As in previous studies, low socio-economic status and low parental educational achievement were associated with participants reporting abuse and neglect age 45y294;318. Unlike other studies298;441, level of maternal education was not associated
remaining in education in the mid-twentieth century, compared to contemporary populations. The established literature indicates that disabled children are more likely to be abused and neglected compared to able-bodied children442. Disability was not associated with retrospective measures of child maltreatment in the 1958 cohort, potentially due to the small number of disabled participants in the cohort.
The prevalence of witnessing abuse, psychological and physical abuse in the 1958 cohort was within the range reported in a recent review29, but the prevalence of sexual abuse was lower (1.6%- 2.9% versus 5-30%). The disparity in estimates possibly reflects differences in study populations, definition of childhood and sexual abuse, and methods of data collection (e.g. agency reports vs. parent or self-reports) 39;429;430. In a meta-analysis of worldwide studies, methodological factors accounted for nearly a quarter of the variance in prevalence estimates of child sexual abuse. Prevalence estimates were lower in: 1) community cohorts compared to college and general practitioner samples, 2) studies that used narrow rather than broad definitions of sexual abuse, 3) those that adopted one as opposed to multiple questions to assess sexual abuse and 4) male cohorts compared to females443. Thus, prevalence estimates may be low in the 1958 cohort because it is a general population study which adopted one question to ascertain child sexual abuse. The low prevalence of sexual abuse may have
implications for the power to detect an effect on later outcomes, such as child-to-adult height growth and pubertal development.
Some emotional neglect measures collected in childhood were highly prevalent, e.g. ‘mother/father little interest in education’, such that two thirds of the cohort had ≥ 1 neglect indicator. This was possibly because most were derived from repeat
measurements. It is likely that reliance on any single indicator of neglect may be misleading. However, the prevalence of neglect in the 1958 cohort (3.5-30.4%) is comparable to that reported in the NSPCC’s 1999 survey in which over a third of participants reported at least one form of emotional neglect444.
Consistent with previous studies, psychological and sexual abuse were retrospectively reported more frequently by females than males23;342;416. Neglect measures at age 45y were also more commonly cited amongst females than males. In contrast, where gender differences were found for prospective indicators of neglect, a greater proportion of
boys were neglected compared to girls. The discrepancy between self and parent- reported neglect measures suggest that males were less likely to report maltreatment experiences at age 45y, than females. In a Canadian study, male participants known to have experienced sexual abuse were less likely to retrospectively report child
maltreatment than female cohort members418. Herbert et al concluded that male victims may be less likely, than females, to disclose their abuse experiences due to greater feelings of shame and self-blame. Therefore, by using both prospective and
retrospective methods to ascertain cases of neglect, it is possible that fewer participants will have been misclassified.
4.5.3 Co-occurrence of child maltreatment
Co-occurrence of different forms of child maltreatment was common in the 1958 cohort, as shown elsewhere39. In other community studies, the proportion of children who experienced multiple forms of abuse (psychological, physical and sexual abuse and witnessed abuse of family member) varies from 42.1–64.4%37;127;445. Estimates from this study are at the upper end of this range: of those abused, 61.9% males and 68.3% females reported another form of maltreatment (including neglect).
The identification of two distinct maltreatment groups in this study was a novel finding. Although co-occurrence of child maltreatment measures was observed in the cohort, a distinct group of participants who had a high risk of neglect but a low risk of abuse was detected. Childhood abuse tended to co-occur with neglect, but less so vice versa. This is consistent with findings from LONGSCAN, where the majority of sexual and
physical abuse cases involved multiple forms of maltreatment, whereas neglect cases rarely reported co-occurring abuse34. Agency reports have found neglect to be the most common maltreatment reported18;99, and community studies have shown that neglect is the most frequent form of maltreatment to occur in isolation446. Results from the 1958 cohort further support findings that indicate that abused participants are more likely to be neglected, whilst those neglected were not necessarily abused.
It is possible that the identification of a neglect ‘only’ group may be due to the high prevalence of prospectively reported indicators of neglect in the 1958 cohort. Over a
interest in education and low parental aspirations, and over two thirds reported at least one indicator of neglect. In the LCA model, prospective and retrospective neglect measures distinctly clustered between the two maltreatment groups: 45y measures in the abuse and neglect group and childhood measures in neglect ‘only’. Furthermore, the correlation between retrospective and prospective measures of neglect was weaker compared to correlations between measures collected during the same period (Appendix 3.3). These differences suggest that indicators of neglect collected in childhood and at age 45y may be measuring different things. Instead, prospective neglect measures may be more reflective of socio-economic disadvantage, than retrospective variables. Comparing the relationship between different indicators of neglect and social class at birth, childhood measures are more strongly related to social class at birth than those reported in adulthood (Appendix 3.7). In addition, the strength of association between maltreatment groups and social class at birth is greater amongst participants classified as neglected ‘only’, compared to none maltreated and abuse and neglect. Thus, the LCA model may have identified participants who were at high risk of socio-economic disadvantage, as well as parental neglect.
4.5.4 Household dysfunction and child maltreatment
Evidence is substantial for both volatile family environment94;437;447 and physical punishment94;447;448 to be associated with child maltreatment. In the 1958 cohort, the relationship between these household dysfunction measures and child maltreatment was stronger than those presented elsewhere. The difference in findings may reflect
methodological differences with prior studies mostly using agency reports to investigate the association430. Alternatively, the larger effect sizes found between certain measures could reflect a lack of distinctness of concepts between some measures. For example, the strong association between physical punishment and physical abuse may be because they are measuring the same thing. Likewise, ‘a lot of conflict’ and witnessing abuse may reflect similar constructs and thus were strongly related.
Overall, findings from the 1958 cohort indicate that household dysfunction measures were associated with a range of child maltreatment measures. Specific markers of household dysfunction may be particularly informative for predicting some forms of child maltreatment. For example, family contact with children’s department or charity
and time in care were more strongly associated with sexual abuse, than other forms of maltreatment. However, the strength of association between most forms of household dysfunction measure and child maltreatment were comparable. Thus a greater number of risk factors may be more predictive of later maltreatment, than any one individual risk factors192. In addition, the result of a stronger association for the abuse and neglect group, than neglect ‘only’ suggests that children from dysfunctional family backgrounds have a higher risk of being maltreated across a broad spectrum than those who are not. In the 1958 cohort, measures of child maltreatment and household dysfunction
accumulated in participants. Such accumulation of adverse childhood experiences may have implications for the later health of cohort members.