Annual Research Review: Stability of psychopathology: lessons learned from longitudinal population surveys

Psychopathology has been long recognized as a fluctuating process with various expressions over time, which can only be properly understood if we follow individuals and their social context from childhood up until adulthood. Longitudinal population-based studies have yielded powerful data to analyze this process. However, the resulting publications have not been reflected upon with regard to (a) the homotypic and heterotypic stability of internalizing and externalizing problems and (b) how transactions between psychopathology and environmental factors shape its development.

In this narrative review, we primarily focused on population-based studies that followed cohorts repeatedly from an early age (<18 years) onwards, across multiple stages of development, using statistical methods that permit inferences about within-person bidirectional associations between internalizing and externalizing problems or psychopathology-environment transactions.

There is robust evidence that mental health problems in childhood or adolescence predict psychiatric problems later in development. In terms of the broadband domains internalizing and externalizing problems, homotypic stability greatly exceeds heterotypic stability and transitions from purely internalizing to purely externalizing problems or vice versa are rare. Homotypic rank-order stabilities seem to increase over time. Findings regarding transactions with environmental factors are less robust, due to widely varying research topics and designs, and a scarcity of studies that separated between-person differences from within-person changes. In general, however, the literature shows little consistent evidence for substantial mutual prospective influences between psychopathology and environmental factors.

Longitudinal surveys have strongly augmented insight into homotypic and heterotypic stability and change. Attempts to unravel the myriad of risk and protective factors that place individuals on particular pathways or deflect them from these pathways are still in a pioneering phase and have not yet generated robust findings. As a way forward, we propose to join forces and develop a common risk factor taxonomy.