Table 3.

Average marginal effects (logistic regression - Model C)

	Marginal effect	95%CI
CANTON
Aargau (reference)
Vaud	-0.040	[-0.045 -0.034]
Ticino	-0.260	[-0.265 -0.255]
PATIENTS' VARIABLES Female	0.003	[-0.001 0.007]
Age:
0-19	0.031	[0.022 0.040]
20-39	0.035	[0.028 0.041]
40-59	0.016	[0.011 0.022]
60-79 (reference)
80+	-0.005	[-0.013 0.002]
Optional health insurance deductible	0.007	[0.003 0.012]
Out-of-pocket participation	0.023	[0.018 0.028]
Treatment complexity:
1-5 drugs	0.014	[0.009 0.020]
6-10 drugs (reference)
11-15 drugs	-0.009	[-0.015 -0.004]
16-20 drugs	-0.017	[-0.024 -0.010]
21+ drugs	-0.029	[-0.037 -0.021]
PHYSICIANS' VARIABLES Specialist prescriber	-0.032	[-0.037 -0.027]
Hospital prescriber	-0.131	[-0.137 -0.125]
SUBSTITUTION GROUPS' VARIABLES Number of generics:
1-2 (reference)
3-4	0.083	[0.075 0.090]
5-6	0.125	[0.118 0.133]
7+	0.085	[0.076 0.095]
Number of treated patients:
1-1000 (reference)
1001-2000	0.018	[0.012 0.025]
2001-3000	0.071	[0.060 0.082]
3001+	0.002	[-0.006 0.010]
Price of brand medication:
1st quintile	0.031	[0.023 0.039]
2nd quintile	-0.001	[-0.011 0.009]
3rd quintile (reference)
4th quintile	0.012	[0.004 0.020]
5th quintile	-0.011	[-0.019 -0.002]
Relative price difference:
1st quintile	-0.050	[-0.058 -0.042]
2nd quintile	-0.028	[-0.036 -0.020]
3rd quintile (reference)
4th quintile	0.004	[-0.004 0.013]
5th quintile	0.038	[0.028 0.048]
Original with special galenic form	-0.077	[-0.088 -0.066]
At least one drug sold under license	-0.051	[-0.058 -0.043]
Multiple package sizes	-0.039	[-0.045 -0.032]

Notes:

* p < 0.05 ** p < 0.01 *** p < 0.001

The marginal effect for discrete and factor variables is the discrete change from the reference level