by swings in a person's mood and energy and will often lead to damaged relationship and poor performance at school or work. About 1% of the American population aged 18 years or more have, in any given year, BD (Regier et al., 1993). To our knowledge there are as yet no published studies focusing on CD and BD. The prevalence of depression in CD varies between 6–57% (6.5% (Ciacci et al., 2003), 14% (Cicarelli et al., 2003), 17% (Siniscalchi et al., 2005), 19–24% (Fera et al., 2003), 32% (Ciacci et al., 1998), 42%(lifetime risk) (Carta et al., 2002), 46–57% (Addolorato et al., 2001)), with the lowest prevalence occurring among individuals with gluten-free diet in one of the largest earlier studies (Ciacci et al., 2003). The 36 individuals with CD in the Carta et al. study were at a 2.7-fold increased risk of major depressive disorders (Carta et al., 2002) but also at an increased risk of recurrent brief depression (Carta et al., 2003). Most other studies have shown a statistically significantly positive association between CD and depression (Addolorato et al., 2001, 1996; Ciacci et al., 1998; Hallert and Astrom, 1982a,b; Hallert and Derefeldt, 1982; Siniscalchi et al., 2005), but not all (Accomando et al., 2005; Cicarelli et al., 2003; Fera et al., 2003; Roos et al., 2006). Despite an increased depression score in a modified version of the Zung Self-Rating

Depression Scale, Fera et al found no increased risk of depression as defined by this instrument in individuals with CD (Fera et al., 2003). Neither did two recent Italian studies (Accomando et al., 2005; Cicarelli et al., 2003); or a Swedish study based on psychological general well-being scores in individuals with treated CD (Roos et al., 2006). Several of the previous studies have been based on a limited number of patients with CD (Addolorato et al., 2001, 1996; Carta et al., 2003; Ciacci et al., 2003, 1998; Fera et al., 2003; Hallert and Astrom, 1982a,b; Hallert and Derefeldt, 1982; Siniscalchi et al., 2005), with only four studies including more than 100 patients with CD (Ciacci et al., 2003; Cicarelli et al., 2003; Fera et al., 2003; Siniscalchi et al., 2005) and among the three studies that used controls (Cicarelli et al., 2003; Fera et al., 2003; Siniscalchi et al., 2005), only one found an increased risk of depression in individuals with CD compared with the control population (17% vs. 0%)(Siniscalchi et al., 2005). In several of the studies, depression was identified solely through self-rating questionnaires (Accomando et al., 2005; Addolorato et al., 2001, 2003; Ciacci et al., 1998; Hallert and Astrom, 1982a,b; Siniscalchi et al., 2005), while the involvement of medical professionals in the diagnosis of MD is not clearly outlined in several other papers (Cicarelli et al., 2003; Fera et al., 2003). Most studies on CD and MD have been crosssectional in nature (Accomando et al., 2005; Addolorato et al., 1996; Carta et al., 2003, 2003; Ciacci et al., 1998; Cicarelli et al., 2003; Fera et al., 2003; Goldberg, 1970; Hallert and Astrom, 1982a,b; Hallert and Derefeldt, 1982; Pynnonen et al., 2002; Siniscalchi et al., 2005) and when both pre- and post-CD diagnoses of psychiatric disease have been studied, temporal sequence has not been taken into account when calculating risk estimates (Addolorato et al., 2001; Carta et al., 2002). This study is designed to address some of the methodological limitations of earlier work and to examine differences in association by temporal sequence of CD and MD (depression and BD) using longitudinal data. Swedish national registers were used to conduct this research among over 13,000 individuals with CD and 66,000 individuals without the disease. 2. Methods 2.1. Participants The Swedish National Board of Health identified all individuals with a hospital discharge diagnosis of CD

J.F. Ludvigsson et al. / Journal of Affective Disorders 99 (2007) 117–126 Table 1 International classification codes used in the current paper

Coeliac disease Diabetes mellitus Thyroid disease a Bipolar disorder Depression

ICD-8

ICD-9

ICD-10

269.00; 269.98 250

579A 250

K90.0 E10–14

240–246

240–246

E00–E07

296.10; 296.20; 296.30; 296A; 296C– F30–31 296.88; 296.99 E; 296W 296.00; 300.40 296B; F32–33 300E; 311

a

Does not include congenital thyroid malformations, thyroid cancer, tuberculosis or parasitic disease in the thyroid gland, or trauma against the thyroid gland.

