Introduction
The association between mental health and common infection agents has been an increasing area of investigation. The specific infections of interest particularly include Toxoplasma gondii (T. gondii) and several infections of the herpesvirus family, such as Cytomegalovirus (CMV), Epstein–Barr virus (EBV), and Herpes Simplex virus Type 1 (HSV-1), all infections that primarily often are asymptomatic or cause only mild symptoms, but form a lifelong latent infection and can later reactivate. Immunoglobulin G (IgG) antibodies to these infections indicate lifetime infection, whereas immunoglobulin M (IgM) antibodies may arise during primary infections or reactivation of infection.
Toxoplasma gondii is an intracellular parasite, most commonly spread by food or water that is contaminated with oocysts shed by cats, or by eating meat-containing tissue cysts, whereas the herpes viruses spread mostly through bodily fluids, and are among the most common viruses in humans. In recent studies of the general population in Finland, the seropositivity prevalence in adults was 20% for
T. gondii [
1], 72% for HSV-1 [
2], 84% for CMV, and 98% for EBV [
3]. Associations between these infections and mental health have been repeatedly found. Toxoplasma may be associated with several mental disorders [
4‐
7]. We have previously published cross-sectional results on the association between IgG antibodies against
T. gondii and anxiety disorders and depressive and psychotic-like symptoms [
1,
8]. Furthermore, CMV has been linked to psychiatric disorders [
9‐
12]. Interestingly, in one study, CMV seropositivity influenced mood disorders differentially among males and females, associating with carrying a
lower risk of mood disorders in males [
13]. Also, the association between mental health and HSV-1 and EBV [
10,
11,
13‐
21] has been studied, with somewhat conflicting results.
In addition to mood and emotions, inflammation and infections can also affect behavior [
22,
23]. The association between exposure to infectious agents and suicidal behavior has recently raised interest, both in population and clinical studies. Comparing 20 European nations, the prevalence of
T. gondii was positively correlated with national suicide rates [
24], especially in women of postmenopausal age [
25]. In a birth-cohort, there was a trend between toxoplasma IgG antibodies and suicide attempts [
26] and in a large population cohort, toxoplasma infection was found to be associated with risk of self-directed violence in women [
27]. Toxoplasma serointensity has also been associated with suicide attempts in psychiatric outpatients [
28] and individuals with mood disorders [
29]. However, conflicting results have also been published in schizophrenia patients [
30,
31].
In individuals with serious mental illness, those having a lifetime suicide attempt had elevated levels of IgM class antibodies to both
T. gondii and CMV, and also an association between the levels of these antibodies and the number of suicide attempts was found, whereas IgG class or EBV antibodies were not associated with suicide attempts [
32]. In participants with schizophrenia or mood disorders, CMV antibodies were associated with risk of suicide, while HSV-1 and EBV antibodies were not [
33]. In another study in the general population, HSV-1 infection was associated with suicidal behavior [
34]. A few other studies have not found evidence of association between seropositivity for CMV or HSV-1 and nonfatal suicidal self-directed violence [
35] or history of suicide attempt in schizophrenia [
36].
One pathway by which infections may influence mental health is inflammation occurring in the central nervous system. The relationship between the inflammatory marker C-reactive protein (CRP) and the studied infections is partly unknown. Higher CRP levels, measuring inflammation, have been reported to be associated with
T. gondii seropositivity in our previous study [
1] and other studies [
37,
38] and the role of CRP has also been studied in the context of HSV-1, CMV, and EBV [
34,
39,
40]. Inflammation may also trigger changes in affective and behavioral modulation, and inflammatory processes may have a role in suicidality [
41].
Information on the possible association on common infections and suicidal outcomes may help prevent self-harm in the population, as the infections may be prevented and treated. In this study, we aimed to investigate whether toxoplasma and herpes infections have a role in the risk for death by suicide or intentional self-harm in the general population. For this purpose, the participants of the Health 2000 study, a large Finnish population survey, were followed-up longitudinally in national health registers for 15 years. As depression is a major risk factor for suicidality [
42], we investigated whether being seropositive to the studied infectious agents at baseline was associated with a) baseline depressive symptoms, and b) death by suicide and self-harm during the follow-up.
In addition, we assessed suicidal ideation and suicide attempts retrospectively in a smaller subsample of participants of the Health 2000 study, Psychoses in Finland, selected for possible psychotic symptoms or as controls, and who had been comprehensively assessed using medical records and interview data [
43].
