Long term effects of early childhood traumatic brain injury on narrative discourse gist and psychosocial functioning

Abstract

Purpose:

To examine the long-term consequences of early childhood traumatic brain injury (TBI), approximately 7 years post-injury, on cognitive communication competency within a complex interplay of individual, contextual, and psychosocial factors. Additionally, we were interested in looking at the association of communication competence with social competence and overall functioning.

Methods:

64 children with orthopedic injury, 40 children with moderate traumatic brain injury, and 14 children with severe traumatic brain injury who were between 3 years and 6 years 11 months at injury completed a narrative discourse task and clinical measures in a single visit at a longer term follow up in early adolescence, an average of 6.8 years post injury. Analyses of covariance were conducted to compare groups on the discourse task, and hierarchical linear regressions were conducted to examine the association of discourse measures with clinical measures of cognitive and psychosocial functioning.

Results:

Children with severe traumatic brain injury performed worse than children with moderate traumatic brain injury and orthopedic injury on all discourse indices and clinical measures. Injury severity, pragmatic skills, and socioeconomic status were associated with discourse gist. Discourse gist was the most sensitive measure of communication competence, and it was significantly associated with psychosocial outcomes independent of group.

Conclusion:

Children who sustain a severe traumatic brain injury in early childhood are at risk for long-term cognitive communication impairments that may be related to a complex interplay of injury, individual, and social factors.

Background

Traumatic brain injury (TBI) is a leading cause of acquired disability in childhood in the United States.¹ Following a TBI, children often experience cognitive communication disorders that place them at an increased risk for lower academic performance and greater difficulty adapting to the demands of everyday living.^2–4 Cognitive communication disorders are communication impairments resulting from underlying cognitive deficits due to neurological impairment.⁵ Although many children regain lower-level language abilities within the first few months after TBI, cognitive-communication impairments go beyond the level of single words or sentences and involve larger units of language such as discourse. Discourse is language in the contextual, narrative, and conversational setting in which is it used and understood, and discourse proficiency involves a complex interaction of cognitive, linguistic, emotional, and contextual processes.⁶Throughout the day, we are constantly processing, encoding, and storing information through different types of discourse (i.e., conversations, reading, writing, etc.). The ability to abstract meaning from the large amounts of information we process is called gist reasoning, and it is at the core of everyday communication at school, home, work, and in social settings.⁷ Impaired discourse, particularly gist reasoning, is the hallmark communication disorder following TBI.^7–8 In view of the essential role gist reasoning plays in everyday communication, impairments can have long-term deleterious effects on academic achievement and psychosocial functioning.^9–13

Communication impairments after TBI are frequent, with reported incidence rates exceeding 75%.^14–16 Following a TBI, children may experience a range of cognitive, linguistic, and behavioral impairments that contribute to some form of communication disorder.³ In MacDonald’s (2017) model of cognitive communication competence, discourse communication is viewed within a complex interplay of individual and contextual factors.¹⁷ After a TBI, discourse may be disorganized or not clearly summarized with central semantic meaning (i.e., gist) and these deficits may arise from underlying problems with language, attention, speed of processing, memory, executive functions, or emotional processing.^4,13,18–19 A number of individual factors have also been linked to discourse impairments following pediatric TBI, including age at injury, severity, and time since injury.^{9–13, 19–20, 22} Additionally, discourse can vary depending on the communication demands of the child’s environment, communication partner (e.g., parents, siblings, peers, etc.), or situational context.^3–4,17 MacDonald’s model illustrates the interactions among cognition, communication, emotion, individual, and contextual factors and highlights the complexity of discourse competency and gist reasoning across settings.¹⁷ Furthermore, because discourse competency and gist reasoning are essential for everyday communication, impairments in both have been associated with poor psychosocial functioning and quality of life.^20–21 Unfortunately, deficits in discourse and gist reasoning are commonly under identified due to reliance on assessment tools that lack the sensitivity to detect subtle but functionally significant communication impairments.

