Table 3.

Parameter estimates for the ESS scores of the models predicting EQ-5D utility scores

Model	Estimate	SE	P
(a) GLM—ESS score as a continuous variable	-0.0068	0.0009	< 0.001
(b) GLM—US/RoW cutoffs (reference 0–10)
Mild EDS (ESS, 11–12)	-0.0134	0.0174	0.44
Moderate EDS (ESS, 13–15)	-0.0505	0.0165	0.002
Severe EDS (ESS, 16–24)	-0.1132	0.0159	< 0.001
(c) GLM—UK cutoffs (reference 0–10)
Mild EDS (ESS, 11–14)	-0.0268	0.0138	0.05
Moderate EDS (ESS, 15–18)	-0.0789	0.0167	< 0.001
Severe EDS (ESS, 19–24)	-0.1513	0.0236	< 0.001
(d) Piecewise linear with breakpoint at 11.29
ESS Slope 1 (ESS, < 11.29)	-0.0028	0.0018	0.13
ESS Slope 2 (ESS, > 11.29)	-0.0134	0.0031	< 0.001
(e) Linear spline with breakpoint at 11
ESS, 0–11	-0.0026	0.0016	0.11
ESS, 12–24	-0.0131	0.0022	< 0.001

Note: P-values for models a-c were calculated using Wald chi-square tests. P-values for models d-e were calculated using t-tests. Due to no adjustments for multiplicity, P-values presented are nominal

EDS Excessive daytime sleepiness, ESS Epworth Sleepiness Scale, GLM Generalized linear model, RoW Rest of world, SE Standard error, UK United Kingdom, US United States