Discussion
First, we found no significant cross-sectional associations between baseline depressive symptoms and antibody levels to the herpes viruses EBV, CMV, or HSV-1 when controlling for background variables associated with infection seropositivity. We have previously reported an association between
T. gondii and depressive symptoms in this sample, even when adjusting for age, gender, region of residence, education, marital status, cat ownership, 12-month diagnoses, CRP, and antidepressant use [
1]. Some previous studies have found CMV but not HSV-1 to be linked with depression [
11,
14,
50], but the association between exposure to herpes infections and depressive symptoms has not been extensively studied.
Second, in a general population cohort of 6250 participants and 15-year follow-up using comprehensive register data and complete follow-up, seropositivity or antibody levels of T. gondii or herpes viruses were not found to be associated with suicides or diagnoses of intentional self-harm.
Third, we assessed suicidal ideation and suicide attempts in a subsample using information from medical records as well as data collected by questionnaire and interview. The participants of this subsample had either a severe psychiatric disorder or were controls; we controlled for this screen status in the regression models. A total of 31% of this subsample had had suicidal thoughts and 13% had attempted suicide during their lifetime. EBV antibody level was associated with a history of a suicide attempt in males, and none of the other infection variables added to the association with the suicide measures. However, those seropositive for IgG class antibodies for CMV, measuring latent infection, had fewer suicide attempts compared to those who were seronegative. This result was especially significant among male participants and also high CMV serointensity showed the same protective effect. The association was specific to CMV, as antibodies to the other studied herpes infections did not show the same relationship, and the association with EBV was in the opposite direction.
Converging evidence suggests that some common infectious agents may predispose to mental disorders and disrupt affective and behavioral modulation [
6,
12,
27,
51], hence also possibly elevating risk of self-harm. Only few prospective studies have been conducted [
27,
33] and there are differences between study design, seropositivity cut-offs used, populations studied, variables controlled for, and the timelines, which can all cause discrepancies between studies. Most of the previous studies have been conducted in clinical samples, only few studies concentrating on suicidality in the general population [
26,
27]. Previous studies have reported self-harm to be associated with toxoplasma infection in the population especially in women [
25,
27]. In psychiatric samples, antibodies for toxoplasma have been positively associated with suicidality [
28,
29,
32,
36,
52], however, not all studies have found such association [
30,
31]. The differences between the current study and a similarly prospective study by Pedersen et al. [
27] include their larger sample and a broader definition of the suicidal outcomes. The current study found no association between
T. gondii antibodies and the suicidal outcomes of interest when controlling for the confounding variables, although
T. gondii was associated with depressive symptoms [
1].
The association between herpes viruses and suicidality has been studied scarcely [
32‐
36] and mostly among psychiatric samples. In previous works, elevated CMV antibodies have been associated with suicide attempts or death by suicide in some psychiatric samples [
32,
33] but not in all [
36]. In a prospective study where elevated levels of CMV antibodies predicted suicide [
33], the participants were mostly schizophrenia or bipolar disorder patients. In our study, we had the reverse result of lower CMV antibodies associating with multiple suicide attempts, but the association was not significant among those with a psychotic disorder. In another previous work, IgG class antibodies against HSV-1 were associated with attempting or committing suicide in the general population [
34]. A few previous studies have reported negative findings on the association between suicidality and EBV or HSV-1 [
32,
33,
35,
36]; to our knowledge, our finding that higher EBV antibodies were associated with risk of a lifetime suicide attempt is new. In a previous longitudinal study, EBV antibody levels were stable at the individual level but there was strong variation among individuals [
53]. As overall EBV antibody level was measured here, the results could look different if we would measure specific EBV proteins, as was done by Dickerson and colleagues [
20].
Although the prevalence of CMV in Finland has decreased significantly during recent decades [
54], the rate was as high as 84% in the population at our baseline in year 2000 [
3]. The increased odds of multiple suicide attempts in persons belonging to the CMV seronegative minority is a new finding. In line with this was our post hoc result of higher CMV antibodies predicting lower risk of self-harm in the register follow-up among those seropositive for CMV. In the same sample as used in the current study, we have also found CMV to protect from new-onset generalized anxiety disorder, but not from new-onset depressive disorders or other anxiety disorders [
55]. Low levels of CMV antibodies in affective disorders have also been found in other studies (Yolken et al., unpublished results). One previous study found that CMV seropositivity associated with higher risk of mood disorders in females, but—in line with our results—with a lower risk in males [
13]. The association between the immune system and mental health may be different in females and in males [
13,
56,
57], although the reasons for this are still unclear, leading to a question whether gender differences may exist in the effect of infections on suicidality. As transmission of CMV requires intimate contact with other people, one could speculate whether persons seronegative for CMV have personality factors that predispose to suicidality as well, such as neuroticism [
58] or social isolation. However, HSV-1 and EBV have similar modes of person-to-person transmission as CMV, and our result was specific to CMV.
