Quasi-Experimental Methods

The matched control group came from adults registered with participating GP practices before 1 September 2020 who did not receive CLICS. A total of 1,304 discharged CLICS patients were matched to 5,216 controls. The study used a quasi-experimental dynamic staggered difference-in-differences design on a propensity-matched cohort because a randomized controlled trial was unavailable and treatment timing varied. The Callaway and Sant’Anna estimator was used to estimate treatment effects when treatment timing varied and to avoid biases from standard two-way fixed-effects methods. Nearest-neighbour matching used a 1:4 CLICS-to-control ratio based on demographics, deprivation, and clinical risk factors.