between 1973 and 2003 through the Swedish National Inpatient Register (IPR). The IPR was set up in parts of Sweden in 1964, but psychiatric diagnoses became available throughout the entire country from 1973. The IPR has covered all hospital admissions in Sweden since 1987. Every record in the IPR can be linked with patients through a unique personal identity number assigned to more than 99.9% of all Swedish residents and immigrants (Lunde et al., 1980). Individuals who received any of the international classification of disease (ICD) codes consistent with a diagnosis of CD between 1973 and 2003 (Table 1) were defined as having CD. A total of 15,201 individuals with CD were identified. For each individual with CD, Statistics Sweden used the National Total Population Register to select at random up to five reference individuals matched for age, sex, calendar year and county of residence. The Total Population Register (Johannesson, 2005) includes information on area of residence, vital status, and dates of immigration or emigration. We then used the IPR to identify individuals with MD among the study participants. Depression and BD were defined according to the relevant ICD codes listed in Table 1. The IPR was also used to provide data on the potential confounding factors diabetes mellitus (DM) and thyroid disease (Table 1). The IPR does not distinguish between type 1 and type 2 DM. 2.2. Exclusion criteria Among those with a diagnosis of CD, 95 were excluded due to data irregularities such as death recorded before the first diagnosis of CD or probable

119

misclassification (e.g. an ICD-code that consisted of both a code for MD and another disease code (e.g. 300E17)). We then excluded another 1330 individuals with CD due to a follow-up shorter than 1 year or A diagnosis of MD before study entry (340 received a diagnosis of depression and 71 of BD before or within the first year after diagnosis of CD). We did so in order to minimize the risk of detection bias (the risk that when CD itself is diagnosed or due to associated symptoms, a patient is investigated for a number of potential diagnoses including depression or BD). Similar exclusion criteria were applied to reference individuals. The analyses of the current study were based on 13,776 individuals with CD and 66,815 reference individuals who never had a diagnosis of CD. Both cohorts were free of a diagnosis of MD at the beginning of follow-up and their characteristics are given in Table 1. Table 2 Characteristics of participants (No. (%)) Characteristics

Reference Coeliac disease (%) (no coeliac disease) (%)

Total

66,815

13,776

Age at first recorded diagnosis of coeliac disease (years) 0–15 – 9228 (67.0) ≥16 – 4548 (33.0) Sex Men Women

27,616 (41.3) 39,199 (58.7)

5702 (41.4) 8074 (58.6)

Calendar period 1973–1983 1984–1993 1994–2003

19,849 (29.7) 30,183 (45.2) 16,783 (25.1)

4069 (29.5) 6204 (45.0) 3503 (25.4)

Socioeconomic index I II III Missing data

7086 (10.6) 9046 (13.5) 19,290 (28.9) 31,393 (47.0)

1463 (10.6) 2120 (15.4) 4950 (35.9) 5243 (38.1)

Individuals with more than 1 year of follow-up after coeliac disease diagnosis or corresponding date in matched individuals. See also text. Socioeconomic index: “I” is highest category. See also text. For reference individuals we have given the number of individuals who constituted the basis for the Cox regression. We actually had data on socioeconomic index in another 6047 reference individuals but these individuals were not part of the internally stratified calculations due to missing values on socioeconomic index in the matched individual with coeliac disease. Adding the 6047 reference individuals to those presented above, the proportion of missing values was similar among individuals with coeliac disease and among those without coeliac disease.

120

J.F. Ludvigsson et al. / Journal of Affective Disorders 99 (2007) 117–126

2.3. Socioeconomic index In a subset of individuals (N = 43,955) we had data on socioeconomic index (SEI) (Table 2) based on a threecategory occupational classification from 1968, where “I” denotes a high SEI (Guteland, 1982). Some 6500 of these were children born after 1990 and were too young to have their own SEI, so they were assigned a socioeconomic code on the basis of the occupation of their mother. We used the SEI of the mother and not of the father since in divorced families the children most often live with their mother. 2.4. Statistical analyses Cox regression was used to estimate the risk of subsequent MD in individuals with CD. Risk estimates are given as Hazard Ratios (HRs). The proportional hazards assumption was tested with a log minus log plot for each outcome measure (see Appendix). The follow-up time started 1 year after study entry (date of first recorded diagnosis of CD and the corresponding date in reference individuals) and ended on the date of first discharge diagnosis of MD, date of emigration, death or the end of the study period (31st December 2003), whichever occurred first. The Cox model was internally stratified for age, sex, calendar year and county therefore resembling a conditional logistic regression model, as each individual with CD was only compared with his/her matched reference controls. In separate analyses we stratified for age at first recorded CD diagnosis (≤15 years vs. ≥ 16 years) and sex. In order to increase the

specificity of our outcome we specifically identified individuals who had MD listed as the main/first diagnosis (rather than secondary diagnosis). In additional analyses we adjusted for thyroid disease, DM or SEI. Conditional logistic regression estimated the association of CD (the dependent variable) in individuals with a prior diagnosis of MD. This analysis was performed to assess whether the associations between CD and MD were independent of temporal sequence. The end of follow-up was defined as date of first CD diagnosis and the same date for the matched reference individuals without CD. Those with 1 year or less between the date of first MD diagnosis and CD diagnosis, and corresponding date in reference individuals, were excluded. Risk estimates are given as odds ratios (OR). Statistical significance was defined as 95% confidence intervals for HRs or ORs not including 1.00. Statistics were calculated using SPSS 11.0 (Chicago, Illinois, 2002). 2.5. Power calculations At a significance level of 5%, we had an 80% power to detect an increased risk of subsequent depression in individuals with CD if the HR was 1.3 or above. For subsequent BD we had the power to detect an increased HR of 1.5 or above. 2.6. Ethics This study was approved by the Research Ethics Committee of the Karolinska Institutet. None of the

