VALIDATION DATA

There was excellent internal consistency with a Cronbach's alpha of 0.95, indicating that the domain questions measure the same construct.

Test and retest reliability. Spearman correlation coefficients between items of FIQ-R performed with a difference of one week. All coefficients reached statistical significance for a p < 0.01. Test-retest reliability was not done in this original study apart from the 10 patients in the focus group who had test-retest reliability of 0.8. A subsequent study of a Turkish version of the FIQR reported test-retest of 0.83

The FIQR has been evaluated using correlation coefficients, which measure the strength of the relationship between two variables. The results indicate that the FIQR has a high correlation coefficient with physical function (0.79) and days feeling good (0.68). It also has a strong correlation with VAS pain (0.75), VAS fatigue (0.66), VAS morning tiredness (0.61), VAS stiffness (0.60), VAS depression (0.67), and VAS anxiety (0.58). The total score of the FIQR has a very high correlation coefficient (0.85). Additionally, the FIQR has a strong correlation with work missed days (0.83) and job ability (0.76). Overall, these correlation coefficients suggest that the FIQR is a reliable and valid tool for evaluating the impact of fibromyalgia on physical function, symptom severity, and overall quality of life.

The questionnaire was developed for use in patients seen at OHSU, who were predominantly a female caucasian population living a typical american style. As the clinical spectrum of FM became more widely recognized, it became evident that the original seven symptom questions in the third domain needed to be expanded to include congnitive function, tenderness, balance and overall sensitivity to environmental stimuli such as bright lights and loud noises.

The goal of giving more weight to the first domain (function) compared to the other two domains appears to have achieved its desired effect. The imbalance between function and symptoms in the FIQ was reflected in weightings of 7% and 74% respectively. In the FIQR this ratio was 28% and 53% respectively; thus approximating the new weighting factors of 30% for function and 50% for symptoms that were adopted in the FIQR.

The FIQR has adequate construct validity as shown by its strong correlation with the FIQ and good correlations with the SF-36. Indeed, multiple regression and partial correlation analyses showed that each of the three FIQR domains predicted corresponding construct-related SF-36 subscale domains.

The four new symptoms (memory, balance, tenderness, and environmental sensitivity) provided good discriminant validity between the FM group and the other three groups (healthy controls, RA/SLE, and MDD). The FM group had substantially higher scores for each individual symptom compared to subjects in the three control populations. All four new symptoms had strong correlations with the total FIQR score. There was a good discriminant validity between FM patients and the three other groups. The mean total score for the FIQR in the FM subjects (56.6) was significantly higher than the other three groups (healthy controls = 12.1, RA/SLE = 28.6, and MDD = 17.3). Thus a high total FIQR score has some diagnostic implications.

There was a good correlation between the total scores for the FIQR and the FIQ (r = 0.88, P <0.001). Thus it should be possible to make reasonable comparisons between papers reporting the original FIQ and future papers reporting the FIQR.

The correlations between all the various FIQR and FIQ formats completed by the focus group was >0.85 (p<0.001). These results were reassuring in that the use of numeric scoring was similar to a VAS format and that the online version behaved similarly to the paper version.