Skip to main content
. Author manuscript; available in PMC: 2020 Nov 3.
Published in final edited form as: Longit Life Course Stud. 2020 Jun 1;11(4):431–458. doi: 10.1332/175795920X15891281805890

Table 3. Summary of findings on consequences of child maltreatment—evidence for an association in the 1958 British cohort.

Outcome Neglect Sexual abuse Physical abuse Psychological abuse
Child development [study number] ~
mental/ emotional health[5] moderate moderate/strong moderate weak/moderate
cognition[5] moderate moderate none none
height[2] moderate none none none
pubertal timing[3] moderate none (M)
inconsistent (F)
none none (M)
inconsistent (F)
adiposity(BMI)[4] None none none none
child-adult BMI gain[4] weak weak (F) weak none
Adulthood
smoker at 23y[4] strong strong strong strong
smoker at 50y[4] strong strong strong strong
heavy drinker at 23y[4] none none none none
heavy drinker at 50y[4] none none none none
physically inactive at 23y[4] weak weak weak weak
physically inactive at 50y[4] weak weak weak weak
Height at 45y[2] weak none none none
general obesity at 45y[9] weak none moderate none
central obesity at 45y[9] weak none moderate none
hypertension at 45y[9] none none none none
HDL-c at 45y[9] weak(F) none* weak(F) weak
LDL-c at 45y[9] none none* moderate none
triglycerides at 45y[9] weak none* None moderate
glycosylated haemoglobin (HbA1c) at 45y[9] weak moderate moderate(M) none
metabolic syndrome at 45y[9] none none* none none
C-reactive protein 45y[10] strong none moderate none
fibrinogen at 45y[10] moderate none none none
cognition at 50y[5] weak none none none
education level at age?[5] strong strong moderate none
LTS at 50y[7] moderate none* moderate moderate
NEET at 50y[7] moderate none* none none
lack of assets at 50y[7] moderate moderate moderate moderate
income-related support at 50y[7] none moderate moderate moderate
financial insecurity at 50y[7] none moderate moderate moderate
manual social class at 50y[7] moderate moderate none none
mental health+ at 50y[12] moderate none none moderate
self-rated health at 50y[12] moderate moderate moderate moderate
physical functioning+ at 50y[12] moderate strong none moderate
premature mortality (45y to 58y)[13] moderate strong moderate none
~

study number given in Table 1; M= males; F=females

LTS and NEET (defined in Table 1) baseline = employed, in education or training.

+

defined in Table 1

Associations between child maltreatment(s) and outcome(s): in general included adjustment for other early life factors (e.g. child SEP) shown previously to be associated with outcomes and, because types of maltreatment are correlated, adjustments for the other maltreatments were made where appropriate. Thus, associations summarised here mostly take account of other forms of neglect/abuse and other early life factors. Associations are consistent with the following criteria:

strong= ORadj (RRadj or HRadj) > 2.0 i.e for binary, multinomial or survival outcomes; standardised mean difference (SMD) >0.5 i.e. for continuous outcomes;

moderate= ORadj (RRadj or HRadj) 1.2-2.0 or SMD 0.2-0.5;

weak= ORadj (RRadj or HRadj) < 1.2 or SMD <0.2;

*

none= no evidence of association using 95% confidence intervals; N for sexual abuse was small and study power was low to detect associations (indicates where associations were moderate/strong but non-significant).

inconsistent=associations vary across different indicators.

Criteria are based on: (i) Cohen's d (standardized mean difference, SMD) for continuous outcomes (0.2, 0.5 and 0.8 are often cited as indicative of small, medium, and large effect sizes respectively)(Cohen, 1988); and (ii) for categorical outcomes, the strength of an association is influenced by the prevalence of the outcome. For example, for an outcome with a 5% prevalence in the non-exposed group, OR reference points for a “weak”, “moderate”, and “strong” association (i.e. equivalent to a SMD of 0.2, 0.5 and 0.8) are 1.52, 2.74, and 4.72 respectively(Chen et al., 2010). An unadjusted OR of 4.72 is high and rarely seen in epidemiological studies. Note that because reference points are unadjusted, whereas associations in the 1958 cohort studies adjust for several potential confounding factors, we used lower cut-offs for adjusted OR, RR and HR as listed above.