Table 3 Risk of depression and BD in subgroups of patients with coeliac disease No. participants

First year excluded No coeliac disease 66,815 Any coeliac disease 13,776 Age at first recorded coeliac disease diagnosis 0–15 years 9228 ≥ 16 years 4548 Sex Male 5702 Female 8074 First year included No coeliac disease Any coeliac disease a

72,532 14,798

Depression

Bipolar disorder

No. events

HR; 95% CI

p-value

No. events

HR a, 95% CI

p-value

529 181

1.0 1.8; 1.6–2.2

b0.001

99 22

1.0 1.1; 0.7–1.7

0.779

54 127

1.6; 1.2–2.2 2.0; 1.6–2.4

0.002 b0.001

8 14

1.4; 0.6–3.1 0.9; 0.5–1.7

0.384 0.808

57 124

2.0; 1.4–2.7 1.8; 1.5–2.2

b0.001 b0.001

11 11

1.8; 0.9–3.7 0.7; 0.4–1.4

0.089 0.365

619 214

1.0 2.0; 1.7–2.4

b0.001

112 25

1.0 1.1; 0.7–1.8

0.574

HR = Hazard ratio. Estimates derived from Cox regression internally stratified for sex, age, year of study entry and county.

J.F. Ludvigsson et al. / Journal of Affective Disorders 99 (2007) 117–126

121

participants was contacted. Patient information was anonymised prior to the analyses. 3. Results 3.1. Characteristics of participants The median age at study entry was 2 years in both individuals with CD and those who never had a diagnosis of CD (range = 0–94 years). The first recorded diagnosis of CD was during childhood for the majority of those with CD and a higher proportion of study participants were female (Table 2). The median age at first recorded diagnosis of depression was 50 years (range: 11–87) among individuals with CD and 56 years (range: 11–94) among reference individuals. Corresponding values in BD were as follows: CD: 47 years and range: 13–77; reference individual: 46 years and range: 15–85. The median duration from study entry (date of diagnosis of CD in individuals with CD) to the first recorded diagnosis of depression was 9.6 years in individuals with CD (range: 1.0–29.0) and 12.5 years in reference individuals (1.1–29.8). Corresponding values for BD were as follows: CD: 10.7 years and range: 1.2– 24.1; reference individuals: 8.6 years and range: 1.1– 29.2. There was a diagnostic overlap between the two mood disorders; 33 / 710 (4.6%) individuals with depression and 33 / 121 (27.3%) individuals with BD had received both diagnoses before the end of follow-up.

Fig. 1. Risk of depression.

Fig. 2. Risk of bipolar disorder.

3.2. CD and subsequent depression Individuals with CD were at increased risk of subsequent depression (HR = 1.8; 95% CI = 1.6–2.2; p b 0.001; see also Table 3 and Fig. 1). There was no statistically significant difference in risk estimates according to sex or age at diagnosis of CD (formal interaction test between CD and sex: p = 0.562; between CD and age: p = 0.314). In a subset of individuals we had data on SEI, in these individuals adjustment for SEI did not affect the risk estimates (crude HR = 1.9; adjusted HR = 1.8; both p b 0.001; based on 97 positive events in 8533 individuals with CD and 243 positive events in 35,422 individuals without CD). CD was also positively associated with subsequent depression listed as the main diagnosis (HR = 1.7; 95% CI = 1.4–2.1; p b 0.001; 112 and 342 positive events). When we excluded individuals who had also received a diagnosis of BD before the end of followup, the positive association between CD and subsequent depression remained statistically significant (HR = 1.9; 95% CI = 1.6–2.3; p b 0.001; based on 177 and 500 positive events). Adjustment for DM (adjusted HR for depression = 1.8; 95% CI = 1.5–2.1; p b 0.001) or thyroid disease (adjusted HR for depression = 1.8; 95% CI = 1.5–2.1; p b 0.001) did not affect the risk estimates. When we included the first year after study entry in the follow-up, CD remained a risk factor for subsequent depression (HR = 2.0; 95% CI = 1.7–2.4; see also Table 3).

