Table 3. Psychometric performance of the Cat-PROM5 items for the pre- and post-operative ‘Cycle 2’ and for all Cycles combined. (Items ordered from low to high visual difficulty from above down).
|
Cycle 2
Baseline and follow-up
(N=255) |
Cycle 2
Test–re-test
(N=53) |
Pilot and Cycles 1, 2 Combined Baseline and follow-up
(N=735‡ Pilot=200; Cy1=280; Cy2=255) |
||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Cat-PROM5 Items | Measurea (SE) | Infit MnSq | Outfit MnSq | Point measure correlation | Linear Weighted Kappa | Quadratic Weighted Kappa | Measure a (SE) | Infit MnSq | Outfit MnSq | Point measure Correlation |
| VSQ_Bad Eye | −0.92 (0.12) | 0.91 | 0.91 | 0.81 | 0.53 | 0.69 | −1.00 (0.07) | 1.13 | 1.14 | 0.78 |
| VSQ_Overall | −0.68 (0.10) | 1.10 | 1.06 | 0.85 | 0.63 | 0.73 | −0.66 (0.06) | 1.01 | 1.00 | 0.87 |
| VSQ_Reading | −0.02 (0.11) | 1.22 | 1.08 | 0.80 | 0.54 | 0.69 | 0.11 (0.06) | 1.10 | 1.08 | 0.81 |
| VCM1_Interfere | 0.04 (0.10) | 0.87 | 0.90 | 0.86 | 0.53 | 0.69 | 0.19 (0.06) | 0.84 | 0.88 | 0.86 |
| VSQ_Doing | 1.58 (0.14) | 0.88 | 0.84 | 0.74 | 0.57 | 0.66 | 1.36 (0.08) | 0.88 | 0.85 | 0.76 |
| Model Indices | Rasch-based reliability 0.88; Person separation: 2.66; Cronbach’s α: 0.88; Cohen’s d: −1.52b or −1.11c Residual eigenvalues 1.5; 2nd:1.3; 3rd:1.2; 4th:0.1; 5th:<0.1 | Rasch Person Measure ICC for 5-Item Scale 0.89 | Rasch-based reliability 0.90; Person separation: 2.98; Cronbach’s α: 0.89; Cohen’s d: −1.45a or −1.09c Residual eigenvalues: 1.5; 2nd:1.3; 3rd:1.2; 4th:1.0; 5th:<0.1 | |||||||
‡The number of patients in the validation sample is 735 which includes all from Pilot, and those sampled and available from Cycle 1 and 2. If a patient from either of these later cycles was randomly selected to contribute their post-operative measurement but this was missing, then they were dropped from the model generation group. All patients were represented in subsequent computations. The parameters of Rasch models were estimated by Joint Maximum Likelihood methods using WINSTEPS v3.72.3 software.
Measure represents the Item Location, also called the item difficulty (it is the average of the Rasch-Andrich model thresholds).
Computed with SD of baseline measures as the denominator.
Computed with SD of total sample as the denominator N for the Rasch performance parameters relates to the sample size used to generate the Rasch model.