A next-generation approach to mental health outcomes: Treatment, time, and trajectories

Over the last several decades, inpatient psychiatric length of stay (LOS) has been greatly reduced to the detriment of patients. Latent variable mixture modeling, can be used to improve the quality of care for patients by identifying unobserved subgroups and optimize treatment variables, including LOS. This study had three objectives (1) to replicate the findings made by Oh et al. in a distinct sample, (2) to examine demographic differences related to inpatient treatment trajectories, and (3) to relate additional variables to each trajectory. We collected data on six key mental illness factors and information on felonies, misdemeanors, history of stopping psychiatric medication and psychotherapy, length of time in psychotherapy, and the number of therapists and psychiatrists from 489 patients at an inpatient psychiatric hospital. We derived latent mental illness scores after applying growth mixture modeling to these data. We identified three distinct trajectories of mental illness change: High-Risk, Rapid Improvement (HR-RI), Low-Risk, Partial Response (LR-PR), and High-Risk, Gradual Improvement (HR-GI). The HR-GI group was more likely to have patients who were female, Asian, younger, Yearly Income (YI) <$20,000, that spent more time in psychotherapy throughout their life, and had the longest LOS while inpatient. The LR-PR group had was more likely to be male, Hispanic/Latino and multiracial, older, YI >$500,000, have a history of misdemeanors, and this group had the shortest LOS (p < .05). These findings replicate and extend our previous findings in Oh et al. (2020a) and highlight the clinical utility of agnostically determining the treatment trajectories.