122

J.F. Ludvigsson et al. / Journal of Affective Disorders 99 (2007) 117–126

3.3. CD and subsequent BD We found no association of CD with subsequent BD (HR = 1.1; 95% CI = 0.7–1.7; p = 0.779; see also Table 3 and Fig. 2). The difference in risk estimates between the sexes (male: HR = 1.8; female: HR = 0.7; see also Table 3) was not statistically significant (formal interaction test between sex and CD: p = 0.065), neither was the difference according to age at diagnosis ( p = 0.296). Adjustment for DM or for thyroid disease did not affect the risk estimates (data not shown); neither did adjustment for SEI in the subset with data on SEI (data not shown). The risk estimates were similar when we included the first year of follow-up (Table 3). CD was not associated with subsequent BD listed as the main diagnosis (data not shown). 3.4. CD and prior history of MD Both prior depression (OR = 2.3; 95% CI = 2.0–2.8; p b 0.001) and prior BD (OR = 1.7; 95% CI = 1.2–2.3; p = 0.001) were associated with an increased risk of a subsequent CD diagnosis. 4. Discussion Using longitudinal data from a national generalpopulation based sample, we found a statistically significant positive association of CD with subsequent depression. This risk increase was also seen when we stratified for age and sex, and when we adjusted for the potential confounders DM, thyroid disease and socioeconomic status. CD was not associated with a subsequent diagnosis of BD. To our knowledge this is the first general population-based cohort study of CD and subsequent mood disorder; and also the first cohort study showing an increased risk of subsequent depression in those who received their diagnosis of CD in childhood. Our study participants were matched for age, sex and calendar year and we used a matched statistical approach. While previous studies have methodological limitations or were based on small numbers, we based our risk estimates on the occurrence of MD in more than 13,000 individuals with CD and over 66,000 reference individuals matched for sex and age. Our study confirms previous cross-sectional reports of a positive association between CD and depression (Addolorato et al., 2001, 1996; Carta et al., 2003; Ciacci et al., 2003, 1998; Hallert and Astrom, 1982a,b; Hallert and Derefeldt, 1982; Siniscalchi et al., 2005). The risk estimate for depression in the current study was slightly lower than that reported by Carta et al. (2002). While

Siniscalchi et al. have suggested that depression becomes more common with age in both individuals with CD and healthy controls (Siniscalchi et al., 2005); Carta et al. reported that the risk of major depressive disorder was greatest in individuals with CD diagnosed in childhood (Carta et al., 2002). In our study, there was no statistically different risk for depression according to age at diagnosis of CD. What makes the increased risk of subsequent depression in the current study particularly interesting is that it contrasts with the lack of positive association of CD with subsequent BD. Our study therefore supports earlier evidence of partly different risk factors for depression and BD (Sadock and Sadock, 2002). The association of CD with subsequent depression but not BD indicates a disease process specific to depression, rather than a spurious association resulting from detection bias. Further evidence that the association of CD with subsequent depression is not due to detection bias is the limited increase in HR when the first year of follow-up after CD diagnosis is included in the analysis. In contrast with CD and subsequent MD, both depression and BD were positively associated with a subsequent diagnosis of CD. These associations, at least in part, are more likely to be due to detection bias: patients admitted to hospital with a diagnosis of depression or BD may undergo investigations and more frequent hospital admissions, increasing the probability that a CD diagnosis is made or recorded in association with an admission. Individuals with MD (Charney et al., 2003) are also at an increased risk of diagnosed co-morbidity. In a study of 30,121 adult psychiatric patients hospitalized in Maryland (US) 1998, 19% of elderly patients with BD and 25% of elderly patients with major depressive disorder had at least seven other medical diagnoses (Brown, 2001) (5% of non-elderly 19–64 years old in both groups had at least seven other medical diagnoses). It is also important to note that MD is sometimes the only sign of undiagnosed CD. A number of mechanisms have been suggested to explain the association between CD and depression. These include a low quality of life after diagnosis of CD (Hallert et al., 1998) for reasons such as an unsatisfactory sex life (Ciacci et al., 2003), and difficulty in adapting to the chronic nature of CD (Fera et al., 2003). Malnutrition related to the gastrointestinal disease itself has also been suggested as a risk factor (Addolorato et al., 1996). Asymptomatic CD is difficult to diagnose. But also when the disease is symptomatic is there often a long delay between symptomatic onset to diagnosis of CD (Dickey and McConnell, 1996); this indicates that many patients with a psychiatric

