Demographics
While sex and age were not different between VPT and FT children, FT children had a significantly higher socio-economic status. VPT children scored in the normal range for processing speed and working memory index on the WISC. In addition, their IQ evaluated with the WISC was also in the norm (for details, see Table 1).
Table 1. Demographic data. Mean ± SD are indicated in parentheses, except for the percentage of girls. Regarding the SES, note that data from 5 individuals were missing for the FT group and from 3 individuals for the VPT group. For the cognitive measures of the WISC, due to the absence of similar measures in the FT group, norms were used.
Measure
|
VPT
|
FT
|
p-value
|
N (% girls)
|
28 (53.57%)
|
23 (43.48)
|
.48
|
Age in yrs
|
8.80 (0.51)
|
9.00 (1.65)
|
.584
|
GA in weeks
|
27 (2.80)
|
NA
|
NA
|
SES
|
57.2 (24.8)
|
78.2 (18.7)
|
.004
|
WISC- Total (IQ)
|
105 (13.3)
|
100 (15)
|
NA
|
WISC – Processing speed
|
99.4 (15.8)
|
100 (15)
|
NA
|
WISC – Working Memory
|
102 (17.2)
|
100 (15)
|
NA
|
Auditory and visual acuity of the VPT children
During the neonatal period, auditory evoked potentials were classified as normal for all VPT infants. And by the age of 8, none of them had any hearing impairment. Regarding vision, two neonates suffered from grade I or II retinopathy of prematurity. One was treated with laser and the second one normalized without treatment. Both had no persistent sequalae. At 8 years old, 10 children were wearing corrective glasses for various pathologies, including 9 with complete correction and 1 without binocular vision. Moreover, one VPT child suffered from strabismus and was undergoing neuro-visual rehabilitation.
Simple Detection Task
Sensory Dominance
Sensory dominance, operationally defined here as the modality resulting in faster unisensory RTs in each participant (see also Romei et al., 2013 and Murray et al., 2018 for similar approaches), was similar between VPT and FT children (χ2 = 0.48, p=0.49). There was a significantly greater proportion of visual dominance compared to auditory dominance in both FT children (78% vs. 22%, p=.011) and VPT children (86% vs. 14%, p<.001).
Accuracy
The Friedman test on mean Accuracy showed a main effect of Stimulus Condition, (χ2(2)=7.32, p=.0.03), which was due to lower accuracy for auditory (M=91.6%, SD=8.32) than for multisensory (M=94.7%, SD=7.92) stimuli (i.e., Dubin-Conover pairwise comparison: p=.007). But visual (M=92.1%, SD=11.1) presented an intermediate accuracy which was not significatively different from auditory and multisensory conditions.
The Kruskall-Wallis test showed that the general accuracy of FT children (M=92.3%, SD=8.45) was similar to the accuracy of VPT children (M=93.2%, SD=6.69), (χ2(2)=0.01, p=0.94), as well as for each condition separately (see Table 2). The inclusion of SES as a covariate did not change this pattern of results.
Table 2. Accuracy and standard deviation for each modality on the simple detection task and its comparison between VPT and FT groups.
Stimulus Condition
|
VPT
|
FT
|
Chi-square
|
p
|
ε2
|
Visual
|
91.6% (12.6)
|
92.7% (9.38)
|
0.003
|
0.953
|
6.83e-5
|
Auditory
|
92.5% (8.11)
|
90.4% (8.61)
|
0.77
|
0.38
|
0.02
|
Multisensory
|
95.4%
(7.06)
|
93.8%
(8.94)
|
0.24
|
0.63
|
0.004
|
Reaction Times
The Friedman test on mean RTs showed a main effect of Stimulus Condition, (χ2(2)=74.9, p<.001), which was due to generally faster RTs for multisensory (M=415ms, SD=92.8) than for either visual (M=457ms, SD=103) or auditory (M=510ms, SD=138) stimuli (all Durbin-Conover pairwise comparisons between each stimulus condition p<.001).
