Improving the K6 short scale to predict serious emotional disturbance in adolescents in the USA

Int J Methods Psychiatr Res. 2010 Jun;19 Suppl 1(Suppl 1):23-35. doi: 10.1002/mpr.314.

Abstract

Effective screening for emotional and behavioral disorders among youth requires brief screening scales with good validity to identify youth requiring further evaluation and to estimate prevalence of target disorders in populations of interest such as schools or neighborhoods. This paper examines the psychometric properties of a very short (six-item) screening scale, the K6, to assess serious emotional disturbance (SED) among youth. The K6, which is made up of symptoms of depression and anxiety, has been shown in previous research to be a strong predictor of serious mental illness (SMI) in adults, but no information is available on the ability of the scale to screen for SED among youth. The current report examines the K6 as a screen for SED in a national survey of US adolescents, the National Comorbidity Survey Replication Adolescent Supplement (NCS-A). The K6 is shown to provide fairly good prediction of SED [area under curve (AUC) = 0.74] that is somewhat higher for internalizing (AUC = 0.80) than behavior (AUC = 0.75) disorders. Based on this result, we augmented the K6 with questions about symptoms of behavior disorders. This improved prediction of SED (from AUC = 0.74 to AUC = 0.83) as well as of SED associated with pure behavior disorders (from AUC = 0.53 to AUC = 0.78). These results show that although the symptoms of depression and anxiety in the K6 are sufficient to detect SMI among adults, high rates of behavior disorders among adolescents require indicators of behavior disorders to be added to the K6 to screen adequately for adolescent SED.

Publication types

  • Research Support, N.I.H., Extramural
  • Research Support, Non-U.S. Gov't

MeSH terms

  • Adolescent
  • Affective Symptoms / diagnosis*
  • Affective Symptoms / epidemiology
  • Area Under Curve
  • Behavioral Symptoms / classification
  • Behavioral Symptoms / diagnosis
  • Female
  • Health Surveys
  • Humans
  • Male
  • Predictive Value of Tests
  • Psychometrics / methods*
  • Reproducibility of Results
  • Retrospective Studies
  • Surveys and Questionnaires
  • United States / epidemiology