J.F. Ludvigsson et al. / Journal of Affective Disorders 99 (2007) 117–126

diagnosis prior to the diagnosis of CD might have had ongoing inflammation when the psychiatric diagnosis was made. Goldberg suggested in 1970 that psychiatric abnormalities seen in idiopathic steatorrhea could be caused by malabsorption of folic acid and vitamin B12 (Goldberg, 1970). Untreated CD is a risk factor for hyperhomocysteinemia (relative risk = 3.4) and folate deficiency (relative risk = 5.1)(Saibeni et al., 2005) and levels of folate and homocysteine are associated with histologic grading of the small-bowel lesion (Saibeni et al., 2005). The recent study by Saibeni et al referred to newly diagnosed CD (Saibeni et al., 2005), but vitamin deficiency and increased levels of homocysteine also occur many years after diagnosis (Hallert et al., 2002). One reason for this may be ongoing low-grade inflammation even after diagnosis of CD (Lee et al., 2003). Folate and homocysteine may play important roles in neurodegenerative and psychiatric disease (Bottiglieri, 2005), and have been implicated in the pathogenesis of depression (Bjelland et al., 2003, 2005; Carney, 1967; Tiemeier et al., 2002). Additional evidence of the role of folate in depression is presented in the four double-blinded trials reviewed by Bottiglieri (2005). In contrast, research on folate levels with BD is inconsistent (Carney et al., 1990; Hasanah et al., 1997; McKeon et al., 1991). If as Carney et al. have suggested (Carney et al., 1990), red cell folate levels are significantly lower in depression than in BD; this could be one explanation for the association with subsequent depression but not with BD. Another potential link between depression and CD can be sought in tryptophan levels (Bell et al., 2005; Russo et al., 2003). Individuals with undiagnosed or diagnosed CD often suffer from low tryptophan levels (Hernanz and Polanco, 1991). Free tryptophan levels seem to be lower in adolescents with untreated CD and depression than in untreated CD without depression (Pynnonen et al., 2002). Tryptophan is metabolized into 5-hydroxytryptophan that is a precursor of serotonin. In 1982 demonstrated low levels of serotonin in the cerebrospinal fluid, possibly mediating the increased risk of depression seen in CD (Hallert et al., 1982b). We do not know of any study of low tryptophan levels and risk of BD (Bell et al., 2005), except that in a pilot study of lithium-remitted patients with BD, tryptophan depletion caused transient worsening of symptoms (Cappiello et al., 1997). Both malnutrition and active inflammation, potentially explaining the positive association between CD and depression, are likely to be more common in individuals with CD who have low dietary compliance. Unfortunately the IPR contains no information on diet-

123

ary compliance, CD serology, the occurrence of small bowel biopsies or other signs of CD activity. While depression is associated with a diagnosis of CD, it is notable that Roos et al. found no increased risk of depressed mood in treated CD when there was objective evidence of remission (return of villous structure at repeat biopsy or absence of IgA endomysial/gliadin autoantibodies after the institution of gluten-free diet) (Roos et al., 2006). As this indicates that depression risk is not notably elevated when patients are in remission due to good dietary compliance, it is reasonable to assume that dietary compliance influences the risk of depression in individuals with CD. Previous research on depression and CD has often been based on rating scales and personality inventories (Accomando et al., 2005; Addolorato et al., 2001; Ciacci et al., 2003, 1998; Hallert and Astrom, 1982a, b; Siniscalchi et al., 2005). In contrast, our outcome measures were diagnosed in a hospital setting by a medical doctor. In this context, the MD has had important clinical implications, as it had often been the main reason for hospitalization indicating relatively severe MD. This means that mild MD is likely to have been underestimated in this study. Also CD was identified through a hospital-based register, and we may not have identified all individuals with CD. Despite the potential under-ascertainment of CD, the current study was based on more than 13,000 individuals with CD. The high number of participants gave this study large statistical power and allowed for meaningful stratifications. There is a risk that individuals with a hospital-based diagnosis of CD had more severe CD than the average patient with CD. However, in the first part of the study period, hospitalization was common as part of the investigative procedures including small-bowel biopsy. Hospital admission may also have been common for small children throughout the entire study period. Also when we adjusted our data for two potential confounders signalling increased autoimmunity (DM and thyroid disease), one of which is also an important cause of hospitalization in individuals with CD (DM), this did not alter notably the association between CD and depression. The advantage of identifying individuals with CD through a hospital-based register is that earlier validation has shown a high specificity for chronic diseases listed in the IPR (Smedby et al recently showed that the specificity for CD in individuals with concomitant lymphoma was above 85%(Smedby et al., 2005)). Diagnostic criteria for CD have changed since 1973 (ESPGHAN, 1990) and so has access to CD serology. Although small bowel biopsy has always

