Statistics from Altmetric.com
Is the Strengths and Difficulties Questionnaire (SDQ) hyperactivity-inattention subscale and predictive algorithm a valid screening tool for attention deficit hyperactivity disorder (ADHD) phenotype in Norwegian school children?
6233 children (7–9 years old). Questionnaires were sent to teachers and parents of all 9137 children enrolled in second through fourth grades in Bergen, Norway during 2002.
Public, private and special education schools in Bergen, Norway; 2002.
SDQ hyperactivity-inattention subscale, reported by both teachers and parents. The SDQ is a 25 item broad-band questionnaire designed for parents, teachers and self-report, and covers child behaviour, emotions, relationships and symptom impact. The inattention-hyperactivity subscale consists of 5 items, and is combined with impact scores using a prediction algorithm to predict ‘unlikely’, ‘possible’ and ‘probable’ cases of ADHD.
18-item SNAP-IV scale (teacher and parent report) for DSM-IV ADHD symptoms. ADHD phenotypes were defined as: ADHD – inattentive (ADHD-I) if both parent and teacher reported ≥6 of the nine inattention symptoms; ADHD – hyperactive/impulsive (ADHD-H/I) if both parent and teacher reported six or more of the nine hyperactive/impulsive symptoms; ADHD – combined (ADHD-Co) if one informant reported six or more symptoms on both the inattentive and hyperactive/impulsive subscales, and the other informant reported six or more symptoms on at least one of the subscales or if one informant reported six or more symptoms on one of the subscales and the other informant reported six or more symptoms on the other subscale.
Area under the curve (AUC), sensitivity, specificity and positive predictive value (PPV).
Using the SNAP-IV standard, 326 children met the criteria for the ADHD phenotype, resulting in an overall prevalence of 5.2% (3.3% ADHD-Co, 1.6% ADHD-I, 0.3% ADHD-H/I). The receiver operating characteristic curve for the teacher report SDQ hyperactivity-inattention subscale estimated the AUC for predicting the ADHD phenotype as 0.95 (95% CI 0.94 to 0.95), and the AUC for parent reports as 0.91 (95% CI 0.90 to 0.92). There were no significant gender differences for parent reports, however, teacher reports showed a significantly better AUC for girls (0.96, 95% CI 0.95 to 0.97) than boys (0.92, 95% CI 0.91 to 0.94). Using a cutoff score of 7 out of 10 on the teacher report SDQ hyperactivity-inattention identified 55% of the boys and 32% of the girls with the SNAP-IV ADHD phenotype. The parent reports using this same threshold identified 38% of boys and 34% of girls with the SNAP-IV ADHD phenotype. Using the predictive algorithm for predicting ‘probable’ cases had a sensitivity of 23% for the ADHD phenotype, a specificity of close to 100% and PPV of 0.93. Including participants identified by the algorithm as ‘possible’ cases increased the sensitivity to 52%, with a specificity of 98% and PPV of 0.58. By ADHD subtype, using the predictive algorithm to identify ‘possible’ or ‘probable’ cases identified 74% of children were with the ADHD-Co phenotype according to the SNAP-IV, 22% with ADHD-I and 14% with ADHD-H/I.
Obtaining good sensitivity for the ADHD phenotype with the 5-item SDQ hyperactivity-inattention subscale may require a lower cutoff than 7/10 in Norwegian school-aged children. It has better sensitivity for the ADHD – combined subtype than for the ADHD – inattentive or ADHD – hyperactive/impulsive subtypes.
Ullebo et al present data to support the use of the Strengths and Difficulties Questionnaire (SDQ) as a screening measure for attention deficit hyperactivity disorder (ADHD). The development of reliable and valid screening tools to assist in identifying children who should be further evaluated for ADHD is of the utmost importance. Given ADHD's status as one of the most common childhood psychiatric conditions, it must be screened by primary care clinicians in order for effective, comprehensive care to be provided. However, because these clinicians are responsible for evaluating many aspects of children's health, the length of any screening measure specific to ADHD must be considered, as such a measure will, in many cases, be embedded within a much longer screening tool designed to capture functioning in multiple domains.
The SDQ offers many practical benefits for clinicians over other screening measures of ADHD symptomatology. Both the measure itself, and its normative data (gathered from several countries) and scoring procedures are readily available online at no cost. The SDQ has been translated into over 60 languages and includes parent-, teacher-, and self-report versions. Thus, the measure is very practical for use by clinicians within a wide range of settings. This study by Ullebo et al suggests that the SDQ's five ADHD screening items, both alone and in combination with ‘impact scores,’ offer reasonable prediction of the ADHD-combined phenotype, but do not provide adequate predictive power for either the inattentive or hyperactive/impulsive phenotype. For these, the measure had a high rate of false negatives and would fail to identify many children with significant symptoms of ADHD. The SDQ, despite its practicality, does not seem to be an appropriate large-scale screening tool for ADHD in the clinical context. This conclusion should be viewed as tentative; however, due to the considerable difference in ADHD prevalence between this study's Norwegian sample and that of other geographic regions, which have an effect on predictive values. Further research in this area is needed before firm conclusions can be drawn regarding the SDQ's utility as a screener for ADHD.
Sources of funding Norwegian Research Council and the Centre for Child and Adolescent Mental Health, Unifob Health, Bergen.
Competing interests None.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.