Fig. 5From: An opportunistic evaluation of a routine service improvement project to reduce falls in hospitalForest plot of the post-intervention time trend in each ward. To explore differential intervention effects across wards, a Poisson regression model was fitted to each ward individually. The estimate here indicates the post-intervention change each month in the rate of falls compared to the previous month. The overall estimate was obtained by fitting a mixed-effect Poisson regression model to the entire dataset. This model included a fixed effect for: intervention, time, and a time by intervention interaction; and random effects for: ward, time, and post-intervention time. IRR: Incidence rate ratio. CI: Confidence intervalBack to article page