Aim: To map health-related quality of life (Qol) with clinical parameters BASFI and ASDAS-CRP measure, and other covariates.
Methods: Our prospective multicenter non-interventional observation study of ankylosing spondylitis (AS) collected data about QoL and clinical outcomes on the initial and four subsequent visits. We employed simple linear regression analysis of a cross-sectional dataset, and fixed effect, random effect and pooled linear regression of a longitudinal dataset.
Results: We showed that BASFI and ASDAS-CRP are very strong, robust predictors of EQ-5D utilities in all regression specifications together with sex (female), invalidity, and activity impairment. Additionally, the longitudinal regression analysis showed that a fixed effect model may be a viable alternative to the most commonly used random effect model or pooled linear regression due to the nature of our dataset.
Conclusion: This is one of the first studies using a fixed effect model in longitudinal patient-level data, although, this method has been widely used in economics.
Keywords: Ankylosing spondylitis; fixed effect regression; health-related quality of life; longitudinal analysis; mapping; outcomes research; regression analysis.