Session-by-session outcome trajectories in disaster counselling: Testing dose-effect and good-enough level models in South Korea

OBJECTIVE: Evidence comparing dose-effect and good-enough level (GEL) models is drawn largely from Western outpatient and university counselling samples treating common mental health problems. It therefore remains unclear whether these models generalize to non-Western, post-disaster counselling contexts. This study examines session-by-session outcome trajectories among South Korean survivors of the humidifier disinfectant disaster and tests whether these patterns align with dose-effect and GEL predictions. METHODS: We analyzed routine outcome monitoring data from humidifier disinfectant disaster survivors in South Korea (180 clients; 2307 sessions; 105 therapists). Outcome Rating Scale (ORS) scores were modelled with multilevel growth curves and compared across aggregate and treatment-length-stratified specifications. RESULTS: A negatively accelerating log-linear trajectory best fit the data, showing rapid early improvement followed by decelerating gains. Stratified models revealed a significant interaction between session progress and treatment length; brief care (≤5 sessions) typically did not reach the ORS clinical cutoff (25) despite exhibiting the steepest initial improvement rate. Reliable and clinically significant change peaked in the 11-20 session group compared to the brief care group, whereas individuals attending more than 20 sessions showed sustained but relatively stabilized recovery pattern. CONCLUSIONS: Findings support complementary dose-effect and GEL perspectives, highlighting the inadequacy of rigid session caps for highly distressed survivors. The higher odds of recovery in moderate-to-long-term group (11-20 sessions) emphasize that survivors require a sufficient dose to consolidate gains. These results underscore routine outcome monitoring and flexible care structures as practical tools for dynamically tailoring treatment lengths in national-scale trauma counselling programmes.