The impact of early adversity on later life health, lifestyle, and cognition | BMC Public Health

ELSA dataset

In the ELSA dataset, most of the early adversities (with the exception of sexual assault, which was not associated with cognition, and deprivation, which was only associated with cognition) were associated with BMI, subjective health, depressive symptoms, smoking status, alcohol consumption, and cognition (i.e., immediate memory and verbal fluency). Specifically, in line with the literature, early adversities were found to be adversely associated with BMI, subjective health, depressive symptoms, smoking status, alcohol consumption, and cognition [2,3,4,5,6, 9,10,11,12,13,14,15,16, 55]. However, some inconsistencies were also found.

First, contrary to its counterpart adversities (i.e., physical assault and parental abuse) having experienced sexual assault before the age of 16 was not associated with immediate memory and verbal fluency performances. These results may either suggest a robustness of immediate memory and verbal fluency to the negative impact of experiencing early sexual abuse or a fully mediated effect of experiencing early sexual abuse on cognition via mental health, lifestyle, and/or education. Having experienced physical assault before the age of 16 was associated with lower BMI, lower levels of depressive symptoms, being non-smoker, higher self-rated health, and higher cognitive performance, but physical assault was also associated with higher frequency of alcohol consumption. These results need special focus and further study to be able to fully grasp them. However, different explanations could be suggested. The low response rate to this specific adversity (n = 340) may bias the results found (i.e., statistical power) but may also suggest that the way this adversity was assessed was not clearly formulated or added confusion in relation to the other adversities assessed and their possible overlap (see supplements Table A, Table B, and Table D). Another explanation could be the role of chronicity. No information regarding chronicity was available in this dataset, therefore it could not be excluded that the respondents had, in majority, experienced a single episode of physical assault which may therefore contribute to the development of their resilience – achieving positive outcomes despite the experience of adversity [56]– [57].

A positive association was found between physical assault and increased alcohol consumption, contrary to the experience of sexual assault and parental abuse which were both associated with lower alcohol consumption. Interestingly, in the literature low-to-moderate alcohol consumption has been positively associated with cognition and protective against cognitive decline [58,59,60]. However, some caution is needed when interpreting the results related to alcohol consumption as chronic and heavy alcohol consumption has a negative impact on health and cognition and any positive effect of low-to-moderate alcohol consumption may also be outweighed by the risk of harmful effects [58,59,60].

Regarding the experience of deprivation, only two associations with cognition were found to be significant, poorer immediate memory and verbal fluency performance. This result is in line with the literature showing a negative effect of low childhood socioeconomic status with later life cognition [61,62,63,64,65]. The finding that deprivation has no significant effect on physical and mental health, neither lifestyle variables, may be due to the severity of the other adversities included in the model and the often-found co-occurrence between adversities (of different severities) which could thus outweigh the potential negative effect of deprivation [8, 11]. Another explanation could be that deprivation only indirectly affects physical and mental health, as well as lifestyle, and thus that other factors are at play in these associations (e.g., education).

UK Biobank dataset

In the UK Biobank dataset, experiencing physical abuse, emotional abuse, physical neglect, sexual abuse, or emotional neglect was associated with depressive symptoms as previously demonstrated in the literature [2, 3, 5, 11,12,13, 15]. In line with studies showing a negative effect of early adversity on subjective health [66,67,68], experiencing emotional abuse or emotional neglect showed a similar pattern of results as both were associated with higher levels of depressive symptoms and lower self-rated health. This similar pattern of results may suggest the involvement of a common mechanism related to the emotional component of adversity in contrast to the literature suggesting a common mechanism related to the threat or deprivation component of adversity [36, 55, 69]. Interestingly, and contrary to the literature findings, all other associations between emotional abuse or emotional neglect and the other outcomes of interest were non-significant [4,5,6, 10, 11, 14, 16, 55]. It may be possible that the experience of emotional adversity may have a strong influence on self-perception/self-esteem but also on personal control (e.g., subjective health and self-reported depressive symptoms) that may indirectly affect the objective measures (e.g., BMI, and cognition), contrary to physical adversity (neglect or sexual abuse) which shows a direct effect on objective cognitive measures [66, 70, 71]. Nonetheless, further studies including mediation links through self-perception/self-esteem as well as the addition of self-perceived cognitive measures may provide a better insight into these results.

For physical adversity (physical abuse, physical neglect, and sexual abuse) no specific pattern of outcomes linked to the physical component could be identified (all the early adversities were associated with depressive symptoms). Interestingly, although, both physical neglect and sexual abuse showed an association with higher levels of depression and slower TMTB completion time, physical neglect showed the most significant associations with the outcomes studied. Physical neglect was associated with higher levels of depressive symptoms, lower frequency of alcohol consumption, and poorer cognitive performance (i.e., visual memory task, TMTB, and fluid intelligence). These results are in line with the literature showing a negative effect of early adversity on mental health, and cognition (e.g., [2, 3, 5, 10,11,12,13, 15]). However, the findings do not support previous research showing a negative effect of early adversity on physical health and lifestyle (e.g., [3, 9, 11, 12, 16]). The role of a mediator may be hypothesized as underlying the association between physical neglect and physical health, as well as smoking status. Interestingly, the results regarding alcohol consumption need to be cautiously interpreted due to some positive associations found in the literature between alcohol consumption and health and cognition [58,59,60].

