Methods and statistical applications in the construction of health related quality of life scales published in Thai medical journals and theses

Abstract

This study aimed to examine methodologies and statistical methods in constructing health related
quality of life scales as reported in Thai health journals and theses from January 2000 to December 2009.
Research articles were searched from E-Databases and by hand searching. Theses were search from the
TDC database. Searching keywords were “health related quality of life”, “illness related quality of life”, “disease
related quality of life” and “disorders related quality of life.”
Three research articles and 10 theses were identified. Standard methods for constructing scales
followed suggestions of textbooks, previous studies and recommendations from experts. All 13 studies
elaborated on 4 issues: 1) methods used to define the construct or definition of a variables wished to measure,
2) methods in generating items, 3) designing the scales and 4) testing the reliability of the scales with the
Cronbach’s alpha coefficient. Twelve out of the 13 studies pre-tested the scales. Eleven studies mentioned
the methods used to assess the content validity. Seven and three studies assessed the content validity by
using expert’s opinions and by calculating the content validity index (CVI) respectively. Nine studies mentioned
the methods used for the item analysis; 8 studies used an exploratory factor analysis. Six studies described
the details of how to use the scales. Three studies described the methods for assessing the construct validity
of the scales; a simple correlation coefficient was calculated and a confirmatory factor analysis was done. Two
studies tested the criterion-related validity by calculating a simple correlation coefficient. None of the studies
reported an evaluation of the sensitivity to change property of the scale.
Although all studies reported steps in constructing a scale covered as the standard methods
suggested, but the process of each steps need to be revised. Scale developer should use several methods in
defining the construct of the variable needed to measure. All studies must assess the construct validity of the
scale.