Journal of Rehabilitation Medicine 51-8 | Page 25

Psychosocial well-being after stroke Assessed for eligibility (n=670) Enrolment Eligible (n=353) Randomized (n=322) Allocation Allocated to intervention (n=166) 561 Excluded (n=317) Reasons: • Not meeting inclusion criteria (n=63) • Declined to participate (n=211) • Lost to competing studies (n=12) • Other reasons (n=31) Declined participation after initially giving consent (n=31) Reasons: • Missed data collection deadline (n=8) • Unavailable after discharge (n=5) • Medical conditions (n=5) • Death (n=4) • Diagnosis change, not longer eligible (n=2) • Overwhelmed (n=1) • Too healthy to participate (n=1) • Did not disclose reason (n=5) Allocated to control (n=156) Follow-Up T2, 6 months Lost to follow-up (n=14) Reasons: • Due to group allocation (n=2) • Other medical condition (n=3) • Unavailable after discharge (n=5) • Did not disclose reason (n=3) • Dead (n=1) Lost to follow-up (n=23) Reasons: • Due to group allocation (n=5) • Felt too healthy to participate (n=1) • Other medical condition (n=8) • Total burden of rehabilitation (n=2) • Unavailable after discharge (n=4) • Did not disclose reason (n=3) Analysis 6 months Intention to treat analysis (n=166) Complete cases (n=143) Intention to treat analysis (n=156) Complete cases (n=142) Fig. 1. Flow chart of the study. the intervention trajectory was complete (≥6interven- tion sessions); it began between 4 and 8 weeks (mean 49 days (n = 147)) after the stroke, and the frequency and total duration were a maximum of 17 weeks from session 1 to session 8. Primary outcome After dichotomizing the sum score, 99 of the 166 patients (59.6%) in the intervention group and 93 of the 156 patients (59.6%) in the control group had nor- mal mood (GHQ-28<5) at 6 months. Table II shows that, compared with that at baseline, the proportion of participants with normal mood increased in both the intervention and control groups. After controlling for the baseline characteristics and recruitment centre, in the logistic regression model, no benefit of the dialogue-based intervention was obser- ved over usual care on mood at T2 (OR: 0.898: 95% CI: 0.54–1.50, p = 0.680) . By separately exploring the results for the interven- tion and control groups, it was found that no baseline characteristics demonstrated statistically significant effects on mood at T2 in the intervention group. Two baseline factors affected the odds of a normal mood at T2 in the control group. Higher SOC (OR: 1.098, 95% CI: 1.01–1.19, p = 0.026) increased the odds of normal mood, while comorbidities (OR: 0.282, 95% CI: 0.09–0.83, p = 0.022) decreased the odds of normal mood. This difference should be further explored when the 12-month data are available. J Rehabil Med 51, 2019