Association of Screening Question with Impact

Association of Screening Question with Impact

SUPPLEMENTARY RESULTS

Association of symptom dimensions with Impairment.

Parent report

When both scales were entered simultaneously into a linear regression model, parent-reported episodic under-control was a significant predictor of impact even after adjusting for the presence of DSM-IV depressive or anxiety disorders, conduct disorders, and ADHD, as well as age and gender (b=0.12; CI 0.06 to 0.19), while parent-reported exuberance was not a significant predictor.Also the association between parent-rated episodic under-control and SDQ-rated impact also remained significant after adjusting for continuous measures of psychopathology, the SDQ emotional, conduct, hyperactivity scores, as well as age and gender (b=0.07; CI 0.01 to 0.12).

Self report

Self-reported episodic under-control was a significant predictor of impact even after adjusting for the presence of DSM-IV depressive or anxiety disorders, conduct disorders, ADHD, as well as age and gender (b=0.13; CI 0.09 to 0.17); the association between self-rated episodic under-control and SDQ-rated impact also remained significant after adjusting for continuous measures of psychopathology, the SDQ emotional, conduct and hyperactivity scores as well as age and gender (b=0.12; CI 0.08 to 0.16).

Fit indices for the latent classes

Supplementary Table 1. Parent reported log likelihood and BIC.

Parent Report / Number of latent classes
1 / 2 / 3 / 4 / 5 / 6
Log Likelihood / -7793.56 / -6597.02 / -6255.6 / -6159.8 / -6070.5 / -6018.49
BIC / 15752.7 / 13531.6 / 13020.7 / 13001 / 12994.4 / 13062.3

Supplementary Table 2. Self reported log likelihood and BIC

Self Report / Number of latent classes
1 / 2 / 3 / 4 / 5 / 6
Log Likelihood / -10370 / -8960.6 / -8637.4 / -8553.1 / -8489 / -8442.5
BIC / 20916.9 / 18280.1 / 17816.6 / 17831 / 17885.6 / 17975.5

Association of latent classes with impact.

Parent Report

The relationship between the parent-reported latent classes and impact was further examined in a multinomial logistic regression model: the top (relative risk ratio (RRR)= 1.36 , CI=1.16 to 1.59) but not the intermediate class (RRR=1.12 , CI=0,98 to 1.28) were significantly associated with impact (compared to the low class that was used as the base outcome) even after adjusting for depressive and anxiety disorders, ADHD, and CD/ODD as covariates.

Self Report

The relationship between the self-reported latent classes was further examined in a multinomial logistic regression model: the top (relative risk ratio (RRR)=1.38, CI=1.15 to 1.65) and the intermediate class (RRR=1.38 , CI=1,15 to 1.66) were significantly associated with impact (compared to the low class that was used as the base outcome) even after adjusting for depressive and anxiety disorders, ADHD, and CD/ODD as covariates.