medwireNews: Reducing the cutoffs of commonly used depression screening measures may increase the sensitivity for detecting depression in adolescents with type 1 diabetes, US study findings indicate.
“While we were surprised to find that cutoff scores should be decreased across most depression screeners to improve sensitivity, one possible explanation could be that adolescents in our sample under-reported depressive symptoms on screeners,” write Susana Patton (Nemours Children’s Health System, Jacksonville, Florida) and co-authors in Diabetes Care.
They continue: “The clinic we recruited adolescents from routinely screens for psychosocial concerns (e.g., depression, anxiety, and disordered eating), suggesting that adolescents could be primed to underreport symptoms to complete measures faster or avoid having to remain in clinic to meet with a mental health professional.”
The study included 100 participants (60% boys) aged 12 to 17 years (mean 15 years) with a mean diabetes duration of 6.0 years and a mean glycated hemoglobin level of 8.9% (74 mmol/mol).
During one visit, they each completed the Beck Depression Inventory (BDI)-II, Center for Epidemiological Studies Depression Scale – Revised (CESD-R), Children’s Depression Inventory (CDI)-2, Patient Health Questionnaire for adolescents (PHQ-9A), and the Short Mood and Feelings Questionnaire (SMFQ), as well as the Kiddie Schedule for Affective Disorders and Schizophrenia (K-SADS-PL) semi-structured diagnostic interview.
Based on the reference standard K-SADS-PL interview, 15 adolescents met diagnostic criteria for any depressive disorder, and the researchers report that their diagnostic category (i.e., depressed vs not depressed) strongly and significantly correlated with total scores on the depression screeners.
However, the sensitivity of the screening measures using the pre-existing cutoff scores was “low,” ranging from 33% to 60%, say Patton et al. Specificity ranged from 94% to 99%.
When the researchers reduced the cutoffs, with a goal of achieving 80% sensitivity, the measures detected depression in 80% to 93% of cases. Specificity at these reduced cutoffs ranged from 69% to 80%. The degree of reduction needed in the cutoffs ranged from 5 points (≥10 to ≥5) for PHQ-9A to 12 points (≥20 to ≥8) for BDI-II.
Data for the short versions of the PHQ-9A (PHQ-2A) and the CDI-2 measures were also analyzed. At the standard cutoff, PHQ-2A had a sensitivity of 67% that increased to 87% when the cutoff was reduced from 20 to 11; the corresponding specificities were 89% and 73%. For the CDI-2 short screener, the initial sensitivity was 100% but specificity was 54%, so the researchers increased, rather than decreased, the cutoff in this case. At the adjusted cutoff (≥5 rather than ≥3), sensitivity and specificity were 87% and 82%, respectively.
Patton et al say their results show that “clinics should consider adjusting depression screener cutoff scores to increase sensitivity and reduce false-negative rates.”
For the best diagnostic accuracy, the researchers suggest that “clinics should administer the CDI-2 Short form, if feasible, to use a measure with proprietary costs or the PHQ-9A and SMFQ if clinics are unable to routinely administer a proprietary measure due to cost or low reimbursement.”
They add that “clinics should avoid using the PHQ-2A alone, as this very short measure had low sensitivity for detecting depressive disorders in our sample.”
medwireNews is an independent medical news service provided by Springer Healthcare Ltd. © 2022 Springer Healthcare Ltd, part of the Springer Nature Group
Diabetes Care 2022; doi:10.2337/dc22-0275