Attempting and dying by suicide may be associated with infections in different ways. Furthermore, we found that CMV seronegativity only added to the risk for several attempts but not to the risk for one attempt, when compared to no suicide attempts. Those attempting suicide once or several times can be clinically different populations, and the risk factors for single or repeated suicide attempts can be different [
59].
Lower socioeconomic status is associated with seropositivity for CMV [
60,
61]. In the general population, suicide risk is also associated with lower socioeconomic status [
62,
63]. Therefore, it seems surprising that antibodies for CMV were associated with lowered suicide risk. However, among patients treated for major depressive disorder, the association between socioeconomic status and suicidality has been the opposite in Finland: higher educational level and family income predict suicide mortality [
64]. In another Finnish study investigating people with depression-based disability retirement, high socioeconomic position did not protect against unnatural and alcohol-related deaths, unlike in the general population [
65]. Our result of the protective role of CMV was significant especially among those with mood disorders. It was not significant when only looking at those with psychotic disorder in the PIF subsample, and suicide attempts were not associated with educational level among those with a psychotic disorder in the PIF study [
43]. In other words, the interplay among socioeconomic status, psychiatric symptoms, and infections in regards to suicide risk seems to be complex.
Another factor to consider is general inflammation. Inflammatory processes may be linked to suicidality, and increased CRP levels are associated with mortality risk in people with mental disorders [
66,
67]. The associations found in the PIF subsample remained when adjusting for CRP, so they were not explained by inflammation. Another reason why general activation of the immune system is not likely to explain our findings was that they were specific to certain infections.
Strengths and weaknesses
A number of limitations should be kept in mind when evaluating the current results. Two samples were used in the study, both having their strengths and weaknesses. In the large general population of sample representative of the whole adult population in Finland, we were able to prospectively investigate the role of serological factors in suicide deaths and self-harm. Suicide deaths were reliably defined in the Causes of Death statistics, although the rareness of the outcome severely limited the statistical power of the analyses. Power calculations show that with only 18 cases, to have a 60% power, a HR as high as 3–4 would be needed. Only the most severe forms of intentional self-harm were captured in the health care register follow-up, and cases of self-harm not resulting in medical care or not diagnosed as self-harm were missed. Furthermore, IgM antibodies were only available for cases and matched controls and for the PIF subsample, while the whole sample could be used when investigating IgG levels. Furthermore, IgM levels informed whether the infection had occurred close to the baseline assessment, but we did not have information on IgM levels at the time of self-harm or suicide.
Using the smaller PIF subsample, we were able to assess suicidal ideation and suicide attempts more reliably, as conclusive, retrospective information from self-report, medical records, and health care registers was available. Persons with severe mental illness were enriched in this subsample, limiting the generalizability of these results to the general population, which is why we controlled for screen status and also looked at this subsample dividing it based on diagnosis group. The subsample was screened using multiple sources of information and included psychosis patients and individuals with any suspicion of psychotic illness as well as about 20% of matched healthy controls. One of the screens was disability pension based on a severe mental disorder (often depressive disorder) and suicidality was thus common in the subsample. In the subsample, lifetime suicidality was assessed, so the suicidal thoughts or acts may have happened before the blood sample was taken in year 2000. These kinds of differences in study design might explain some of the discrepancies noted among past studies. Genders were investigated separately in post hoc analyses but the small sample sizes limit the value of these findings, especially in females having small cell numbers in some analyses.
In analyzing the associations between infections and self-harm, we adjusted for various background variables related to the measured infections, including demographic factors and inflammation. We did not assess for all the factors that could contribute to suicide such as trauma history, psychiatric symptoms, substance use, or personality factors [
68]. Finally, multiple comparisons were not corrected for in this exploratory study. If the false discovery rate had been controlled for using the Benjamini–Hochberg procedure [
69], the associations would have not remained significant.
Conclusions
In a large sample representing the whole Finnish adult population, antibodies to CMV, EBV, or HSV-1 were not associated with depressive symptoms. Antibodies to T. gondii or the herpes viruses were not associated with heightened risk for subsequent suicide deaths or diagnoses of intentional self-harm in a 15-year register follow-up. In a subsample consisting mostly of participants with severe mental disorders, the infections were not associated with a heightened risk for suicidal thoughts or acts. However, the finding of heightened suicidality risk among persons not infected with CMV calls for further research.