How to Implement Adherence-Promoting Programs in Clinical Practice? A Discrete Choice Experiment on Physicians’ Preferences

Figures & data

Figure 1 An example of the applied choice sets. The DCE card shows two different hypothetical APP alternatives derived from a combination of the defined attributes and the respective levels. Physicians needed to decide between option A, B or an opt-out option.

Notes: The figure shows one choice set as presented to participating physicians. All in all, physicians were asked to decide on 16 different choice sets.

Table 1 Characteristics of Interviewed Physicians

	Overall Sample	GPs	Cardiologists	Neurologists	Ophthalmologists
N (%)	222 (100.0%)	47 (21.2%)	47 (21.2%)	83 (37.4%)	45 (20.3%)
Age in years - mean (Valid N | SD)	49.6 (164 | 8.8)	50.9 (36 | 8.9)	48.9 (34 | 8.8)	51.1 (66 | 7.9)	44.9 (28 | 9.4)
Female gender – N (%)	81 (36.5)	25 (53.2)	7 (14.9)	32 (38.6)	17 (37.7)
Type of medical practice
Single practice – N (%)	74 (37.6)	17 (36.2)	12 (25,5)	31 (37.3.)	15 (33.3)
Joint/community practice – N (%)	78 (39.6)	17 (36.2)	19 (40,4)	27 (32.5)	14 (31.1)
Practice cooperation - N (%)	10 (5.1)	2 (4.3)	1 (2.1)	5 (6.0)	2 (4.4)
Medical care center– N (%)	28 (14.2)	2 (4.3)	8 (17.0)	9 (10.8)	9 (20.0)
Others– N (%)	7 (3.6)	1 (2.1)	2 (4.3)	4 (4.8)	0 (0.0)
Patients per quarter – mean (Valid N | SD)	1,691.8 (159 | 1526.8)	1,579.1 (35 | 1177.9)	1,667.7 (31 | 1537.9)	1,263.3 (67 | 1074.1)	2,976.5 (26 | 2175.9)
Total number of employees – mean (Valid N | SD)	9.4 (154 | 10.5)	6.3 (35 | 5.0)	10.3 (32 | 8.3)	7.0 (60 | 6.3)	17.3 (27 | 18.9)
Years since start of practice – mean (valid N | SD)	13.1 (169 | 9.1)	15.2 (38 | 10.2)	10.4 (34 | 8.2)	13.7 (67 | 8.9)	12.0 (30 | 8.8)
Number of physicians – mean (Valid N | SD)	2.6 (158 | 2.3)	2.1 (35 | 1.7)	2.6 (34 | 2.1)	2.4 (62 | 2.4)	3.5 (27 | 2.9)
Location of the practice
City Center – N (%)	124 (63.9)	18 (38.3)	30 (63,8)	48 (57.8)	29 (64.4)
Countryside – N (%)	23 (11.9)	8 (17.0)	4 (8.5)	5 (6.0)	6 (13.3)
Suburban – N (%)	33 (17.0)	11 (23.4)	5 (10.6)	14 (16.9)	3 (6.7)
Medical Center – N (%)	14 (7.2)	4 (8.5)	1 (2.1)	7 (8.4)	2 (4.4)

Table 2 Logistic Regression Exploring the Probability to Choose the Opt-Out Option

			Odds Ratio	95% Confidence Interval	p-value
Attribute Levels		N of Decisions
Intervention tool	No option had the “comprehensive package tool”	888	Reference
	At least one option had the “comprehensive package tool”	2,664	0.789	0.649–0.960	0.018
Time commitment	No option was “10 mins of the staff”	888	Reference
	At least one option was “10 mins of the staff”	2,664	0.396	0.328–0.477	p<0.001
Compensation	No option was “80€”	1,776	Reference
	At least one option was “80€”	1,776	0.512	0.431–0.609	p<0.001
Physician characteristics		N of physicians
Age ≤50 years		140	Reference
Age >50 years		82	0.938	0.751–1.172	p>0.100
Male gender		141	Reference
Female gender		81	1.187	0.984–1.433	0.073
GP		47	Reference
Cardiologist		47	0.916	0.692–1.213	p>0.100
Neurologist		83	0.787	0.621–0.998	0.048
Ophthalmologist		45	1.253	0.958–1.640	0.099
≤10 years of practice		84	Reference
≤20 years of practice		46	1.164	0.915–1.481	p>0.100
>20 years of practice		38	0.560	0.421–0.745	p<0.001
Single practice		148	Reference
Other type of medical practice		74	1.367	1.133–1.649	0.001

Figure 2 The results of the conditional logit estimation based on 222 physicians. For each attribute describing alternative APP options, its relative importance for overall (hypothetical) physicians’ decisions for/against a specific APP is reported in % (light grey areas). Furthermore, utility per attribute level including significance of utility differences in comparison to the reference attribute level is shown (black bars).

Figure 3 The relative importance of the different attributes across subgroups.