Modern Psychometric Approaches to Analysis of Scales for Health-Related Quality of Life

Abstract

In recent years, much effort has been invested in the development of new instruments for assessment of health-related quality of life (HRQOL). For many new instruments, modern psychometric methods, such as item response theory (IRT) models, have been used, either as supplemental to classical psychometric testing or as the primary methodological approach. We will use the term modern psychometric methods to refer to psychometric methods for multi-item scales that (1) examine the contribution of each item to the measurement properties of the overall scale and (2) recognize that items are categorical. The models include Rasch models (Rasch 1980; Fischer and Molenaar 1995), other IRT models (Samejima 1969; van der Linden and Hambleton 1997), and factor analytic models for categorical data (Muthén 1984). “Modern” psychometric methods have actually a rather long history within psychiatric research (both focusing on self-reported scales (Bech et al. 1978) and psychiatric outcome rating scales (Bech et al. 1984)). During the past 25 years, modern psychometric methods have increasingly been used in the analysis of patient-reported outcome measures (Teresi et al. 1989; Haley et al. 1994). For example, the NIH-sponsored Patient-Reported Outcomes Measurement Information System (PROMIS) project relies primarily on modern psychometric methods (Reeve et al. 2007). Similarly, modern psychometric analyses have started to be adopted for analysis of patient-reported HRQOL measures for patients with schizophrenia (D’haenen 1996; Pan et al. 2007; Boyer et al. 2010; Reise et al. 2011a; Laurens et al. 2012; Mojtabai et al. 2012; Chen et al. 2013; Michel et al. 2013; Park et al. 2015; Galindo-Garre et al. 2015; Norholm and Bech 2006). The present chapter provides an introduction to modern psychometric methods and discusses their potential use for analyses of HRQOL data from patients with schizophrenia. Rather than focusing on one particular approach, we will show what the methods have in common and how they can supplement each other.