124

J.F. Ludvigsson et al. / Journal of Affective Disorders 99 (2007) 117–126

been mandatory prior to a diagnosis of CD in Sweden, we cannot exclude that individuals with CD diagnosed in the early part of the study period differ from those in the later part. We did however stratify for calendar period and so slight changes in diagnostic criteria should be no major concern of this study. Both depression and BD have previously been studied in the IPR (Osby et al., 2001). Although we know of no validation of these disorders in the Swedish Inpatient Register, a validation of another psychiatric diagnosis, schizophrenia found that 86% of individuals with an early diagnosis of schizophrenia met the DSM-IV-criteria for that disease (Dalman et al., 2002). In conclusion, our study found a statistically significant positive association of CD with subsequent depression but not with subsequent bipolar disease. We suggest this is due to malnutrition or active bowel inflammation in individuals with CD; and although this study contains no information on dietary compliance it is reasonable to assume that good dietary compliance in such individuals will influence overall CD remission and the risk of depression. Both mood disorders were associated with subsequent CD, possibly due, at least in part to screening bias. Diagnosis and treatment of depression in individuals with CD is important since depression may itself result in lower dietary compliance (Addolorato et al., 2004).

Box 1

Clinical relevance Coeliac disease (CD) affects up to 1% of the western population. This study found a positive relationship between CD and subsequent depression but not with subsequent bipolar disorder. • The positive association of CD with depression may in part be explained by malnutrition and ongoing inflammation. • The risk increase for CD in individuals with prior depression and bipolar disorder may be due to screening for CD among those with mood disorder. •

Acknowledgement JFL was supported by grants from the Örebro University Hospital and the Swedish Medical research Council while writing this article. This project was supported by The Swedish Society of Medicine, the Sven Jerring Foundation, the Juhlin Foundation, the Clas Groschinsky Foundation, the Karolinska Institute Funds, the Örebro Society of Medicine, the Majblomman Foundation and the Swedish Coeliac Society.

Appendix A The proportional hazards assumption was tested with log minus log plots. Parallel lines indicate that the proportional hazards assumption is not violated. NB. Y-axis scales are not identical.

J.F. Ludvigsson et al. / Journal of Affective Disorders 99 (2007) 117–126

References Accomando, S., Fragapane, M.L., Montaperto, D., Trizzino, A., Amato, G.M., Calderone, F., Accomando, I., 2005. Coeliac disease and depression: two related entities? Dig. Liver Dis. 37 (4), 298–299. Addolorato, G., Stefanini, G.F., Capristo, E., Caputo, F., Gasbarrini, A., Gasbarrini, G., 1996. Anxiety and depression in adult untreated celiac subjects and in patients affected by inflammatory bowel disease: a personality “trait” or a reactive illness? Hepatogastroenterology 43 (12), 1513–1517. Addolorato, G., Capristo, E., Ghittoni, G., Valeri, C., Masciana, R., Ancona, C., Gasbarrini, G., 2001. Anxiety but not depression decreases in coeliac patients after one-year gluten-free diet: a longitudinal study. Scand. J. Gastroenterol. 36 (5), 502–506. Addolorato, G., De Lorenzi, G., Abenavoli, L., Leggio, L., Capristo, E., Gasbarrini, G., 2004. Psychological support counselling improves gluten-free diet compliance in coeliac patients with affective disorders. Aliment. Pharmacol. Ther. 20 (7), 777–782. Bell, C.J., Hood, S.D., Nutt, D.J., 2005. Acute tryptophan depletion. Part II: clinical effects and implications. Aust. N. Z. J. Psychiatry 39 (7), 565–574. Bjelland, I., Tell, G.S., Vollset, S.E., Refsum, H., Ueland, P.M., 2003. Folate, vitamin B12, homocysteine, and the MTHFR 677C-NT polymorphism in anxiety and depression: the Hordaland Homocysteine Study. Arch. Gen. Psychiatry 60 (6), 618–626. Bottiglieri, T., 2005. Homocysteine and folate metabolism in depression. Prog. Neuropsychopharmacol. Biol. Psychiatry 29 (7), 1103–1112. Brown, S.L., 2001. Variations in utilization and cost of inpatient psychiatric services among adults in Maryland. Psychiatr. Serv. 52 (6), 841–843. Bushara, K.O., 2005. Neurologic presentation of celiac disease. Gastroenterology 128 (4 Suppl 1), S92–S97. Cappiello, A., Sernyak, M.J., Malison, R.T., McDougle, C.J., Heninger, G.R., Price, L.H., 1997. Effects of acute tryptophan depletion in lithium-remitted manic patients: a pilot study. Biol. Psychiatry 42 (11), 1076–1078. Carney, M.W., 1967. Serum folate values in 423 psychiatric patients. Br. Med. J. 4 (578), 512–516. Carney, M.W., Chary, T.K., Laundy, M., Bottiglieri, T., Chanarin, I., Reynolds, E.H., Toone, B., 1990. Red cell folate concentrations in psychiatric patients. J. Affect. Disord. 19 (3), 207–213. Carta, M.G., Hardoy, M.C., Boi, M.F., Mariotti, S., Carpiniello, B., Usai, P., 2002. Association between panic disorder, major depressive disorder and celiac disease: a possible role of thyroid autoimmunity. J. Psychosom. Res. 53 (3), 789–793. Carta, M.G., Hardoy, M.C., Usai, P., Carpiniello, B., Angst, J., 2003. Recurrent brief depression in celiac disease. J. Psychosom. Res. 55 (6), 573–574. Charney, D.S., Reynolds III, C.F., Lewis, L., Lebowitz, B.D., Sunderland, T., Alexopoulos, G.S., Blazer, D.G., Katz, I.R., Meyers, B.S., Arean, P.A., Borson, S., Brown, C., Bruce, M.L., Callahan, C.M., Charlson, M.E., Conwell, Y., Cuthbert, B.N., Devanand, D.P., Gibson, M.J., Gottlieb, G.L., Krishnan, K.R., Laden, S.K., Lyketsos, C.G., Mulsant, B.H., Niederehe, G., Olin, J.T., Oslin, D.W., Pearson, J., Persky, T., Pollock, B.G., Raetzman, S., Reynolds, M., Salzman, C., Schulz, R., Schwenk, T.L., Scolnick, E., Unutzer, J., Weissman, M.M., Young, R.C., 2003. Depression and Bipolar Support Alliance consensus statement on

