Survivors of childhood brain tumor are historically thought to perform worse on measures of executive functioning, including cognitive flexibility (CF; e.g., set-shifting), when compared to their peers. Commonly utilized measures, such as subtests from the Delis-Kaplan Executive Function System (D-KEFS), have baseline conditions that attempt to measure performances independent of but critical for CF tasks (e.g., motor speed on trail making, letter fluency on verbal fluency). However, in research, conditions measuring CF are often included in analyses without accounting for these important baseline conditions. The aim of the current study is to explore differences in CF performance between survivors and their healthy peers when controlling for baseline conditions. The variance explained by each baseline condition on CF condition performance in survivors is also explored.

A sample of 107 long-term survivors of childhood brain tumor (Mage=21.81, SD=5.99, 50.5% female) and 142 healthy controls (Mage= 23.25, SD=6.61, 61.3% female) were administered the Trail Making Test (TMT), Color-Word Interference (CWI), and Verbal Fluency (VF) subtests from the D-KEFS. For the TMT, baseline conditions include visually scanning for a target, motor speed, and letter and number sequencing. For the CWI subtest, baseline conditions include rapid color naming, word reading, and reading words in a different colored ink. On the VF subtest, baseline conditions include rapidly naming words with a specific letter and from a specific category. An analysis of covariance was conducted for each subtest to determine if groups differed in performance on the CF condition (i.e., Number-Letter Switching, Inhibition/Switching, Category Switching Accuracy) when controlling for baseline conditions. In survivors only, linear regressions investigated the amount of variance explained by each baseline condition on the CF conditions of each subtest.

Groups did not differ in CF performance of each subtest when controlling for baseline conditions (ps>.10). Across subtests, baseline conditions significantly predicted CF performance in survivors. On the TMT, Letter Sequencing (p=.003, unique-R2=.05), but not Visual Scanning, Number Sequencing, or Motor Speed, was a significant predictor of Number-Letter Sequencing performance (p<.001, R2=.50). On the CWI subtest, Word Reading (p<.001, unique-R2=.09) and Inhibition (p<.001, unique-R2=.05), but not Color Naming, were significant predictors of Inhibition/Switching performance (p<.001, R2=.67). On the VF subtest, Letter Fluency (p=.009, unique-R2=.06) and Category Fluency (p<.001, unique-R2=.08) were significant predictors of Category Switching Accuracy performance (p<.001, R2=.37).

Findings suggest that CF may not differ between survivors and their healthy peers, but that other factors of executive functioning, such as processing speed, drive performance differences on measures of CF. As these tasks rely heavily on speed, survivors may be slower than their healthy counterparts, but may not perform worse on set-shifting. In addition, these results highlight the importance of controlling for lower-order processes in analyses to help isolate CF performance and more accurately characterize potential differences between groups. While replication of findings in survivors and other clinical groups (e.g., congenital heart disease, traumatic brain injury) is still needed, this work can help inform which processes are most important to account for, which is not yet established.