Predictors of treatment response in a child and adolescent psychiatry inpatient unit: A retrospective chart review

Predictors of treatment response in child and adolescent psychiatry inpatient units remain incompletely understood. This study examined predictors of treatment response, concordance between clinician-rated and patient-reported response criteria, and response patterns across diagnostic groups in a tertiary adolescent psychiatry inpatient unit. This retrospective single-center chart review included 127 adolescent inpatients with documented group information. The primary analysis group comprised 52 patients with matched admission and discharge Clinical Global Impression-Severity (CGI-S) data. Diagnoses were established according to Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5) criteria. Treatment response was defined as a ≥ 2-point reduction in CGI-S. Overall, 59.6% of patients were classified as responders. CGI-S, Beck Depression Inventory, and Suicide Probability Scale scores improved significantly with large effect sizes. In the multivariable model, higher admission CGI-S (OR = 7.21; 95% CI: 2.20-23.63; P = .001) and male sex (OR = 0.16; 95% CI: 0.03-0.81; P = .03) were associated with treatment response; sensitivity analyses indicated that the association with baseline severity largely reflected baseline-change coupling, so both findings should be regarded as exploratory. Concordance between clinician-rated and patient-reported response definitions was at chance level (Cohen’s kappa = -0.015; 95% CI: -0.36 to 0.34). Diagnostic group differences were at the threshold of statistical significance (Kruskal-Wallis H = 11.08; P = .050), with the lowest response rate in the personality disorder group (25%). Adolescent psychiatric inpatient treatment was associated with clinically meaningful improvement in most patients. Associations of baseline severity and sex with response were exploratory and need confirmation in prospective studies. The limited concordance between clinician and patient perspectives supports multi-informant response assessment. The low response rate in the small personality disorder subgroup was a preliminary observation that requires confirmation.