125

the unmet needs in diagnosis and treatment of mood disorders in late life. Arch. Gen. Psychiatry 60 (7), 664–672. Ciacci, C., Iavarone, A., Mazzacca, G., De Rosa, A., 1998. Depressive symptoms in adult coeliac disease. Scand. J. Gastroenterol. 33 (3), 247–250. Ciacci, C., D'Agate, C., De Rosa, A., Franzese, C., Errichiello, S., Gasperi, V., Pardi, A., Quagliata, D., Visentini, S., Greco, L., 2003. Self-rated quality of life in celiac disease. Dig. Dis. Sci. 48 (11), 2216–2220. Cicarelli, G., Della Rocca, G., Amboni, M., Ciacci, C., Mazzacca, G., Filla, A., Barone, P., 2003. Clinical and neurological abnormalities in adult celiac disease. Neurol. Sci. 24 (5), 311–317. Dalman, C., Broms, J., Cullberg, J., Allebeck, P., 2002. Young cases of schizophrenia identified in a national inpatient register—are the diagnoses valid? Soc. Psychiatry Psychiatr. Epidemiol. 37 (11), 527–531. Dickey, W., McConnell, J.B., 1996. How many hospital visits does it take before celiac sprue is diagnosed? J. Clin. Gastroenterol. 23 (1), 21–23 (see comments). Dube, C., Rostom, A., Sy, R., Cranney, A., Saloojee, N., Garritty, C., Sampson, M., Zhang, L., Yazdi, F., Mamaladze, V., Pan, I., Macneil, J., Mack, D., Patel, D., Moher, D., 2005. The prevalence of celiac disease in average-risk and at-risk Western European populations: a systematic review. Gastroenterology 128 (4 Suppl 1), S57–S67. ESPGHAN, 1990. Revised criteria for diagnosis of coeliac disease. Report of Working Group of European Society of Paediatric Gastroenterology and Nutrition. Arch. Dis. Child. 65 (8), 909–911. Fera, T., Cascio, B., Angelini, G., Martini, S., Guidetti, C.S., 2003. Affective disorders and quality of life in adult coeliac disease patients on a gluten-free diet. Eur. J. Gastroenterol. Hepatol. 15 (12), 1287–1292. Goldberg, D., 1970. A psychiatric study of patients with diseases of the small intestine. Gut 11 (6), 459–465. Green, P.H., Jabri, B., 2003. Coeliac disease. Lancet 362 (9381), 383–391. Guteland, G., 1982. Socioekonomisk indelning (SEI). Swedish socioeconomic classification. 1982:4. Reports on Statistical Coordination. Statistics Sweden (SCB— Statistiska centralbyrån), Stockholm. Hallert, C., Astrom, J., 1982a. Psychic disturbances in adult coeliac disease. II. Psychological findings. Scand. J. Gastroenterol. 17 (1), 21–24. Hallert, C., Astrom, J., Sedvall, G., 1982b. Psychic disturbances in adult coeliac disease. III. Reduced central monoamine metabolism and signs of depression. Scand. J. Gastroenterol. 17 (1), 25–28. Hallert, C., Derefeldt, T., 1982. Psychic disturbances in adult coeliac disease. I. Clinical observations. Scand. J. Gastroenterol. 17 (1), 17–19. Hallert, C., Granno, C., Grant, C., Hulten, S., Midhagen, G., Strom, M., Svensson, H., Valdimarsson, T., Wickstrom, T., 1998. Quality of life of adult coeliac patients treated for 10 years. Scand. J. Gastroenterol. 33 (9), 933–938. Hallert, C., Grant, C., Grehn, S., Granno, C., Hulten, S., Midhagen, G., Strom, M., Svensson, H., Valdimarsson, T., 2002. Evidence of poor vitamin status in coeliac patients on a gluten-free diet for 10 years. Aliment. Pharmacol. Ther. 16 (7), 1333–1339. Hasanah, C.I., Khan, U.A., Musalmah, M., Razali, S.M., 1997. Reduced red-cell folate in mania. J. Affect. Disord. 46 (2), 95–99. Hernanz, A., Polanco, I., 1991. Plasma precursor amino acids of central nervous system monoamines in children with coeliac disease. Gut 32 (12), 1478–1481.

