Table 1 Model overview

Model I: ISI 7 items

$Utilit{y}_{modelI}=1-exp\left(\alpha +\sum _{j=0}^7{X}_i{\beta }_{ij}\right)$ Where $\sum _{j=0}^0{X}_i{\beta }_{i0}\equiv 0$and, i = 1,...,7 representing 7 ISI items, j = 0,...,4, representing 5 levels of the Likert scale.

Model II: ISI summary scores (0-28)

$Utilit{y}_{ModelII}=1-exp\left(\alpha +X\beta \right)$ Where X is the ISI summary scores range from 0(min) to 28 (max) and treated as a continuous variable in the model

EQ-5D disutility values

$Disutility=exp\left(\alpha +\sum X\beta \right)$ Utility = 1-Disutility

Model III: ISI 4 clinical categories

$Utilit{y}_{ModelIII}=1-exp\left(\alpha +\sum _{j=1}^{4}X{\beta }_{ij}\right)$ Where $\sum _{j=1}^1{X}_i{\beta }_{i1}\equiv 0$and, j = 1,...,4 representing 4 clinical classifications based on insomnia summary scores.

Model IV: Insomnia symptoms and demographics

$Utilit{y}_{ModelIV}=1-exp\left(\alpha +\sum _{i=1}^8{X}_i{\beta }_i+X_{duratio{n}^2}^2{\beta }_{duratio{n}^2}\right)$ Where i = 1,...,8 representing 5 symptoms, 2 demographic variables and 1 variable indicating the presence/absence of comorbidity. The 5 symptom variables were: sleep duration, quality, latency, next day sleepiness, and number of wakeup times during the night. The 2 demographic variables were age and gender. Within the model, a sleep duration squared term was an additional predictor adjusting the concave function of sleep duration on health state utilities.