Assess the test-retest reliability for each of the domains of CSF-36 based on the study described in Example10.5.

Transcribed Image Text:

Example 10.5 Consider the random subsample of 197 patients in the CSF-36 study who were asked to take the questionnaire again 1-2 days after admission. As the time is too short for patients to show a significant improvement in their quality of life, we may use the data at the two time points to assess the test-retest reliability of the translated instrument. Shown in Table 10.3 are the estimates, standard errors, and 95% confidence intervals of Pearson correlation, CCC, and ICC between the two assessment times for the PF domain based on the subset of 197 subjects. The three indices yield quite similar values, and the high values for all the three coefficients indicate good test-retest reliability for this particular domain of the CSF-36. TABLE 10.3: Pearson correlation, CCC, and ICC between admission and 1-2 days post-admission to hospital for the PF domain Index Estimate (SE) 95% CI PPearson PCCC PICC 0.76 (0.022) (0.72, 0.80) 0.75 (0.031) 0.76 (0.051) (0.69, 0.81) (0.66, 0.86) The near identical estimates across the three indices provide a strong indication that there is very little "drift" in the domain outcome from admission to post-admission to the hospital, as expected for such a short period, since otherwise the estimated pccc would have been substantially smaller than its counterpart PPearson as noted in Example 10.3.