Does Early Age at Brain Insult Predict Worse Outcome? Neuropsychological Implications

Vicki Anderson, PhD; Rani Jacobs, PhD; Megan Spencer-Smith, PhD; Lee Coleman, MD; Peter Anderson, PhD; Jackie Williams, PhD; Mardee Greenham, BA (Hons); Rick Leventer, PhD

Disclosures

J Pediatr Psychol. 2010;35(7):716-727. 

In This Article

Results

Sample Demographics

No group differences were identified for gender, SES, or handedness. A significant age at test difference was identified, F(5, 158) = 3.21, p =.009, η 2 =.09, revealing that the LC group was older than the CON (p =.04), INF (p =.007), PRE (p =.02), and MC (p =.037) groups. To account for the possible effect of age, MANCOVA was conducted for measures where raw scores were employed. Group differences were also present for age at diagnosis, F(5, 149) = 64.25, p <.001, η2 =.68, time since diagnosis, F(5, 149) = 39.95, p <.001, η2 =.57 and presence of seizures, χ2(5, N = 158) = 17.44, p =.004, V =.332, with a large proportion of children in the CON group with epilepsy/seizures (SR = 1.7) and a small proportion in the PRE group.

Analysis of mechanisms of insult detected significant group differences, χ2 (20, N = 163) = 150.10, p <.001, V =.48 (Table II). There were no group differences for region [frontal, χ2 (5, N = 164) = 7.84, p =.17, V =.22; extra-frontal, χ2 (5, N = 164) = 9.74, p =.08, V =.24; subcortical, χ2 (5, N = 164) = 5.08, p =.41, V =.18], or extent of insult (unifocal/multifocal), χ2 (5, N = 164) = 5.46, p =.36, V =.18.

Comparing EBI to Normative Expectations

As illustrated in Table III, using total group data, children with EBI achieved poorer scores than the normal population (p <.001) on all measures, thereby supporting our first hypothesis. All p-values were below the strictest adjusted alpha level calculated using Holm's sequentially rejective Bonferroni procedure, where α =.002. For the majority of variables (17/22), and across all domains, effect sizes (ES) were large (>.75). For 6/22 measures, ES were very large (>1.0). These deviations from normal were observed across a range of domains: language (VOC), attention (SS:TPT, SSDT:DEC), executive function (CWI:I/S, TMT:COM), and visuo-spatial skills (TMT:VS).

Comparisons Across AL Groups

Analysis across all domains included either presence of seizures and/or age at testing as covariates. Tests of the homogeneity of slopes for each domain revealed no violations of this assumption (all p >.05).

Language Seizures covaried significantly with the language domain, Wilks's =.93, F(4, 129) = 2.61, p =.039, as did age at testing, Wilks's =.89, F(4, 129) = 3.93, p =.005. After partialling out the variance associated with these variables, the multivariate contrast for AL group remained significant, Wilks's =.84, F(4, 129) = 6.17, p <.001, η2 =.16. Univariate contrasts identified expected group differences, demonstrating that the CON, PERI and INF groups combined scored significantly more poorly than the PRE, MC, and LC groups combined on all measures. Significant differences remained after corrective adjustments were made to alpha levels (Table IV), providing support for hypothesis 2a.

Visuo-spatial Skills Seizures covaried significantly with the visuo-spatial domain, Wilks's =.89, F(4, 131) = 3.89, p =.005, as did age at testing, Wilks's =.83, F(4, 131) = 6.64, p <.001. After partialling out these effects, the multivariate contrast for AL group remained, Wilks's =.87, F(4, 131) = 4.73, p =.001, η2 =.13. Univariate contrasts identified expected group differences, demonstrating that the CON, PERI, and INF groups combined recorded significantly lower results than the PRE, MC, and LC groups combined on all measures. Significant differences remained after corrective adjustments were made to alpha levels (see Table 4), also supporting hypothesis 2a.

Attention Again, seizures covaried significantly with the attention domain, Wilks's =.88, F(5, 142) = 3.94, p =.002, as did age at testing, Wilks's =.92, F(5, 142) = 2.40, p =.040. The first contrast, comparing the CON, PERI, and INF groups combined to the PRE group identified no significant multivariate effect (p >.05) and only one significant univariate difference: CC:TOT (Table IV). The second contrast, comparing the CON, PERI and INF groups combined to the MC and LC groups combined identified a significant multivariate effect of AL group, Wilks's =.84, F(5, 142) = 5.30, p <.001. Univariate contrasts identified expected group differences, with the CON, PERI, and INF groups combined recording significantly lower scores than the MC and LC groups combined on 3/5 measures: CC;TOT, SS:TPT and LNS. The third contrast, comparing the PRE group to the MC and LC groups combined identified no significant multivariate effect or univariate differences (all p >.05). These results provided partial support to hypothesis 2b.

Executive Function Seizures covaried significantly with the executive domain, Wilks's =.90, F(4, 140) = 3.73, p =.006, but age at testing did not, p >.05. The first contrast, comparing the CON, PERI, and INF groups combined to the PRE group identified no significant multivariate effect (p >.05) and only one significant univariate difference: TMT:COM (Table IV). The second contrast, comparing the CON, PERI, and INF groups combined to the MC and LC groups combined identified a significant multivariate effect of AL group, Wilks's =.87, F(4, 140) = 5.23, p =.001. Univariate contrasts identified expected group differences, with the CON, PERI and INF groups combined recording significantly lower scores than the MC and LC groups combined on all four measures. The third contrast, comparing the PRE group to the MC and LC groups combined identified no significant multivariate effect or univariate differences (all p >.05). This again provided only partial support to hypothesis 2b.

Memory Skills No group differences were identified for memory measures. Seizures, Wilks's =.882, F(7, 124) = 2.37, p =.026, and age at testing, Wilks's =.891, F(7, 124) = 3.40, p =.002, co-varied significantly with the memory domain, and no significant multivariate effect for group remained after accounting for these two variables, Wilks's =.743, F(35, 524.1) = 1.10, p =.33, η2 =.06. Univariate analyses showed no group differences for any of the memory measures (Table IV), which does not support hypothesis 2c.

Processing Speed Seizures did not covary significantly with information processing, Wilks's =.950, F(3, 131) = 2.31, p =.079, but age at testing did, Wilks's =.917, F(3, 136) = 4.11, p =.008. After partialling out the variance associated with age at testing, a significant multivariate effect remained, Wilks's =.798, F(15, 375.84) = 2.13, p =.008, η2 =.07. After adjusting alpha levels, univariate analyses showed a significant group difference on CWI:NRT and SSM remained. Post hoc analyses revealed significant findings only for CWI:NRT, with the MC group recording significantly higher scores than both the CON (p =.04) and PERI (p =.01) groups. This provides partial support to hypothesis 2c.

Comments

3090D553-9492-4563-8681-AD288FA52ACE
Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.
Post as:

processing....