Evidence showing that infectious diseases in childhood can play a crucial role in the etiopathogenesis of mental disorders is growing (1). This relation has been a long-standing issue in psychiatric epidemiology, but research activity has recently intensified. Current topics include prenatal infections with Toxoplasma gondii, rubella, and other infectious diseases (2, 3) as risk factors for neurodevelopmental disorders and schizophrenia. Furthermore, a series of postnatal infectious agents was reported in association with schizophrenia/psychosis (4–6). The pediatric autoimmune neuropsychiatric disorders associated with streptococcal infections (PANDAS) model also postulates a crucial role for autoantibodies against basal ganglia tissue appearing after infections with group A streptococci in the development of attention deficit hyperactivity disorder, Gilles de la Tourette syndrome, and obsessive–compulsive disorder (OCD) (7, 8). Finally, recent analyses have shed new light on the role of infectious diseases in anxiety disorders (9).

On a formal level, there are four models of how infectious diseases could affect the risk for mental disorders. The models can be illustrated using the associations between Epstein–Barr virus (EBV) and multiple sclerosis (MS) or psychotic disorders. The first model relates to the mere occurrence of an infection. For example, an EBV infection is a necessary precondition for later MS onset: while the seroprevalence for EBV is high, i.e., it covers 90% of the adult population or more (10), almost no one from the negative seroprevalence group develops MS. The second model includes the severity of an infection, for example, due to a dysfunctional immune response and/or due to a combination with genetic or other types of vulnerabilities. The third model considers the timing of childhood infections. Deviations from typical timing, in particular delayed childhood infectious diseases such as chickenpox, measles, mumps, or rubella, are well known to lead to a more severe disease course with more complications (11–14). This also applies to delayed EBV infections, which increase the risk for mononucleosis and probably also for MS (15). Finally, the fourth model focuses on the sequence of childhood infections. This is a possible reason for why infectious disease occurring earlier in life than typically expected might have harmful effects. For example, EBV infections in the first years of life have been linked to psychotic experiences (16). Both the timing and sequence of childhood infections can be assumed to be subjected to evolutionary optimization.

There is a lack of research systematically examining how the timing of infections and other inflammatory insults (17) impacts the risk for mental disorders. This study focused on the timing of childhood diseases with regard to early-onset anxiety disorders (separation anxiety disorder, overanxious disorder, specific phobias, and social phobia). The aim of the study was to examine whether deviations in age of onset of childhood infectious diseases are associated with the early-onset anxiety disorders. We used a two-step approach. First, we examined whether the age of onset of infections differs in cases with and without a specific disorder. The analyses were adjusted for common confounding factors and other early-onset anxiety disorders. In the case of delayed onset, we examined whether the sequence of a childhood infectious disease and the first symptoms of an anxiety disorder might have induced the result.

The mean age of onset of first symptoms was 7.0 (SD 3.5) years in overanxious disorder, 5.2 (SD 2.2) years in separation anxiety disorder, 12.1 (SD 9.8) years in specific phobias and 10.9 (SD 8.2) years in social phobia. Table 1 displays the average onset age of common viral and bacterial childhood diseases in early anxiety disorders.

Table 2 shows the results for social phobia regressed on the age of onset of chickenpox, measles, and mumps for the whole sample and the results for separation anxiety disorders (only males) regressed on the age of onset of mumps. The analyses in other anxiety disorders did not yield significant estimates (results not shown). Neither the adjustment for covariates (other early anxiety disorders and sex) and potential confounders (education level, age of subjects, and childhood adversities) nor the replication of the analyses after excluding subjects born in 1963 or later yielded any noteworthy differences. However, replication by sex-specific analyses showed that in social phobia the onset of chickenpox, measles, and mumps was consistently delayed in both sexes. The odds ratios based on square root smoothed values for age were 1.8 (1.2–2.7), 2.1 (1.3–3.2), and 2.0 (1.3–3.0) in males, and 1.4 (1.0–1.9), 1.7 (1.2–2.2), and 1.4 (1.0–2.0) in females.

