Objective: 'Clinimetric' and 'psychometric' approaches are currently used to develop health related quality of life questionnaires. The Quality of Life after Myocardial Infarction questionnaire (QLMI) was originally developed using 'clinimetric' criteria; it was subsequently modified (McNew QLMI) and a new domain structure was defined using factor analysis. The objective of this study was to compare the measurement properties of the McNew QLMI scores when both approaches for scoring are used.
Methods: The McNew QLMI and SF-36 were administered to patients 2 weeks and 2 months after myocardial infarction. Two sets of scores for the McNew QLMI were computed using the original 'clinimetric' and the subsequent 'psychometrically' derived scoring systems. Reliability statistics for the two sets of domains were compared and construct validity was assessed by establishing a priori hypotheses on the expected correlation between each score and the dimensions of the SF-36.
Results: Both sets of scores had similar reliability (Cronbach's alpha between 0.64 and 0.93) and responsiveness (SRMs between 0.17 and 0.87) while validity was better for the 'clinimetric' set of scores (concordance between observed and expected correlations was moderate for the 'clinimetric' scores and fair for the 'psychometric' scores).
Conclusion: Since overall measurement properties of the 'clinimetrically' scored McNew QLMI are better than the 'psychometrically' scored version, we suggest that either the original 'clinimetric' system is used or that an improved 'psychometric' version is developed.