Mental illness among relatives of successful academics: implications for psychopathology‐creativity research

The relationship between creativity and psychopathology is a long standing topic of research1. Creativity is defined as the ability to produce something novel, original, useful and valued, for instance in the domains of art, science or technology. It is being debated if the nature of creativity is general or domain‐specific1. The assumed relationship between creativity and psychopathology is depicted as an inverted U curve, i.e. vulnerability to or low levels of psychopathology are believed to be associated with creativity, which declines with increased psychopathology1.

Kyaga et al² coupled register information on psychiatric diagnosis with census information on self‐reported occupational status. They found that individuals with bipolar disorder and healthy siblings of people with schizophrenia or bipolar disorder were overrepresented among the scientific and artistic professions. Power et al³, in a population study in Iceland, found that higher polygenic risk scores for schizophrenia and bipolar disorder were associated with artistic society membership or creative profession, which could not be accounted for by increased relatedness between creative individuals and those with psychoses.

Typically, we consider someone to be creative post hoc, on the basis of his/her recognized achievements. However, the contemporary measures of creativity typically rely on psychometric tests1 or self‐reported occupational status2, 3. Such approaches have limited validity because they may, in fact, measure either a hypothetical disposition or personal aspiration.

We therefore applied a novel approach to the issue by studying the frequency of mental illness among the relatives of successful academics, i.e., people employed in tenured positions at universities. We assumed that such population would reflect a quasi‐objective creative achievement compared to the background population.

We designed a study with elements from matched cohort studies and case‐control studies. We received the personal identification numbers of all scientific employees in tenured positions at three Danish universities: Copenhagen, Aarhus and Southern Denmark. They were in total 11,803 individuals (referred to as “academics”). These academics were matched 1:6 on age, gender and municipality of residence with randomly selected controls from the background population. Through the Danish Civil Register, we identified first‐ and second‐degree relatives of academics and controls. We divided this population into five subgroups: children, parents, grandparents, siblings and nephews/nieces. Grandchildren were excluded due to low age.

From the Psychiatric Central Research Register, we obtained information on psychiatric diagnoses in academics, controls and their relatives, and grouped these diagnoses following the ICD‐10 hierarchy: schizophrenia, non‐affective psychosis, bipolar disorder, melancholia, any other mental disorder, or no psychiatric diagnosis.

In comparing the relatives of academics and controls, we adjusted for age and gender. Furthermore, we adjusted for intelligence level, as this has been shown to be a significant epidemiological risk factor for schizophrenia4 and therefore represents a confounder. We used the educational level (obtained from Statistics Denmark) as a proxy for intelligence.

The five subgroups of relatives were analyzed in a logistic model, with “relation to academic or control” as the dependent variable and the six diagnostic outcomes as the independent variable, adjusted for education, gender and age. The academics and controls were analyzed separately without covarying for educational level.

All data were anonymized, and the authors had no access to any data that could identify individuals. The study was approved by the Danish National Committee on Health Research Ethics and by the administrations of the Universities.

The total population comprised 588,532 individuals: 11,805 academics; 70,818 controls; 69,325 relatives of academics and 436,584 relatives of controls. The odds ratio (OR) for the academics to be diagnosed with any mental disorder was significantly (p<0.05) lower than for the controls (OR: 0.44, 95% CI: 0.40‐0.49). This also applied to both bipolar disorder (OR: 0.43, 95% CI: 0.27‐0.70) and schizophrenia (OR: 0.17, 95% CI: 0.11‐0.26).

There was a significantly increased risk for schizophrenia among siblings (OR: 1.92, 95% CI: 1.62‐2.27), children (OR: 1.85, 95% CI: 1.38‐2.48) and nephews/nieces (OR: 1.50, 95% CI: 1.15‐1.96) of the academics. For bipolar disorder, the OR was significantly increased among the academics' parents (OR: 1.38, 95% CI: 1.10‐1.74), grandparents (OR: 1.43, 95% CI: 1.03‐1.98) and nephews/nieces (OR: 1.62, 95% CI: 1.04‐2.50), while significance was borderline (p=0.05) for the academics' siblings. The risk for schizophrenia was significantly increased in academics' maternal, but not paternal, half‐siblings. The risk for any other mental disorder was significantly lower among the academics' children (OR: 0.75, 95% CI: 0.69‐0.82) and nephews/nieces (OR: 0.72, 95% CI: 0.67‐0.78).

This study shows that, while successful academics as a group are less prone to mental disorders than the background population, there are increased rates of schizophrenia and bipolar illness among their biological relatives. Other mental disorders, on the other hand, are less frequent among the relatives of academics. Because of our a priori hypothesis, we believe that this study supports the idea of a link between creativity and vulnerability to mental illness. We acknowledge, however, that the association between academic status and increased rates of schizophrenia and bipolar disorder in the relatives may be caused by multiple other factors.

The hypothesized relationship between creativity in successful academics and the increased risk for schizophrenia and bipolar disorder in their relatives seems to be mediated by a vulnerability that is not manifested as overt mental disorder in the academics, consistent with the inverted U curve model.