The Kruskall-Wallis test showed a main effect of Group; FT children (M=366ms, SD=37.9) being generally faster than VPT children (M=538ms, SD=79.1), χ2(1)=35.3, p<.001, ε2=0.71 (Fig. 1A). This was the case for each stimulus condition (i.e., visual: MFT=370ms, SD=38.2 vs MVPT=529ms, SD=80.7, χ2(1)=33.8, p<.001, ε2=0.68; auditory: MFT=395ms, SD=47.8 vs MVPT=605ms, SD=113, χ2(1)=34.7, p<.001, ε2=0.69; and multisensory: MFT=335ms, SD=31.6 vs MVPT=481ms, SD=79.1, χ2(1)=35.3, p<.001, ε2=0.71). This pattern of results remained the same when adding the SES as a covariate.
Variability
The Friedman test on IQRs showed a main effect of Stimulus Condition, (χ2(2)=56.7, p<.001), which was due to smaller IQR for multisensory (M=76.8, SD=76.8) than for either visual (M=97.9, SD=58.4) or for auditory (M=140, SD=82.7) stimuli; the latter of which also significantly differed from each other (all Durbin-Conover pairwise comparisons between each stimulus condition p<.001).
The Kruskall-Wallis test showed a main effect of Group; FT children (M=60.7 SD=11.3) having generally smaller IQR than VPT children (M=141, SD=60.8), χ2(1)=33.6, p<.001, ε2=0.67 (Fig. 1B). This was the case for each stimulus condition (i.e., visual: MFT=55.8, SD=12.3 vs MVPT=133, SD=58.7, χ2(1)=25.5, p<.001, ε2=0.51; auditory: MFT=83.4, SD=30.6 vs MVPT=186, SD=83.5, χ2(1)=21.4, p<.001, ε2=0.43; and multisensory: MFT=43.0, SD=9.78 vs MVPT=105, SD=95.2, χ2(1)=21.8, p<.001, ε2=0.44).
Multisensory Gain
Reaction Times. The percentage multisensory gain on RT was similar in FT children (M=9.08%, SD=2.29) and in VPT children (M=7.52%, SD=7.87), (t(32.4)=1, p= .33; Cohen’s d=0.27).
Variability. The percentage multisensory gain on IQR was similar in FT (M=16.6%, SD=9.61) and in VPT children (M=9.61%, SD=48.7), (t(38.6)=0.69, p= .50; Cohen’s d=0.19).
Race Model Inequality
There was slight violation of the Race Model inequality in VPT children for the 0, 10, 15, 25, 30, and 35 quantiles (Fig. 2). However, in no case was the cumulative probability for the multisensory condition significantly higher than the cumulative probability for the values predicted by the race model (for details, see Table 3). By contrast, the race model was significantly violated in FT children for the 0, 5, 10, 15, 20, 25, 30, 35, and 40 quantiles. The cumulative probability for the multisensory condition was significantly higher than the cumulative probability for the values predicted by the race model for these eight quantiles (for details, see Table 3).
Table 3. The average cumulative probability (and standard deviation) for the simultaneous condition with that predicted by the race model as well as the corresponding paired t-test and p-values for all quantiles in VPT children (left) and FT children (right).
|
VPT
|
FT
|
|
Cumulative Probability
|
|
|
Cumulative Probability
|
|
|
Quantile (%)
|
Simultaneous
|
Predicted
|
t
|
p value
|
Simultaneous
|
Predicted
|
t
|
p value
|
0
|
0.03(0.03)
|
0.03(0.03)
|
0.00
|
0.5
|
0.01(0.01)
|
0.002(0.01)
|
2.87
|
0.004**
|
5
|
0.06(0.09)
