Skip to main content
. 2015 Apr 21;16:50. doi: 10.1186/s12875-015-0261-6

Table 3.

Results of logit models of decision to refer/not refer

Variables Model 1a Model 1b
Intercept −1.062*** −13.12***
(0.371) (4.503)
Nevus 5.008*** 5.901***
(0.474) (0.737)
Melanoma 5.549*** 5.409***
(0.598) (0.680)
Psoriasis 1.746*** 1.761***
(0.279) (0.358)
Waiting time 0.000184 0.0000828
(0.000938) (0.00118)
Distance −0.00403 −0.00286
(0.00325) (0.00411)
Pressure 0.0671 0.219
(0.246) (0.316)
Age 0.606**
(0.245)
Male −0.186
(0.650)
Distance hospital 0.0400**
(0.0162)
Distance private −0.0293
(0.0214)
List size 0.0000163
(0.0000227)
Telemedicine −1.911***
(0.603)
Special interest −1.149**
(0.526)
Health status: bad 0.423
(0.481)
Health status: good −1.698
(1.377)
Age-squared −0.00680**
(0.00293)
Observations 721 473
AIC 514.1 327.2
Log likelihood −249.1 −145.6
Chi-squared 152.5*** 95.37***
Hosmer & Lemeshow Chi-squared 12.22 13.84*
% pred. correctly 82.25% 86.47%
Area under ROC 0.8763 0.9215

***significant at 1% level; **significant at 5% level; *significant at 10%. Dependent variable is a dummy variable indicating whether a referral was made. Base categories for explanatory dummy variables: Need – Keratosis; Pressure – No; Telemedicine – No; Special interest – No; Health status – Neither good nor bad; standard errors in parenthesis.