Investigating preferences for support with life after stroke: a discrete choice experiment

Table 7 Subgroup analysis by health utility: patient sample

Attribute	Eq5d health utility < 0.69 (n = 39)							Eq5d health utility ≥ 0.69 (n = 40)
Attribute	Β	95% CI		P value	MRS (mins)	95% CI		Β	95% CI		P value	MRS (mins)	95% CI
Format of service
_One to one	0.9543*	0.7393	1.4900	0.0000	197.06	127.88	336.51	0.5233*	0.2431	1.0217	0	75.52	28.89	130.22
Service provider _HOSPITAL	0.1503	-0.0601	0.3849		n/s			0.2257*	0.0427	0.5280		32.57	5.40	65.12
_Community health team	-0.0272	-0.2704	0.2375	0.8080	n/s			0.1234	-0.0824	0.3940	0.248	n/s
_Voluntary organisation	-0.1232	-0.4034	0.0987	0.2580	n/s			-0.3491*	-0.7099	-0.2000	0.001	-50.38	-86.41	-22.30
Journey time (minutes)	-0.0048*	-0.0081	-0.0033	0.0000				-0.0069*	-0.0113	-0.0059	0
Additional social and leisure
_Provided	0.3858*	0.1137	0.8241	0.0060	79.66	20.91	168.99	0.3715*	0.1274	0.7776	0.008	53.61	13.84	99.18
Constant	-0.1424	-0.5636	0.1774	0.3540				-0.2474	-0.7071	0.1181	0.127
No. observations = 323								No. observations = 330
No. individuals = 39								No. individuals = 40
Wald chi2 (5) = 59.09								Wald chi2 (5) = 62.71
Log likelihood = -179.45								Log likelihood = -185.21856

*Statistically significant at P < 0.05.
95% confidence intervals generated using non-parametric bootstrapping.
Marginal rate of substitution values = β-coefficient for significant attribute/β-coefficient for journey time.
Service provider base case (Hospital stroke team) calculated by assuming estimate for effects coded omitted variable = -1*(sum of estimated levels).

ISSN: 1472-6963