Do patients’ preferences prevail in hospital selection?: a comparison between discrete choice experiments and revealed hospital choice

Table 3 Results of mixed logit models (main analyses)

		Breast cancer			cataract
		beta	SE ^a	P value	beta	SE ^a	P value
ASC ^b		0.810	0.064	< 0.01	0.611	0.052	< 0.01
Attributes
1) Patient experiences	Below average (ref.)
	Average	0.816	0.087	< 0.01	1.007	0.062	< 0.01
	Above average	1.515	0.091	< 0.01	1.217	0.069	< 0.01
2) Clinical outcome indicator
Breast cancer:	Tumor-positive resection margin (in %)	−0.150	0.010	< 0.01
Cataract:	Per-operatively performed vitrectomy (in %)				−2.567	0.130	< 0.01
3) Waiting time	(In working days)	−0.058	0.004	< 0.01	− 0.016	0.002	< 0.01
4) Travel distance	3 km (ref.)
	8 km	− 0.351	0.066	< 0.01	− 0.181	0.057	< 0.01
	15 km	−0.422	0.064	< 0.01	− 0.674	0.072	< 0.01
5) Recommendation	Nobody (ref.)
	Friends and Family	0.659	0.077	< 0.01	−0.029	0.053	0.58
	GP	1.259	0.088	< 0.01	0.507	0.062	< 0.01
SD of random parameters
1) Patient experiences	Average	0.764	0.120	< 0.01	0.227	0.130	0.08
1) Patient experiences	Above average	−0.855	0.114	< 0.01	− 0.857	0.086	< 0.01
2) Clinical outcome indicator
Breast cancer:	Tumor-positive resection margin (in %)	0.190	0.012	< 0.01
Cataract:	Per-operatively performed vitrectomy (in %)				2.679	0.135	< 0.01
3) Waiting time (in working days)		−0.072	0.004	< 0.01	− 0.033	0.003	< 0.01
4) Travel distance	8 km				−0.410	0.142	< 0.01
4) Travel distance	15 km				−1.143	0.108	< 0.01
5) Recommendation	Friends and Family	−0.811	0.117	< 0.01	0.280	0.151	0.06
5) Recommendation	GP	0.530	0.147	< 0.01	−0.783	0.095	< 0.01
Number of individuals		631			1109
Number of observations		6304			10,980
Model fit	Log-likelihood	− 2920.42			− 4915.45
Model fit	BIC	5972.07			9989.06

All random parameters were assumed to be normally distributed and were simulated using 5000 Modified Latin Hypercube Sampling draws
^areflects bootstrapped standard errors
^bcoded as 1 for the first (left) alternative and 0 for the second alternative. The coefficient reflects the utility derived from any given hospital presented on left hand side of the choice set and thus accounted for any left-to-right bias
ASC Alternative Specific Constant, BIC Bayesian information criterion, GP General Practitioner, KM Kilometer, NA Not Available, Ref. Reference, SD Standard deviation, SE Standard error

ISSN: 1472-6963