Table 3

Results for triple difference estimation (%)

 logTotal Model 1(CCG deficit) Model 2(CCG demonstrator) Model 3(NHS hospitals) Phase I 0.69***(0.67 to 0.71) 0.57***(0.56 to 0.59) 3.90***(3.79 to 3.99) EBI −13.07***(−14.74 to 11.40) −12.03***(−13.75 to –10.31) −9.69***(−12.48 to –6.90) EBI×phase I 0.13***(0.11 to 0.14) 0.10***(0.09 to 0.11) 0.48***(0.39 to 0.56) EBI×X 2.26(−0.02 to 4.53) −1.24(−3.43 to 0.94) −2.79(−6.85 to 1.27) Phase I×X −0.30***(−0.32 to to 0.28) −0.05***(−0.07 to –0.02) −3.08***(−3.19 to –2.98) EBI×phase I×X −0.06***(−0.08 to to 0.04) 0.00(−0.02 to 0.02) −0.35***(−0.45 to –0.26) Constant 4.15***(4.05 to 4.25) 4.18***(4.08 to 4.29) 2.84***(2.76 to 2.93) Observations 6750 6750 16 559 Units of observation 135 CCGs 135 CCGs 382 hospitals
• Co-efficients are reported as percentages, and can be understood as follows: phase I is the percentage difference in change in volumes between the treatment and control groups. EBI is the percentage difference in volumes for all procedures before and after the implementation of the EBI programme. EBI×phase I is the percentage DiD of volumes before and after the implementation of the EBI programme between the treatment and control groups. EBI×X is the average percentage DiD in volumes for all procedures before and after the implementation of the EBI programme between different organisational characteristics defined by X. Phase I×X is average percentage difference in changes in volumes for the treatment group for different organisational characteristics defined by X. EBI×phase I×X is the average percentage DiD in volumes before and after the implementation of the EBI programme for the treatment group between different organisational characteristics defined by X. In model 1, X is 1 for CCGs which posted a deficit in financial year 2018/19, and 0 for those which did not. In model 2, X is 1 for CCGs which were part of the demonstrator community, and 0 for those which were not. In model 3, X is 1 for NHS hospitals and 0 for independent sector providers. 95% CIs in parentheses. All models used fixed effects, therefore errors are clustered at the CCG or hospital level. This table was created by the coauthors of this manuscript.

• *P<0.05; **p<0.01; ***p<0.001.

• CCG, clinical commissioning group; EBI, evidence-based interventions programme; NHS, National Health Service.