Posttraumatic Stress Dynamics and 15-Year Persistence in Great East Japan Earthquake Survivors.

IMPORTANCE: Long-term persistence of posttraumatic stress reactions (PTSRs) is a major public health challenge, yet early prognostic markers for decadal outcomes remain poorly characterized. Specifically, the importance of early symptom fluctuations compared with initial cross-sectional severity is not well understood. OBJECTIVE: To determine whether early postdisaster symptom dynamics within 1 year are associated with persistent, clinically meaningful PTSRs 15 years after the 2011 Great East Japan Earthquake (GEJE). DESIGN, SETTING, AND PARTICIPANTS: This 15-year prospective longitudinal cohort study (2011-2025) included residents of Shichigahama, Miyagi, Japan, aged 20 years or older, who experienced disaster-certified housing destruction after the GEJE. EXPOSURES: Early symptom dynamics (2011-2012) based on Impact of Event Scale-Revised (IES-R) scores and classified as (1) high-escalating (IES-R score ≥25 and increasing), (2) high-improving (IES-R score ≥25 and nonincreasing), and (3) nonpersistent early (IES-R score <25 at either wave). MAIN OUTCOMES AND MEASURES: The primary outcomes were persistent, clinically meaningful PTSRs (IES-R score ≥25) at 15 years. Multivariable logistic regression with 40-fold multiple imputation was used to estimate adjusted odds ratios (aORs). RESULTS: Among 1291 participants (685 women [53.1%]; mean [SD] age, 53.9 [16.6] years), group-based trajectory modeling of IES-R scores (2011-2025) identified 5 distinct PTSR patterns. At 15 years, 14.7% of participants (190 participants) had clinically meaningful PTSRs. Symptom escalation was most frequent during the first-to-second year interval compared with all subsequent periods. This trend remained consistent even under stringent escalation criteria (>5-point increase). The high-escalating pattern was the one most associated with 15-year persistence (aOR, 5.31; 95% CI, 3.85-7.32). The prognostic model incorporating early dynamics demonstrated improved discrimination (area under the curve [AUC], 0.756; 95% CI, 0.717-0.763) compared with models using baseline symptom severity alone (AUC, 0.608; 95% CI, 0.600-0.612) or delayed single-point assessments (AUC, 0.726; 95% CI, 0.663-0.738). CONCLUSIONS AND RELEVANCE: In this 15-year cohort study of GEJE survivors, symptom escalation during the first-to-second year interval was identified as the most prevalent pattern and was a factor associated with long-term persistent PTSRs. Because early symptom dynamics offer superior prognostic accuracy compared with single-point assessments, these findings support a serial screening approach during the immediate postdisaster period to optimize risk stratification and early intervention strategies.