Skip to main content

Table 3 Estimation of preferences from the mixed logit model

From: Investigating patients’ preferences for new anti-diabetic drugs to inform public health insurance coverage decisions: a discrete choice experiment in China

Attributes

β coefficients

RI

Mean (SE)

SD (SE)

HbA1c control

16.24

  Not as expected (ref)

-0.724***(0.073)

   

  As expected

0.724***(0.073)

0.825***(0.102)

 

Serious side effects

7.39

  Sometimes (ref)

-0.377***(0.072)

   

  Occasionally

0.096 (0.061)

-0.031 (0.135)

 

  Never or rarely

0.281***(0.061)

-0.235 (0.208)

 

Length of extended life

26.48

  Extended life (per year)

0.787***(0.074)

0.991***(0.091)

 

Change in HRQoL

34.25

  Worse (ref)

-1.670***(0.124)

   

  No improvement

0.287***(0.070)

-0.432**(0.138)

 

  Improvement

1.383***(0.110)

1.121***(0.109)

 

Dosing frequency

3.27

  Twice a day (ref)

-0.132*(0.056)

   

  Once every other day

-0.028 (0.060)

0.423***(0.101)

 

  Once a week

0.160**(0.061)

-0.202 (0.183)

 

Out-of-pocket costs per month

12.38

  Cost (per CNY50)

-0.138*** (0.021)

0.327*** (0.030)

 

  Log likelihood

-2600.171

 

  Participants

639

 

  Observations

10,224

 
  1. RI, relative importance; Ref, reference; SE, standard error; SD, standard deviation; *p < 0.05; **p < 0.01; ***p < 0.001. Each participant completed eight pairs of DCE choice scenarios, and the number of observations per participant was 16. So there were 10,244 observations for 639 participants