Table 3.

Difference-in-differences regression results

	Year 1				Year 2
	ATT	SE	%a	Absoluteb	ATT	SE	%a	Absoluteb
Cost outcomes
Third party payer
Pharmaceutical costs	0.008	0.038	0.803	8.065	− 0.040	0.043	− 3.921	− 41.503
Of these due to T2D	0.055***	0.018	5.654	24.173	0.001	0.035	0.100	31.582
Outpatient costs	− 0.065*	0.036	− 6.293	− 53.241	− 0.024	0.039	− 2.371	− 18.909
Inpatient costs	0.058	0.166	5.971	− 47.053	0.002	0.168	0.200	1.788
Other costs	− 0.091	0.143	− 8.698	− 128.662	− 0.114	0.146	− 10.774	− 184.358
Total costs	− 0.047	0.030	− 4.591	− 184.686	− 0.033	0.032	− 3.246	− 136.302
Patients
Co-payments	− 0.160***	0.023	− 14.786	− 20.179	− 0.248***	0.042	− 21.964	− 26.925
Of these due to T2D	− 0.330***	0.016	− 28.108	− 16.881	− 0.438***	0.036	− 35.467	− 16.931
Observations	2389				2389

ATT average treatment effect on the treated, SE standard error, T2D type 2 diabetes mellitus

*p < 0.10, **p < 0.05, ***p < 0.01

^aLog-level model estimates proxy of percentage change. Exact percentage change is calculated as ($\exp \left(A,T,T,),-,1\right)\times 100\%$

^bFor pharmaceutical costs, for example, the absolute change in year 1 is computed as follows: $(0.008\times 1016.13)/(1+0.008)$, where 1016.13 refers to the mean pharmaceutical cost for the fixed-dose combination cohort in year 1. The corresponding mean values can be found in Table S3 of the OSM