AIM: To improve the psychometric properties of the St Georges Respiratory Questionnaire (SGRQ) to enhance its reliability and sensitivity.
METHOD: Examine the properties of each of the 50 questions using a combination of classical test theory and item-response theory (Rasch analysis) on a large population of COPD patients. Baseline data were taken from 5 studies giving a total of 3414 patients from 26 countries with a mean age of 65 years (range 39-89), 2455 were male, 1602 were smokers and the mean FEV1was 46% predicted (range 10-94%). Mean SGRQ Total score 49 (range 2-94). Sensitivity and internal consistency were measured in all 5 studies and questions that failed the psychometric criteria in at least 4 studies were defined as failures. Responsiveness (poor = unchanged response over 1 year in more than 90% of patients) was measured in 2 studies and items meeting the criteria in both studies were defined as unresponsive. Criterion validity was measured against SF36 scores in 1 study.
Results
[Table]
Thirteen questions did not meet the criteria for inclusion in the revised version of the SGRQ. Tests using Rasch analysis showed that item responses were unrelated to gender, age or country. Rasch analysis also showed that response categories in the Symptoms component should be collapsed to improve their performance.
Conclusions
The SGRQ has demonstrated sensitivity to change within and between populations. Removal of unreliable items with poor psychometric properties should improve its sensitivity through an improved signal to noise ratio.

Monday, May 24, 2004 8:15 AM