Skip to main content
. Author manuscript; available in PMC: 2016 Oct 24.
Published in final edited form as: Am Econ Rev. 2016 Aug;106(8):2110–2144. doi: 10.1257/aer.20151080

Table 5.

Choice Model of Patient Allocation across Conditions

Condition (1) (2) (3) (4)
AMI HF Pneumonia Hip/Knee
Mean Miles to Chosen Hospital 12.48 8.27 7.49 13.16
SD Miles to Chosen Hospital 20.06 13.25 11.92 18.85
Risk-Adjusted Survival
MRS(1 pp risk-adjusted survival, miles) −1.793
(0.158)
−1.029
(0.129)
−0.378
(0.057)
Patients 165,005 275,671 317,904
Risk-Adjusted Readmission
MRS(1 pp risk-adjusted readmission, miles) 1.138
(0.173)
1.040
(0.122)
0.451
(0.109)
2.385
(0.268)
Patients 158,086 274,667 317,374 222,673
Process of Care Z-Score
MRS(1 SD process of care, miles) −4.418
(0.383)
−2.238
(0.221)
−1.325
(0.110)
Patients 158,032 270,773 309,623
Patient Survey Z-Score
MRS(1 SD patient survey, miles) 0.324
(0.388)
−0.093
(0.205)
0.036
(0.151)
−1.604
(0.382)
Patients 167,429 266,915 298,185 224,451

This table reports the marginal rates of substitution (MRSs) of quality for distance derived from the conditional logit model (see equation 6). For the survival and readmission rates, the MRS given by equation (6) is divided by 100 to put it into percentage point terms. Only one quality measure is used at a time in each logit model. Standard errors are analytic and clustered at the market level.

The sample is all patients with the condition in 2008 who stayed in their market of residence for treatment. The choice set for a patient is all hospitals in his market with the quality measure available that treated at least one patient in 2008. The mean and SD miles statistics are taken from the patients in the column’s risk-adjusted survival sample (risk-adjusted readmission for hip/knee replacement). All MRSs in a column are evaluated at this mean.