Initial Results of a Randomized Controlled Trial of Computerized Working Memory Training in Pediatric Sickle Cell Disease

Introduction: Disease-related neurocognitive deficits are common in sickle cell disease (SCD), even in the absence of stroke. Working memory (WM) is among the domains most commonly affected, likely a result of the high frequency of disease-related cerebral insults localized to the prefrontal cortex. Computerized cognitive training programs have garnered support for their efficacy in improving working memory across multiple pediatric populations, but have not been thoroughly explored in SCD. We present initial findings from a randomized controlled trial of the feasibility and efficacy of a home-based, computerized WM training intervention (Cogmed) with children with SCD.

Methods: Youth with SCD between the ages of 7 and 16 years old were recruited to complete a randomized (intervention or waitlist-control), dose-controlled, home-based computerized WM training program (Cogmed). After an initial neuropsychological assessment, participants with WM deficits were loaned an iPad on which they accessed Cogmed at home. Cogmedconsists of 12 increasingly challenging exercises that target skills involving visuo-spatial and verbal WM, practiced over 25 training sessions. At the end of the training period, participants completed a post-intervention neuropsychological assessment, including tests of auditory WM (Digit Span subtest, Wechsler Intelligence Scale for Children - Fifth Edition [WISC-V]) and visuo-spatial WM (Picture Span subtest, WISC-V; and Block Design-Forward and Block Design-Backward subtests, Wechsler Intelligence Scale for Children - Fourth Edition [WISC-IV]). Subtest scores are represented using scaled scores (M = 10; SD = 2). Scores on WISC-V WM subtests were used to calculate a composite Working Memory Index (WMI), represented by a standard score (M = 100; SD = 15).

Results: To date, 59 participants (M = 10.45, SD = 2.90; 53% female) have enrolled and completed a baseline neuropsychological assessment. Forty-nine percent (n = 29) exhibited WM deficits and were randomized to either begin Cogmed immediately (n = 14) or wait 5-8 weeks before beginning Cogmed (n = 15). Among those who have received the intervention and reached the end of their training (n = 24), 12 participants (50%) completed at least 10 sessions and 7 participants (29%) completed at least 20 sessions. The mean number of Cogmed sessions completed was 11.83 (SD = 7.97). As expected, participants in the waitlist condition did not exhibit improvements in WM, as demonstrated by stable performance on the WMI (t[8] = -1.520, p = .167), and on the Digit Span (t[9] = -0.449, p = .664), Picture Span (t[9] = -0.708, p = .497), Spatial Span-Forward (t[9] = 0.261, p = .802), and Spatial Span-Backward (t[7] = -0.261, p = .802) subtests. Participants who exhibited WM deficits at baseline and were given Cogmed exhibited reliable improvements in WMI (Baseline M = 89.79, SD = 12.36; Post-test M = 96.83, SD = 14.97), Digit Span (Baseline M = 7.62, SD = 2.24; Post-test M = 8.61, SD = 3.07), Picture Span (Baseline M = 8.93, SD = 2.87; Post-test M = 10.39, SD = 3.01), Block Design-Forward (Baseline M = 8.39, SD = 2.50; Post-test M = 9.06, SD = 3.35), and Block Design-Backward (Baseline M = 7.61, SD = 1.95; Post-test M = 9.39, SD = 2.62). Paired Samples t-tests conducted with participants who completed at least half of the Cogmed program (i.e., 10 sessions), revealed significant improvements on the WMI (t[7] = -3.361, p = .012), and on the Digit Span (t[7] = -3.175, p = .016), Spatial Span-Forward (t[7] = -2.448, p = .044), and Spatial Span-Backward (t[7] = -4.233, p = .004) subtests. Partial correlations controlling for respective baseline scores, indicated that the number of Cogmed sessions completed was positively correlated with post-test scores on the WMI (r = .568, p = .017), and on the Digit Span (r = .623, p = .008), Spatial Span-Forward (r = .518, p = .033), and Spatial Span-Backward (r = .612, p = .009) subtests.

Conclusions: Children with SCD exhibit significant neurocognitive deficits, particularly in WM. Initial results suggest that a home-based, computerized WM training program leads to significant improvements in WM. A dose-effect was observed, as participants who completed more Cogmed sessions evidenced the greatest improvements in WM. Home-based cognitive training programs may ameliorate SCD-related WM deficits but innovative approaches to motivating and supporting patients as they complete home-based interventions are needed to enhance adherence.