Comparing three short questionnaires to detect psychosocial dysfunction among primary school children: A randomized method

Abstract

Background. Good questionnaires are essential to support the early identification of children with psychosocial dysfunction in community based settings. Our aim was to assess which of three short questionnaires was most suitable for this identification among school-aged children. Methods. A community-based sample of 2,066 parents of children aged 7-12 years (85% of those eligible) filled out the Child Behavior Checklist (CBCL) and - randomly determined - one of three questionnaires to be compared: the Strengths and Difficulties Questionnaire with Impact Supplement (SDQ), the Pediatric Symptom Checklist (PSC) and the PSYBOBA, a Dutch-origin questionnaire. Preventive Child Healthcare professionals assessed children's psychosocial functioning during routine health examinations. We assessed the scale structure (by means of Structural Equation Modelling), validity (correlation coefficients, sensitivity and specificity) and usability (ratings by parents and professionals) of each questionnaire and the degree to which they could improve the identification based only on clinical assessment (logistic regression). Results. For the three questionnaires, Cronbach's alphas varied between 0.80 and 0.89. Sensitivities for a clinical CBCL at a cut off point with specificity = 0.90 varied between 0.78 and 0.86 for the three questionnaires. Areas under the Receiver Operating Curve, using the CBCL as criterion, varied between 0.93 and 0.96. No differences were statistically significant. All three questionnaires added information to the clinical assessment. Odds ratios (95% confidence intervals) for added information were PSC: 29.3 (14.4-59.8), SDQ: 55.0 (23.1-131.2) and PSYBOBA: 68.5 (28.3-165.6). Parents preferred the SDQ and PSYBOBA. Preventive Child Health Care professionals preferred the SDQ. Conclusions. This randomized comparison of three questionnaires shows that each of the three questionnaires can improve the detection of psychosocial dysfunction among children substantially. © 2009 Vogels et al; licensee BioMed Central Ltd.