Early sexual abuse was associated with cognitive tasks involving executive functions (i.e., trail-making test and fluid intelligence), suggesting that early sexual abuse may have an effect on the prefrontal and frontal cortices. Importantly the effect of sexual abuse on grey matter volume in the frontal cortex has been demonstrated in a specific time period (i.e., 14–16 years old) [72]. Therefore, it may be possible that sexual abuse, in this sample, may have been experienced specifically in this time frame, which may also explain the absence of a significant association between sexual abuse and memory (i.e., hippocampus) in this sample.

Physical abuse was only significantly associated with higher levels of depressive symptoms. In general, the non-significant associations between adversity and the outcomes of interest may be explained by education (except for smoking status). Indeed, education was significantly associated with all the outcomes of interest except smoking status. Therefore, it may be possible that education in part or fully mediates the association between adversity and the outcome of interest. Interestingly, all early adversities were not significantly associated with BMI and smoking status contrary to the literature findings and the ELSA dataset. Although it is possible that the effect of early adversity on BMI is mediated by education (in contrast to smoking status), for smoking status, it is important to highlight that in the UK biobank dataset, a high number of missing values were present in this variable (i.e., missing values = 497,006), and in addition, only a few participants smoked at the assessment time (n = 118). This may, in part, explain why no significant association was found with the smoking variable but also the differences found between these two cohorts for smoking. In addition, all early adversities were also not significantly associated with reaction time. This result may be explained by the significant association found between age and reaction time. Indeed, age is known to affect reaction time, with increasing age slower reaction time, and thus age may explain most of the reaction time variance in this model [73, 74].

Cohort comparisons

Overall, and in both cohorts, most of the early adversities studied were found to be negatively associated with physical health, mental health, lifestyle, and cognition. Importantly, in both cohorts, differences according to the specific adversity experienced were found, in line with the literature suggesting the importance of considering the specific adversity experienced and the potential role of in part differential underlying mechanisms [32, 36]. In addition, across the two cohorts, almost all early adversities (i.e., physical assault, sexual assault, and parental abuse in the ELSA dataset and physical abuse, emotional abuse, physical neglect, sexual abuse, and emotional neglect in the UK Biobank dataset) were shown to be consistently associated with depressive symptoms. In line with the literature, except for physical assault in the ELSA dataset, each adversity experience was associated with higher levels of depressive symptoms [2, 3, 5, 11,12,13, 15]. The same arguments as the ones previously mentioned may explain this result (i.e., low response rate, the role of chronicity, and increased resilience). Importantly, physical assault in the ELSA dataset has no adversity equivalent in the UK Biobank, since physical abuse is closer to parental abuse. Although the early adversities have been cautiously selected in both cohorts to be the most comparable, there is no perfect overlap between them with the greatest overlap between parental abuse (ELSA) and physical abuse (UK Biobank) as well as sexual assault (ELSA) and sexual abuse (UK Biobank).

The other discrepancies observed between the two cohorts may be determined by the different assessments (i.e., adversity and outcomes assessments). Cohort characteristics (i.e., education but also age) may also explain these results mainly for subjective health and cognition. Indeed, the ELSA cohort is older, and therefore decrease in subjective health and cognitive impairment may be more detectable in an older population.

Mediation

Education was associated with numerous outcomes in both cohorts, therefore education may (fully and partially) mediate the effects of early adversity on the outcomes of interest in both cohorts [9]. Importantly, education has been found to be important for resilience [75]. To test this hypothesis, additional analyses were performed with education as a mediator. The supplementary analyses support this mediation explanation. In the ELSA dataset, all the associations between deprivation and the outcomes of interest (i.e., BMI, subjective health, depressive symptoms, smoking status, alcohol consumption, immediate memory, and verbal fluency) are mediated by education. The associations between early adversities (i.e., physical assault, sexual assault, and parental abuse) and the outcomes of interest are not significantly mediated by education (see supplements Table I and Table J). In the UK Biobank, all the early adversities (i.e., physical abuse, emotional abuse, physical neglect, sexual abuse, and emotional neglect) are mediated by education for all of the outcomes of interest (i.e., BMI, subjective health, depressive symptoms, alcohol consumption, visual memory, Trail Making Test B, reaction time, and fluid intelligence) except for smoking status (see supplements Table W and Table X). These results, emphasize the mediating role of education in the associations between adversities (in general including deprivation and abuse) and physical health, mental health, lifestyle, and cognition.

Strength and limits

This study has the advantage of using large sample sizes, which may be more representative of the general population, as well as cross-cohort models, which may increase the generalization of the effects found [76]. Therefore, at minimum, the associations between early adversities and depressive symptoms (measured differently according to the cohorts) seem robust.

However, in this study, the early adversities selected, although close, are not perfectly overlapping. Therefore, further studies with a perfect overlap between the cohort adversities would not only strengthen the association found with depression, but also replicate and thus strengthen the other results found in this study. It is also important that future studies use different cohorts from different countries and with different characteristics to generalize the results found. In addition, the inclusion of longitudinal (instead of cross-sectional) data would provide access to temporality and insight into the causal relationship between the variables of interest. Furthermore, the lack of consistency in the adversity assessment across the studies makes the replicability and, thus, generalization of the results difficult. In addition, the population included in these cohorts and the adversity assessment (self-reported and retrospective) are subject to various biases (e.g., survival, selection, and resilience bias, recall, social and mental health bias) [77,78,79,80,81,82,83]. However, it is important to note that the reliability of self-reported adversity has already been emphasized and might be even under-reported [84, 85].