Skip to main content
Have a personal or library account? Click to login
Before the damage is done: Early childhood hyperactivity difficulties in adolescents with deliberate self-harm – findings from the DALSC cohort Cover

Before the damage is done: Early childhood hyperactivity difficulties in adolescents with deliberate self-harm – findings from the DALSC cohort

By:  and    
Open Access
|Jan 2021

Figures & Tables

TABLE 1.

Comparing means by t-tests

DSH (n=649 )No DSH (n=2978 )p-value
Variablenmeannmean
Female46872,1%130443.8%<.0001
Suicide attempted (age 18)12519,5%190.6%<.0001
Mental disorder or psychotropic drugs (age 0–18)29947,1%39513.5%<.0001

1 Note. The number of missing observations was 119 for self-harm; 0 for gender; 17 for suicide; 78 for psychiatric disorder

TABLE 2.

DSH and ADHD

nPct. of sample (n=3627)Pct. of reportedDSH Pct. of ADHD diagnosedPct. of ADHD symptoms
DSH, total64917.9%100%34.9%28.8%
ADHD diagnosed952.5%4.6%100%6.3%
ADHD symptoms103727.7%43.5%68.4%100%

1 Note. The number of missing observations was 119 for self-harm; 1 for ADHD; 8 for ADHD symptoms

FIGURE 1.

SDQ 1999–2011

TABLE 3.

Logistic regression of the specified model

Analysis of maximum likelihood estimates
ParameterEstimateStandard ErrorWald Chi-SquarePr > ChiSq
intercept-1.94190.0954414.0452<.0001
female0-0.66790.0543151.4615<.0001
Hyperactivity difficulties, 3 years0.06850.03254.44030.0351
Hyperactivity difficulties, 7 years0.08410.022414.14930.0002
Peer difficulties, 7 years0.08390.03884.68030.0305
Parents divorced0-0.23480.059315.6575<.0001
Maternal smoking0-0.22110.050319.3254<.0001
Maternal mental health, 3-7 years0.12050.04626.78930.0092

1 Note. N = 3,091. C = 0.707. Binary exposure variables are set to zero; hence, the coefficient signs show the effect on the risk of self-harm when the exposure variable is zero

FIGURE 2.

Odds ratios from the logistic regression of the specified model

FIGURE 3.

Mediation analysis

Note. ***= p < 0.0001; Each arrow represents a regression model (Baron & Kenny, 1986). A was performed as a linear regression with the following control variables: gender, maternal smoking, maternal mental health, parents’ divorce. B was performed as a logistic regression, involving the same controls. The mediation analysis was performed as a bootstrap, and the parentheses shown are bootstrap CI. Percentage mediated = 19.3***

Language: English
Page range: 176 - 188
Published online: Jan 4, 2021
Published in print: Jan 1, 2020
Published by: Center for Evidence-based Psychiatry
In partnership with: Paradigm Publishing Services
Publication frequency: 1 issue per year

© 2021 Therese A. Evald, Bo Møhl, published by Center for Evidence-based Psychiatry
This work is licensed under the Creative Commons Attribution 4.0 License.