SEARCH

SEARCH BY CITATION

Keywords:

  • depressive disorder;
  • adolescent;
  • mental health;
  • hospital, pediatric;
  • diagnosis;
  • ROC curve;
  • sensitivity and specificity

Background

This study examines the criterion validity of the Patient Health Questionnaire 9-item (PHQ-9) and 2-item (PHQ-2) version as a depression-screening instrument for adolescents.

Methods

Three hundred twenty-two adolescents aged 13–16 were recruited from pediatric hospitals. Criterion validity of the PHQ-9 and PHQ-2 was assessed against diagnoses of any depressive disorder provided by a structured diagnostic interview. Areas under the receiver operating characteristics curve (AUCs) and sensitivities and specificities at optimal cutoff points were computed for both versions of the PHQ. Besides the dimensional algorithm, a categorical algorithm was applied for the PHQ-9. Validity measures of both scoring procedures of the PHQ-9 as well as PHQ-2 were compared statistically. In addition, unaided clinical depression diagnoses by the attending pediatricians were evaluated.

Results

Using the dimensional algorithm, the AUCof the PHQ-9 (93.2%) was significantly higher than that of the PHQ-2 (87.2%). At optimal cutoffs, there was no significant difference in sensitivity (PHQ-9: 90.0%, PHQ-2: 85.0%), but in specificity (PHQ-9: 86.5%, PHQ-2: 79.4%). Although the categorical algorithm of the PHQ-9 was most specific (94.7%), sensitivity was just above chance (52.5%). The unaided clinical diagnoses yielded a sensitivity of 12.5% and a specificity of 96.0%.

Conclusions

The dimensional algorithm of the PHQ-9 demonstrated high criterion validity, whereas the categorical algorithm should not be applied due to its low sensitivity. Even though the PHQ-2 performed well, validity of the PHQ-9 was still superior. Hence, the PHQ-9 can be recommended as depression screener for adolescents to improve recognition rates in pediatric care.