Skip to main content

Table 3 Two taste classes of the scale-adjusted latent class (SALC) model

From: Exploring the importance of controlling heteroskedasticity and heterogeneity in health valuation: a case study on Dutch EQ-5D-5L

 

Taste class 1

EQ-5D-5L values set

Taste class 2

garbage class

 

Coef

p-value

Coef

p-value

Mobility

Level 1–2

0.0586

 < 0.001

0.2954

0.004

Level 2–3

0.0290

 < 0.001

 − 0.1132

0.348

Level 3–4

0.1205

 < 0.001

0.1903

0.159

Level 4–5

0.0090

0.017

 − 0.0825

0.518

Self-care

Level 1–2

0.0553

 < 0.001

 − 0.2140

0.143

Level 2–3

0.0246

 < 0.001

0.2202

0.127

Level 3–4

0.0630

 < 0.001

 − 0.0745

0.531

Level 4–5

0.0082

0.030

0.2378

0.048

Usual activity

Level 1–2

0.0516

 < 0.001

0.0386

0.794

Level 2–3

0.0297

 < 0.001

 − 0.0073

0.964

Level 3–4

0.1115

 < 0.001

 − 0.0431

0.767

Level 4–5

0.0002

0.967

0.0544

0.745

Pain/discomfort

Level 1–2

0.0686

 < 0.001

0.1218

0.438

Level 2–3

0.0290

 < 0.001

 − 0.0898

0.585

Level 3–4

0.1085

 < 0.001

0.0930

0.575

Level 4–5

0.0346

 < 0.001

 − 0.1657

0.341

Anxiety/depression

Level 1–2

0.0558

 < 0.001

0.3042

0.037

Level 2–3

0.0412

 < 0.001

0.0648

0.643

Level 3–4

0.0753

 < 0.001

0.0241

0.885

Level 4–5

0.0262

 < 0.001

0.1455

0.354

Prob (11111 > 55555) **

.998

 

.554

 
  1. Coefficients are showing as incremental change in the level of severity on a pits scale where value (55555) = 0 and value (11111) = 1; Detailed results are in Appendix 1
  2. **The probability of choosing the best over the worst EQ-5D-5L profile is less than 56% in taste class 2 (calculating probability from the difference between v (11111) and v (55555) on a log-odds scale which is the pits value .2161; log (p/((1-p)) = 0.2161). In this study, taste class 2 is called the garbage class because the responses were unrelated to the ordinal attributes