Purpose: Quality of life in multiple sclerosis has been often measured through the SF-36 questionnaire. In this study, validation of the SF-36 summary scores, its 'physical' component, and its 'mental' component was attempted by exploring the joint predictive power of disability (EDSS score), of anxiety and depression (HADS-A and -D scores, respectively), and of disease duration, progression type, age, gender and marital status.
Method: The sample consisted of 75 patients suffering from multiple sclerosis admitted to an inpatient rehabilitation unit. The interplay between potential predictors was assessed through a particular regression model (classification and regression tree, CART). Two main advantages of this technique are its robustness with respect to distributional assumptions (rarely met by scores coming in from questionnaires) and its sensitivity to high-order interactions, between independent variables, difficult to detect through conventional multiple regression.
Results: Predictive variables for physical component of the SF-36 were EDSS and HADS-D (36.8% variance explanation). The only predictive variable for mental component of SF-36 was HADS-D (39.1% variance explanation).
Conclusion: Results confirm previous findings showing that in patients with multiple sclerosis quality of life is heavily determined by person's mood, whatever his/her neurological or functional severity. The usefulness and validity of the SF-36 as an index representative of quality of life is debatable, as long as depression explains much of its variance. Further refinement of quality of life definition and measurement is worth further psychometric and statistical research.