|
0.07(0.06)
|
-0.41
|
0.66
|
0.03(0.03)
|
0.003(0.01)
|
4.00
|
<.001***
|
10
|
0.15(0.20)
|
0.14(0.13)
|
0.77
|
0.22
|
0.09(0.08)
|
0.03(0.06)
|
4.35
|
<.001***
|
15
|
0.22(0.24)
|
0.21(0.20)
|
0.33
|
0.37
|
0.22(0.16)
|
0.09(0.13)
|
5.69
|
<.001***
|
20
|
0.31(0.27)
|
0.31(0.27)
|
-0.01
|
0.50
|
0.38(0.22)
|
0.19(0.19)
|
7.49
|
<.001***
|
25
|
0.44(0.30)
|
0.43(0.29)
|
0.79
|
0.22
|
0.53(0.23)
|
0.33(0.23)
|
6.88
|
<.001***
|
30
|
0.56(0.28)
|
0.56(0.30)
|
0.14
|
0.45
|
0.65(0.20)
|
0.50(0.24)
|
7.11
|
<.001***
|
35
|
0.57(0.28)
|
0.64(0.30)
|
1.31
|
0.10
|
0.73(0.19)
|
0.65(0.22)
|
3.93
|
<.001***
|
40
|
0.74(0.19)
|
0.75(0.18)
|
-0.43
|
0.66
|
0.79(0.17)
|
0.76(0.19)
|
1.20
|
0.12
|
45
|
0.81(0.15)
|
0.84(0.12)
|
-1.53
|
0.93
|
0.82(0.17)
|
0.83(0.18)
|
-0.52
|
0.69
|
50
|
0.86(0.12)
|
0.89(0.08)
|
-1.85
|
0.96
|
0.84(0.17)
|
0.87(0.15)
|
-1.39
|
0.91
|
55
|
0.88(0.10)
|
0.92(0.07)
|
-2.11
|
0.98
|
0.85(0.17)
|
0.9(.015)
|
-2.43
|
0.99
|
60
|
0.90(0.10)
|
0.94(0.05)
|
-2.41
|
0.99
|
0.87(0.17)
|
0.92(0.14)
|
-2.72
|
0.99
|
65
|
0.91(0.10)
|
0.95(0.04)
|
-2.87
|
1
|
0.88(0.17)
|
0.93(0.13)
|
-2.86
|
1
|
70
|
0.92(0.09)
|
0.96(0.04)
|
-3.16
|
1
|
0.88(0.17)
|
0.93(0.13)
|
-3.22
|
1
|
75
|
0.93(0.08)
|
0.98(0.03)
|
-3.08
|
1
|
0.89(0.17)
|
0.94(0.12)
|
-3.30
|
1
|
80
|
0.93(0.08)
|
0.98(0.03)
|
-3.15
|
1
|
0.89(0.17)
|
0.94(0.11)
|
-3.34
|
1
|
85
|
0.94(0.08)
|
0.98(0.03)
|
-3.00
|
1
|
0.89(0.17)
|
0.95(0.11)
|
-3.15
|
1
|
90
|
0.94(0.07)
|
0.98(0.02)
|
-3.29
|
1
|
0.89(0.17)
|
0.95(0.11)
|
-3.15
|
1
|
95
|
0.95(0.07)
|
0.99(0.02)
|
-3.09
|
1
|
0.89(0.17)
|
0.95(0.11)
|
-3.16
|
1
|
100
|
0.96(0.07)
|
0.99(0.02)
|
-3.27
|
1
|
0.90(0.17)
|
0.96(0.10)
|
-3.03
|
1
|
Relationships between multisensory gains and neuropsychological scores in VPT
VPT children scored in the normal range on all selected scales of the SPM questionnaire and WISC-V tests (for details, see Table 4). By controlling for gestational age and SES, partial Spearman correlations were carried out. The multisensory gain was not significantly correlated with any of the clinical or cognitive measures (for details, see Table 5).
Table 4. Neuropsychological tests for very preterm children. Note. SPM = Sensory Processing Measure, WISC = Wechsler Intelligence Scale Children, Mean = T-scores
|
Mean (SD)
|
Norms
|
IQR
|
SPM – Vision
|
48.9 (6.80)
|
<60
|
13
|
SPM – Audition
|
48.8 (8.85)
|
<60
|
13
|
SPM – Total
|
51.3 (7.11)
|
<60
|
6
|
WISC – Processing Speed
|
99.4 (15.8)
|
100 (15)
|
16
|
WISC – Working Memory
|
102 (17.2)
|
100 (15)
|
18
|
WISC - Total
|
105 (13.3)
|
100 (15)
|
15
|
Table 5. Partial correlations between the multisensory gain and clinical and cognitive measures, controlling for SES and gestational age.
Variables
|
Multisensory Gain
|
p value
|
SPM – Vision
|
-0.24
|
0.32
|
SPM – Audition
|
-0.05
|
0.86
|
SPM – Total
|
0.11
|
0.66
|
WISC – Working Memory
|
0.02
|
0.92
|
WISC – Processing Speed
|
-0.02
|
0.94
|
WISC – IQ Total
|
-0.15
|
0.51
|
Digit Span
|
0.12
|
0.59
|
Digit Span - Ascending
|
0.06
|
0.78
|