Fig. 1

Estimated causal effects of the DIP on inpatient costs, length of stay, and in-hospital mortality using the SDID method. Notes: Each panel of each column is a separate SDID estimates without covariates. In the first row, we show trends in inpatient costs, length of stay, and in-hospital mortality for Guangzhou and synthetic Guangzhou in 2012–2020. The weights used to synthesize pre-treatment periods of candidate regions are presented at the bottom of the graphs. The red vertical bar indicates when DIP starts to be implemented. In the second row, we show the region-by-region adjusted outcome difference. The weights are indicated by dot size. The weighted average of these differences, namely the estimated effect, is indicated by a horizontal line. Observations with zero weight are denoted by an × -symbol