The results from analyses addressing the sequence of each childhood disease and the onset of first symptoms of social phobia (or separation anxiety disorder) are displayed in Table 3. In social phobia, the delay in the onset of chickenpox, measles, and mumps did not induce any noteworthy shift between observed and expected cases. However, in separation anxiety disorders, the delay of the onset of mumps was accompanied by an increase of cases with a mumps infection after encountering preliminary separation anxiety symptoms thus indicating that an altered sequence of onsets was involved and induced artificial results.

Additional analyses examining the associations between social phobia and reported chickenpox, mumps, and measles infections consistently yielded odds ratios above one, but reached significant levels only regarding mumps (OR = 1.26, CI 1.01–1.57).

This study was the first to examine the timing of childhood diseases in early-onset anxiety disorders (separation anxiety disorder, overanxious disorder, specific phobias, and social phobia). We found that social phobia was associated with a delayed age of viral infectious diseases, in particular with respect to chickenpox, measles, and mumps. Notably, these results were separately replicable in males and females and could not be explained by a reciprocal sequence of reported conditions.

Thus, we offer four possible interpretations of our results. First, the timing of the viral infections could interfere with the changing activity of the immune system, as hypothesized regarding EBV infection in/after adolescence and the EBV–MS link (15, 26). However, social phobia symptoms typically have an early onset before adolescence (27). Since the average delay of chickenpox, measles and mumps infections found in persons with social phobia is rather small (below 1 year), it cannot fill the gap between childhood and adolescence. Therefore, there is no support for a crucial role for adolescence and its concomitants in this disorder.

The second interpretation is more trivial, suggesting that the reporting of chickenpox, measles, and mumps basically relies on the visibility of symptoms such as exanthema. It is noteworthy that these childhood diseases had reached a high seroprevalence of above 90%, before the vaccination campaigns began, thus inducing a ceiling effect in the analyses (12, 14). As mentioned earlier, a higher age at infection is a risk factor for more severe symptoms. This is compatible with the increased frequencies of at least mumps infections reported by subjects with social phobia. With this interpretation, some viral infections are directly associated with the risk for social phobia and mediated by a more severe course and symptoms of the infectious disease.

The third interpretation is closely linked to the second one mentioned earlier. It cannot be excluded that an unknown common factor simultaneously increases the risk for more severe forms of childhood diseases and for social phobia. However, no such factor is currently apparent.

Last but not least, the association between social phobia and delayed age of viral infectious diseases could emerge as an implication of coping with early social phobia symptoms, for example, by avoiding social contact in free time and out-of-school activities. A similar rationale would also apply to separation anxiety and overanxious disorder. However, in these instances, it was not supported by the data.

Thus, the most plausible interpretation for the moment is that viral infections in childhood not only precede but also contribute to the risk for social phobia. The reporting of viral infections in a survey such as PsyCoLaus depends on the perception of manifest symptoms by the subjects. In turn, manifest symptoms are related to a more severe course of these infections and thus also to a stronger involvement of the immune system. Therefore, we hypothesize that the increased risk for social phobia is mediated by a relatively stronger involvement of the immune system. This is in line with models linking infectious diseases with neurodevelopmental disorders—for example, the PANDAS model (8)—or with postinfectious fatigue and depression (28–30). However, in contrast to the PANDAS model, viral—and not, or not only, bacterial—pathogens are involved in social phobia. Moreover, the small delay suggests that apart from the severity of viral childhood diseases a specific age stage, i.e., a specific stage of brain development, should also be considered in social phobia.

While the results of this study show a new facet of how the immune system interferes in the etiopathogenesis of social phobia, they should not be generalized to the whole spectrum of social phobia subtypes. As in other phobias (24) or OCD (31), the analysis of epidemiological parameters, risk factors, and comorbidity patterns split by sex and age at onset of first symptoms indicates some hidden heterogeneity of this disorder, as do clinical studies (32).

The analysis of the timing of childhood diseases in early anxiety disorders has shed new light on the connection between viral childhood diseases—chickenpox, measles, and mumps—and social phobia. While the clinical implications of this study are minor, its theoretical implications are challenging. The results suggest a role for the immune system in impacting the development of early onset anxiety disorders.

MP, EC, and CV designed the PsyCoLaus study and acquired the data. VA-G, AA, SR, and MM carried out the analysis. WK, WR, RK, MargM, and RL discussed the preliminary results. VA-G and AA wrote the paper. All the authors contributed to the interpretation of the results and to the critical revision of the manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.