COMPARISON OF VALUATIONS BY PATIENTS, HEALTH CARE PROFESSIONALS, AND LAYPERSONS USING MULTIPLE-OUTCOME HEALTH STATE VIGNETTES

PURPOSE To compare feasibility and assigned values for health states of patients, health care professionals and laypersons. Aim is to gain insight in different valuation patterns across valuation methods and respondent groups using obstetric health state vignettes that involve both maternal and neonatal outcomes.

METHODS Participants from three groups (patients (n=24); health care professionals (n=30); laypersons (n=27)) valued 40 individual vignettes with the visual analogue scale (VAS) and time trade-off (TTO) methods, and 20 paired vignettes in a discrete choice experiment (DCE). Each vignette covered five attributes: maternal health ante partum, time between diagnosis and delivery, process of delivery, maternal outcome, and neonatal outcome. Generalization theory was used to determine relative importance of attributes, respondent group and rating method. Test-retest reliability was assessed by intra class correlation (ICC) for VAS and TTO and Cohen’s kappa (κ) for DCE. Within-group consistency of VAS and TTO was assessed with ICC and ANOVA. Relative weights from each rating method were compared per group using Kendall’s Tau-b (τ).

RESULTS All groups ranked DCE as the easiest and TTO most difficult method. Of total variance in VAS valuations, 66% was explained by vignette and 15% by respondent group (TTO: 62% and 20% respectively). Test-retest reliability was highest for VAS (ICC=0.61-0.73) and lowest for DCE (κ=0.15-0.37) in all groups. Within-group reliability was highest for VAS (ICC=0.70-0.73), intermediate for DCE (κ=0.56-0.76) and lowest for TTO (ICC=0.20-0.66). Correlations between relative attribute weights were highest for VAS (τ=0.64-0.76) and lowest for DCE (τ=0.12-0.41) for all respondent group comparisons.

CONCLUSION While the effect of respondent group appears prominent, the effect of valuation method is dominant. Valuation differences between groups are largest for the severe health states.