Exploring potential bidirectional causality between psychotic experiences and religiosity in a UK longitudinal cohort study (ALSPAC)

Background: Psychotic experiences (PEs) are common in the general population, and can be an early sign of psychotic disorders, which can have a large impact on people’s lives. Understanding the causes and consequences of PEs is therefore important, both for identifying potential causal risk factors for PEs and for exploring how PEs may subsequently affect people’s beliefs and behaviours. To investigate this, we focus on potential bidirectional causality between PEs and religiosity – a topic which remains under-researched and currently with a weak evidence base – using large-scale data from a UK longitudinal birth cohort. Methods: We used data from the Avon Longitudinal Study of Parents and Children (ALSPAC). PEs were assessed using semi-structured interviews at age 24 (for PEs since age 12) and self-reported questionnaires at age 32 (for PEs in the past year). Religiosity was self-reported at age 28, and included questions on religious beliefs, identity and service attendance. Multivariable regression models, adjusted for relevant confounders, analysed bidirectional associations between PEs and religiosity (i.e., whether PEs from age 12-24 potentially cause religiosity at age 28, and whether religiosity at age 28 potentially causes PEs at age 32). Multiple imputation was used to impute missing data and boost statistical power, with g-computation used to calculate our marginal causal contrasts of interest.Results: Interview-rated PEs between age 12-24 were associated with a greater probability of religious belief; participants with PEs were 7.8%-points (95% confidence/compatibility interval [CI] = 1.9% to 13.7%) less likely to answer ‘no’ to believing in God, and 5.7%-points (95% CI = 0.7% to 10.7%) more likely to answer ‘yes’. Similar patterns were observed for religious identity (PEs associated with a 5.6%-point [95% CI = 0.2% to 11.0%] increase in identifying as religious), but with weaker evidence of an association with religious service attendance (PEs associated with a 1.7%-point [95% CI = -1.0% to 4.4%] increase in regular attendance). Religious belief at age 28 was also associated with an increased probability of self-reported PEs at age 32 (5.7%-points [95% CI = 1.9% to 9.5%]), with effects for religious identity (2.1%-points [95% CI = -0.9% to 5.1%] increase in PEs) and religious attendance (5.6%-point [95% CI = -1.8% to 12.9%] increase in PEs) in the same direction but weaker and/or plausibly null. Conclusion: To the extent these results can be given a causal interpretation, these findings suggest a potential bidirectional causal relationship between PEs and religiosity, especially regarding religious beliefs. Further research is needed to explore whether these results are replicable and generalisable across populations, in addition to whether religiosity may moderate or mediate the long-term impact of PEs on mental health outcomes.