Ecological momentary assessments predict clinical anxiety and depression status at 3 months post-stroke

Background: Stroke survivors frequently experience post-stroke depression (PSD) and post-stroke anxiety (PSA), which may be underexamined by clinicians if patient mood status between discharge and follow-up is unknown. Digital monitoring may enable early identification of mood disturbances. Ecological momentary assessments (EMA) have been tested to capture mood fluctuations in patients regular environments, but no standardized procedure for monitoring PSA or PSD has been developed. Objective: Determine the feasibility and validity of EMA for evaluating subacute PSA and PSD in non-severe stroke patients. Methods: Patients with non-severe stroke discharged home responded to daily smartphone EMA questions concerning mood and activity for twelve weeks until follow-up. Questions based on symptoms of anxiety and depression were used to derive EMA Anx and EMA Dep indices, computed over five 15-day intervals. Convergent validity was assessed using follow-up Generalized Anxiety Disorder-7 (GAD-7) and Centre for Epidemiological Studies-Depression (CES-D) scores. Logistic regression models predicted PSA and PSD status at follow-up, with baseline models comprising stroke severity, age, sex, and mood status, to which EMA Anx or Dep were added. Results: Data from 118 patients were analyzed. EMA adherence was moderate, with 89 patients (75.4%) responding on 75 or more days. Correlations between EMA Anx and GAD-7 and between EMA Dep and CES-D generally strengthened at later intervals, with the average mood index generally showing the strongest associations. Adding EMA data from the third time interval (31-45 days) significantly improved model fit beyond the baseline models (Anxiety: {chi}-square = 7.08, p = 0.008, {delta}R-square = 0.060; Depression: {chi}-square = 28.0, p < 0.001, {delta}R-square = 0.256). Discussion: EMA is feasible in mild-to-moderate stroke patients and may facilitate early identification of post-stroke mood disturbances. Further analyses in more diverse populations are required to validate these findings.