In contrast, in the post-implementation phase, 80 patients had 277 episodes over 1,496 person-days (18.5 events per 100 person-days), representing a highly significant decrease (χ 2 (1)=145.26, p <0.001) following PUSH implementation.
This reduction was marginally non-significant in the crude analysis (IRR 0.751, 95% CI 0.5411–1.043, p =0.087) but became significant after adjusting for age and sex (IRR 0.667, 95% CI 0.474–0.938, p =0.020).
16 Although non-stroke patients showed a significant reduction in restraints, trends for reduced delirium and harm events did not reach statistical significance, possibly owing to lower statistical power.