Background Adverse childhood experiences (ACEs) have been related to poorer health across the life course. Previous studies typically relied on cumulative risk scores or individual adversities measured through retrospective self-reports. However these approaches have important limitations. Cumulative risk scores assume equal weighting of adversities and the single adversity approach ignores the high probability that adversities co-occur. In contrast, latent class analysis (LCA) offers an alternative approach to operationalise ACEs that respects the clustering of adversities and may identify specific patterns of ACEs important for health outcomes. Furthermore, prospective and retrospective reports of ACEs show poor agreement. Therefore, it is important to compare findings based on prospective and retrospective measures in the same individuals. The aim of this study was to compare LCA, single adversity and cumulative risk approaches to operationalising ACEs with inflammation in mid-life, comparing prospectively and retrospectively-reported ACEs data.

Methods Using data on 8,810 members of the 1958 British birth cohort we investigated 12 ACEs – physical, psychological and sexual abuse, physical and emotional neglect, parental mental health problems, witnessing abuse, parental conflict, parental divorce, parental offending, parental substance misuse and parental death. LCA was applied to explore the clustering of prospectively and retrospectively reported ACEs separately. Associations between latent classes, cumulative risk scores and individual adversities with three inflammatory markers (C-Reactive Protein, fibrinogen and von Willebrand Factor) were tested using linear regression.

Results There was co-occurrence between adversities, and particularly for retrospectively reported adversities. Three latent classes were identified in the prospective data – ‘Low ACEs’, ‘Household dysfunction’ (2.8%) and ‘Parental loss’ (1.5%) which were related to increased inflammation in mid-life, as were high cumulative risk scores and individual measures of offending, death, divorce, physical neglect and family conflict. Four latent classes were identified in the retrospective data – ‘Low ACEs’, ‘Parental mental health and substance misuse’, ‘Maltreatment and conflict’ and ‘Polyadversity.’ The latter two (5.2%) were related to raised inflammation in mid-life, as was a retrospective ACE score of 4+ (8.3%) and individual measures of family conflict, psychological and physical abuse, emotional neglect and witnessing abuse.

Discussion Specific ACEs or ACE combinations might be important for chronic inflammation. LCA is an alternative approach to operationalising ACEs data but further research is needed. Identifying the specific ACEs or combinations of ACEs which are most strongly related to inflammation is important for investigating the mechanisms involved and the planning of effective interventions.