Developmental trajectories of forward and backward digit spans in deaf children with cochlear implants

Abstract

Background: Accounting for outcome variability among pediatric cochlear implant (CI) recipients is an enduring clinical challenge. Short-term memory and working memory (STM/WM), as indexed by digit span forward (DSF) and digit span backward (DSB), have been shown to be strongly correlated with speech and language (S/L) outcomes. The enormous variability observed in conventional outcome measures of S/L may reflect individual differences in STM/WM.

Methods: Repeated measure auditory digit spans were obtained from 110 children (age 3-15 years; mean 7.2 years) with at least 2 years of CI use. Growth curves were computed for each child, and linear functions were fit to both DSF and DSB. Slopes and intercepts were used as parameters in mixed-models to assess relations between STM/WM capacity change over time and S/L outcome measures including vocabulary (PPVT), open-set set spoken word recognition (PBK), and sentence perception (HINT-C).

Results: For DSF, the percent of the sample more than 1 SD below the norm at each age ranged from 54 to 90% (mean = 66.5%). For DSB, the percent of the sample more than 1 SD below the norm at each age ranged from 23 to 42% (mean = 34.5%) at ages where there were at least five children. Four subgroups within our CI sample emerged: (Subgroup 1) children demonstrating age-appropriate growth in both DSF and DSB scores over time (49/110, 44.55%); (Subgroup 2) children demonstrating age appropriate growth in DSF over time but below average growth in DSB over time (23/110, 20.91%); (Subgroup 3) children demonstrating below average growth in DSF over time but age-appropriate growth in DSB over time (19/110, 17.27%); and (Subgroup 4) children demonstrating below average growth in both DSF and DSB over time (19/110, 17.27%). For all tests except CELF-3, Subgroup 4 demonstrated the poorest performance among the four DS slope subgroups. Significant differences were observed between Subgroup 1 and Subgroup 4 on last visit PBK-Word (P = 0.029), PPVT (P = 0.018), and HINT-C in Quiet (P = 0.001), but not CELF-3 (P = 0.433).

Conclusion: The findings from this longitudinal study suggest that differences in the rate of development of STM/WM may influence S/L outcomes in children with CIs. The clinical implications of these findings are significant because they indicate that the rate of development of STM/WM, and not just the actual level of STM/WM at a single time point, predicts later S/L development in this clinical population. Targeted interventions to improve developmental rate of verbal STM/WM may hold promise for enhancing S/L skills in children with CIs.

Figure 1

Red lines are from linear regression of each child’s digit span scores over time; black lines represent the average growth curve of the sample. Panel A shows digit recall forward across chronological age. Panel B shows digit recall backward across chronological age. Means and standard deviations of norm data from a mixed effects model specifying random slopes and intercepts (from CMS) are shown at each age by black dots and bars.

Figure 2

Scatterplot showing digit span slopes over time for all subjects. Black cross-hatch lines represent the normative slopes over time for normal hearing peers for digits forward (vertical line) and digits backward (horizontal line) obtained from the CMS. The right upper quadrant shows Subgroup 1 (++), children who demonstrated age-appropriate slopes for both DSF and DSB. The right lower quadrant shows Subgroup 2 (+−), children who demonstrated age-appropriate DSF slope, but delayed DSB slope. The left upper quadrant shows Subgroup 3 (−+), children who demonstrated delayed DSF slope but age-appropriate DSB. The lower left quadrant shows Subgroup 4 (−−), children who demonstrated both delayed DSF slope and DSB slope.

Figure 3

Mean scores with 95% confidence interval bars for the final test session’s performance on the four S/L outcome measures (A: PBK-Word; B: CELF; C: PPVT; D: HINT-C in Quiet) across the four digit span slope subgroups. rs = raw score; ss = standard score.