Cardiff University | Prifysgol Caerdydd ORCA
Online Research @ Cardiff 
WelshClear Cookie - decide language by browser settings

PMS2 Estimating health-based utility from clinically assessed disease severity in ankylosing spondylitis [Abstract]

Poole, Christopher David, Singh, A., Freundlich, B., Koenig, A. and Currie, Craig John 2010. PMS2 Estimating health-based utility from clinically assessed disease severity in ankylosing spondylitis [Abstract]. Value in Health 13 (3) , A122. 10.1016/S1098-3015(10)72590-0

Full text not available from this repository.


OBJECTIVES: We sought to conduct a statistical mapping analysis between a standard investigator assessment of disease severity in ankylosing spondylitis (AS) and domain responses in a standard index of health utility; and secondly, to implement the above mapping in an optimised algorithm to estimate utility. METHODS: Multinomial logistic regression was used to estimate response probabilities to each domain of the EQ5D from the Bath Ankylosing Spondylitis Metrology Index (BASMI) among patients enrolled into an RCT studying the use of either etanercept infusion vs. oral sulphasalazine (ASCEND). Other covariates tested were gender, age, co-morbidity, AS duration, DMARD history, and concurrent medications. Predicted EQ5Dindex was estimated by Monte Carlo bootstrap simulation. The predictive ability of the response mapping was assessed by comparing estimated and directly measured utility derived from the UK tariff. RESULTS: Evaluable data were available for 566 predominantly white (87%) patients, 74% of whom male, with a mean baseline age of 41 years (sd 12) and median AS duration 4 years (IQR 1 to 11). Average BASMIlinear was 4.1 (sd 1.8) whilst median observed EQ5D utility was 0.587 (IQR 0.193 to 0.691). The linear definition of the BASMI was optimal in an algorithm that also adjusted for gender, AS duration, number of co-morbid body systems, number of historic DMARDs, number of current non-DMARD drugs, and current NSAID use. The mean utility predicted by the optimized algorithm was 0.552 (sd 0.101) and 0.559 (sd 0.295) by estimation directly from EQ5D responses (p = 0.238). The mean squared error between the actual and predicted utilities was 0.076 (sd 0.111). Adjusted utility was defined by −0.044*BASMIlinear+0.715, with an R2 of 0.62. CONCLUSIONS: In this study, response mapping of AS disease activity to the EQ5Dindex produced reliable estimates of preference-based health-related utility. Future analysis will compare the relative ability of patient-reported, AS-specific, functional assessment measures in predicting health-related utility.

Item Type: Article
Date Type: Publication
Status: Published
Schools: Medicine
Subjects: R Medicine > R Medicine (General)
Additional Information: ISPOR Fifteenth Annual International meeting Research Poster Abstracts
Publisher: Wiley-Blackwell
ISSN: 1098-3015
Last Modified: 04 Jun 2017 03:14

Actions (repository staff only)

Edit Item Edit Item