VALIDATION DATA

MEASUREMENT PROPERTIES
Limitations
Observations
1. RELIABILITY
A. Internal Consistency Tested
Cronbach's (Describe)

Moderate internal consistency: Cronbach's alpha of 0.71–0.85

B. Reliability intraobserver or test-retest Tested
Continuous scores: intraclass
correlation coefficient (ICC)
Dichotomus: Cohen kappa (Describe)

Good test-retest reliability: estimated kappa of 0.78 to detect low compliance.

C. Reliability interobserver or Measurement error Not Tested
Standard error of measurement (SEM),
smallest detectable change (SDC) or
Limits of agreement (LoA) (Describe)
2. VALIDITY
A. Content validity: face validity Tested
Expert opinion (relevance and
comprehensiveness) (Describe)

In the development of the CQR, thirty-two patients participated in semistandardized home interviews about their attitude toward their antirheumatic medication, actual drug intake, and reasons for not taking medication and a focus group interview with 7 patients (3 RA, 2 PMR, 2 gout) was held.

B. Construct Validity:
Structural validity
Not Tested
Hypotheses-testing Not Tested
Cross-cultural validity Not Tested
Brief Description
C. Criterion validity Not Tested
Comparison with a 'gold standard' Continuous scores:
correlations, ROC curves Dichotomus:
Sensitivity & Specificity (Describe)

A discriminant analysis was performed to test the predictive value of the CQR. The classification table of the calculated discriminant function of the CQR versus MEMS data to detect taking compliance ≤ 80% shows a sensitivity and specificity of 62% (95% confidence interval 56.8, 67.3%) and 95% (95% CI 92.2, 97.1%), respectively. The likelihood ratio for a positive test result (e.g., to detect taking compliance ≤ 80%) was 11.6 (95% CI 6.7, 20.1). Sensitivity and specificity to detect unsatisfactory correct dosing (≤ 80%) was 89% (95% CI 85.0, 91.9%) and 70% (95% CI 64.7, 74.7%), respectively. The corresponding likelihood ratio to detect correct dosing ≤ 80% is therefore 2.9 (95% CI 1.5, 5.9).