iSAT於思覺失調症患者之驗證結果 (1020 renew)

Validation of an iPad-based Selective Attention Test in patients with schizophrenia

N=51 (M=25, F=26) mean age=47.2yrs; mean education years=11.2yrs; mean age of onset=24.8yrs; range of CGIS=1-4

1. Test-retest reliability, random measurement error and practice effect: 
Low load indices: good
High load indices: need to be improved due to substantial random measurement error and moderate practice effect

Indices	ICC   (95% CI)	SEM (SEM%)	MDC (MDC%)	Effect size (s: significant; ns: not significant)
Low load (L)	0.89  (0.80-0.94)	0.11  (9.5%)	0.29  (26.3%)	0.14 (s)
Low load with Distractor (LD)	0.88   (0.78-0.93)	0.12  (10.8%)	0.35  (29.8%)	0.15 (s)
High load (H)	0.75   (0.33-0.89)	0.40  (14.6%)	1.11  (40.5%)	0.43 (s)
High load with Distractor (HD)	0.74   (0.41-0.87)	0.47  (16.8%)	1.31  (46.4%)	0.39 (s)

(1) The ICC for the L and LD indices ranged from 0.88 to 0.89, indicating that both low load indices had high test-retest reliability.

(2) The ICC for the H and HD indices ranged from 0.74 to 0.75, indicating that both high load indices had acceptable test-retest reliability.

(3) Our results showed that both low load indices had acceptable random measurement error, whereas both high load indices had substantial random measurement error.
*The criterion for the MDC% was set at 30%. MDC%<30% indicates acceptable random measurement error.

(4) A significant practice effect was shown between repeated assessments for all four indices. For both low load indices, the practice effect was negligible; whereas for both high load indices, the practice effect was around moderate.

2. Convergent validity 
(1) Correlation with MoCA: acceptable

L	LD	H	HD
0.42	0.43	0.48	0.47

All four indices had fair correlations with the MoCA, indicating acceptable convergent validity. (2) Correlation with CGIS: insufficient Only the L and LD indices had weak correlations (r=0.28-0.29) with the CGIS, indicating insufficient convergent validity. 3. Discriminative validity: Yes (1) MoCA≧26 [good global cognition, N=14] vs. <26 [poor global cognition, N=37]: All 4 indices had discriminative validity to discriminate participants with/without poor global cognitive function. (2) Young adult (n=16) vs. middle-aged adult and elderly (n=35): All four indices had discriminative validity to discriminate young adult and middle-aged/elderly adult participants. 4. Ecological validity: partial, correlation with the Personal and Social Performance (PSP) Only the H and HD indices had significant weak correlation (r=0.29-0.31) with the PSP, indicating limited ecological validity. 5. Construct validity: fitting with perceptual load theory The results of the iSAT in patients with schizophrenia confirmed the hypothesis of perceptual load theory.  Participants' response time became slower (>0.04 seconds) when the distractor showed up in the low load condition. However, the distractor did not affect the participants' response time in the high load condition. *How to reduce moderate practice effect in the high load condition: To set up a criterion for response time to reach a plateau phase of selective attention performance.