Causal Inference in Psychopathology: A Systematic Review of Mendelian Randomisation Studies Aiming to Identify Environmental Risk Factors for Psychopathology

Abstract

Psychopathology represents a leading cause of disability worldwide. Effective interventions need to target risk factors that are causally related to psychopathology. In order to distinguish between causal and spurious risk factors, it is critical to account for environmental and genetic confounding. Mendelian randomisation studies use genetic variants that are independent from environmental and genetic confounders in order to strengthen causal inference. We conducted a systematic review of studies (N = 19) using Mendelian randomisation to examine the causal role of putative risk factors for psychopathology-related outcomes including depression, anxiety, psychological distress, schizophrenia, substance abuse/antisocial behaviour, and smoking initiation. The most commonly examined risk factors in the reviewed Mendelian randomisation studies were smoking, alcohol use and body mass index. In most cases, risk factors were strongly associated with psychopathology-related outcomes in conventional analyses but Mendelian randomisation indicated that these associations were unlikely to be causal. However, Mendelian randomisation analyses showed that both smoking and homocysteine plasma levels may be causally linked with schizophrenia. We discuss possible reasons for these diverging results between conventional and Mendelian randomisation analyses and outline future directions for progressing research in ways that maximise the potential for identifying targets for intervention.