126

J.F. Ludvigsson et al. / Journal of Affective Disorders 99 (2007) 117–126

Johannesson, I., 2005. The Total Population Register of Statistics Sweden. New Possibilities and Better Quality. Örebro. Kagnoff, M.F., 2005. Overview and pathogenesis of celiac disease. Gastroenterology 128 (4 Suppl 1), S10–S18. Lee, S.K., Lo, W., Memeo, L., Rotterdam, H., Green, P.H., 2003. Duodenal histology in patients with celiac disease after treatment with a gluten-free diet. Gastrointest. Endosc. 57 (2), 187–191. Lunde, A.S., Lundeborg, S., Lettenstrom, G.S., Thygesen, L., Huebner, J., 1980. The person-number systems of Sweden, Norway, Denmark, and Israel. Vital Health Stat 2 2(84),1–59. McKeon, P., Shelley, R., O'Regan, S., O'Broin, J., 1991. Serum and red cell folate and affective morbidity in lithium prophylaxis. Acta Psychiatr. Scand. 83 (3), 199–201. Murray, C.J., Lopez, A.D., 1997. Alternative projections of mortality and disability by cause 1990–2020: Global Burden of Disease Study. Lancet 349 (9064), 1498–1504. Osby, U., Brandt, L., Correia, N., Ekbom, A., Sparen, P., 2001. Excess mortality in bipolar and unipolar disorder in Sweden. Arch. Gen. Psychiatry 58 (9), 844–850. Pynnonen, P.A., Isometsa, E.T., Verkasalo, M.A., Savilahti, E., Aalberg, V.A., 2002. Untreated celiac disease and development of mental disorders in children and adolescents. Psychosomatics 43 (4), 331–334. Regier, D.A., Narrow, W.E., Rae, D.S., Manderscheid, R.W., Locke, B.Z., Goodwin, F.K., 1993. The de facto US mental and addictive disorders service system. Epidemiologic catchment area prospective 1-year prevalence rates of disorders and services. Arch. Gen. Psychiatry 50 (2), 85–94. Roos, S., Karner, A., Hallert, C., 2006. Psychological well-being of adult coeliac patients treated for 10 years. Dig. Liver Dis. 38 (3), 177–180.

Russo, S., Kema, I.P., Fokkema, M.R., Boon, J.C., Willemse, P.H., de Vries, E.G., den Boer, J.A., Korf, J., 2003. Tryptophan as a link between psychopathology and somatic states. Psychosom. Med. 65 (4), 665–671. Sadock, B.J., Sadock, V.A., 2002. Kaplan and Sadock's Synopsis of Psychiatry: Behavioral Sciences/Clinical Psychiatry. Lippincott Williams & Wilkins. Saibeni, S., Lecchi, A., Meucci, G., Cattaneo, M., Tagliabue, L., Rondonotti, E., Formenti, S., De Franchis, R., Vecchi, M., 2005. Prevalence of hyperhomocysteinemia in adult gluten-sensitive enteropathy at diagnosis: role of B12, folate, and genetics. Clin. Gastroenterol. Hepatol. 3 (6), 574–580. Siniscalchi, M., Iovino, P., Tortora, R., Forestiero, S., Somma, A., Capuano, L., Franzese, M.D., Sabbatini, F., Ciacci, C., 2005. Fatigue in adult coeliac disease. Aliment. Pharmacol. Ther. 22 (5), 489–494. Smedby, K.E., Akerman, M., Hildebrand, H., Glimelius, B., Ekbom, A., Askling, J., 2005. Malignant lymphomas in coeliac disease: evidence of increased risks for lymphoma types other than enteropathy-type T cell lymphoma. Gut 54 (1), 54–59. Sollid, L.M., 2000. Molecular basis of celiac disease. Annu. Rev. Immunol. 18, 53–81. Tiemeier, H., van Tuijl, H.R., Hofman, A., Meijer, J., Kiliaan, A.J., Breteler, M.M., 2002. Vitamin B12, folate, and homocysteine in depression: the Rotterdam Study. Am. J. Psychiatry 159 (12), 2099–2101.