Abstract
OBJECTIVE: To examine the prevalence and correlates of child involuntary mental health detentions through evaluation of legal documentation embedded in medical records and children's electronic health information. METHOD: Medical records were analyzed from 3,440 children ages 10 to 17 years with MH-related emergency department visits in a large academic health system over 2 years (2017-2019). Bivariate analyses and random forests were deployed to identify child-, neighborhood-, and systems-level correlates of involuntary MH detentions. RESULTS: Nearly 1 in 4 (n = 769, 22.4%) visits involved an involuntary detention. Half of detained children (n = 357, 46.4%) arrived on a detainment that was discontinued after MH provider evaluation. Odds of detention were greater among Black (odds ratio 1.33 [95% CI 1.02-1.73]) and publicly insured (odds ratio 1.63 [95% CI 1.37-1.94]) children. Children detained in prehospital settings resided in census tracts with greater social vulnerability scores (χ(2) 13.42, p < .001). Machine learning classifiers (area under the curve 0.83, [95% CI 0.81-0.84]) revealed that strongest indicators of detainment included psychiatric chief concern, prior year psychiatric hospitalization, Social Vulnerability Index, and ICD-10-CM code for suicide or self-harm. CONCLUSION: Medical record-embedded legal documentation supports transparency in the use of detentions, which are common and jointly predicted by children's clinical need and social vulnerability.