Table 5.
Results of the multinomial logit model
| Attribute levels | Model A** | Model B*** | ||
|---|---|---|---|---|
| OR | (95%CI) | OR | (95%CI) | |
| ASC* | 1.33 | (1.10; 1.62) | 1.32 | (1.12; 1.55) |
| Out of pocket costs per year (per €100,-) | 0.74 | (0.62; 0.88) | 0.72 | (0.60; 0.87) |
| Reduction in number of exacerbations per year (per one exacerbation) | 2.48 | (2.08; 2.96) | 2.51 | (2.09; 3.02) |
| Multidose/unidose system | ||||
| Unidose system (reference level) | 1.00 | - | 1.00 | |
| Multidose system | 1.34 | (1.02; 1.74) | 1.37 | (1.02; 1.85) |
| Risk of side effects | ||||
| No side effects (reference level) | 1.00 | - | 1.00 | |
| Mild side effects | 0.64 | (0.43; 0.96) | 0.66 | (0.43; 0.99) |
| Moderate to severe side effects | 0.10 | (0.07; 0.14) | 0.09 | (0.06; 0.15) |
| Impact on CO₂ emissions | ||||
| High GWP (reference level) | 1.00 | - | 1.00 | |
| Low GWP | 3.39 | (2.56; 4.49) | 1.15 | (0.96; 1.37) |
| Systemic preference heterogeneity | ||||
| > 10 years working in general practice x low GWP | - | 1.58 | (0.72; 3.45) | |
| GP (excl. nurse practitioner/specialist) x low GWP | - | 2.48 | (1.15; 5.36) | |
| Member of organisation or network for environmentally friendly healthcare x low GWP | - | 0.87 | (0.35; 2.18) | |
| AIC | 672.79 | 616.58 | ||
Bold: Statistically significance at 5% level
Abbreviations: OR odds ratio, CI confidence interval, ASC alternative specific constant, GP general practitioner, AIC Akaike Information Criterion
*The alternative specific constant (ASC) took the left–right bias of the choice processes of the participants (i.e. participants always choosing option A or B) into account
**Model A includes the results of the multinomial logit model
***Model B includes the results of the multinomial logit model plus systematic preference heterogeneity with interaction terms
The coefficients from the models represent marginal utilities for different attribute levels and were converted from utilities to odds ratios (OR). A statistically significant OR (p-value < 0.05) indicates that the attribute level had an impact on the choice process of the participants. If the OR for an attribute level is higher than one, this indicates that this level is preferred over the reference level of the same attribute. The higher the OR, the greater the preference. For a correct interpretation of the results, it is important to notice the different units of measurement, e.g. out of pocket costs is a continuous variable that is measured per euro, whereas multidose system is a categorical variable that are compared to their reference level