Growing evidence indicates that narrative discourse measures are more sensitive than standardized language tests for assessing cognitive communication impairments after pediatric TBI.^7,22 Unlike traditional language measures, narrative discourse tasks draw on both linguistic and cognitive skills to provide a higher-level measure of verbal expression. Discourse in general, and gist reasoning specifically, has been shown to be a sensitive measure of cognitive communication competency after TBI.^22–25 Because gist reasoning is a higher-order cognitive function that uses both bottom-up and top-down cognitive processes to abstract meaning from the language input, it is hypothesized to be the most sensitive discourse measure. For example, to successfully retell or summarize a story, a child must coordinate lexical, syntactic, semantic, and pragmatic language functions, as well as cognitive functions such as attention, memory, planning, sequencing, and organization to abstract and summarize the gist of the story.²⁵ Both children and adolescents with TBI frequently experience difficulty abstracting the gist meaning from the language input, which can negatively impact their cognitive communication competence in school (i.e., summarizing, outlining, identifying main points/ideas) and social settings (i.e., understanding the gist of conversations, jokes, puns, nonliteral language).^23–25 Research suggests that assessments using a variety of discourse tasks, communication partners, and contexts that are similar to a child’s daily life are likely to be most revealing of their communication skills.²⁶ Narrative tasks have the additional advantage that they can be used with young children. Most children have an idea of what a story is by age 3 years and should know the basic structure of a story by age 5 (i.e., character, setting, event, beginning, middle, and end), although narrative complexity and gist reasoning continues to develop into adolescence and adulthood.

This study examined the long-term consequences of early childhood TBI on cognitive communication competency during the transition to adolescence in a cohort of children injured between 3 years and 6 years 11 months of age. Few, if any, studies have investigated long-term narrative discourse or gist reasoning skills following early childhood TBI in the context of individual and social-contextual factors, nor have they considered the association of discourse proficiency (i.e., communication competence) with psychosocial functioning. Consistent with MacDonald’s model of influences on communication competence, we considered discourse communication skills within a complex interplay of individual, contextual, and psychosocial factors.¹⁷We hypothesized that children with TBI would display deficits on the narrative discourse task, including gist reasoning, compared to children with orthopedic injury (OI), and that children with severe TBI would show the greatest impairments. We also investigated whether individual (age at injury, injury severity), cognitive-affective (social information processing), cognitive-linguistic (language pragmatics), and contextual factors (home environment, SES) were predictors of discourse competency. Finally, we explored associations of discourse measures (information structure, language structure, and gist) with long-term psychosocial functioning.

Methods

Study Design

As part of a larger, prospective study examining long-term functional outcomes following early childhood TBI, Children and caregivers participated in assessments shortly after injury; 6, 12, and 18 months post-injury; and at a longer term follow up 5 to 8 years post-injury (average 6.8 years). The original parent study enrolled 206 children with a severe TBI (n = 23), moderate TBI (n = 64), or OI (n = 119). At the longer-term follow-up, participants included 130 children and their caregivers, including 16 with severe TBI, 42 with moderate TBI, and 72 with OI. The discourse measure was administered to 118 of the 130 children, including 54 children with TBI (14 severe and 40 moderate) and 64 with OI (see Table 1 for demographic characteristics of the sample). Discourse data were missing for 12 children; one did not complete the task and audio files for 11 were lost or deleted.

Table 1.

Group demographic and family characteristics

Group	Severe TBI n = 14	Moderate TBI n = 40	OI n = 64
Age at injury in years, mean (range)	4.86 (3.64–6.52)	5.22 (3.06–6.93)	5.06 (3.15–6.88)
Age in years at 7 year post-injury assessment, mean (range)	11.89 (10.66–13.62)	11.98 (10.13–15.17)	11.80 (10.41–14.09)
Years from injury to assessment, mean (SD)	7.02 (1.11)	6.75 (1.09)	6.73 (.88)
Current grade, mean (range)	5.86 (5–7)	6.08 (5–9)	5.92 (5–8)
Males, n (%)	8 (57%)	24 (60%)	34 (53%)
Caucasian, n (%)	8 (57%)	30 (75%)	47 (73%)
Maternal education greater than high school, n (%)	6 (42%)	23 (57%)	42 (65%)
Census Tract median family income, mean (SD)	$53,681 ($23,222)	$66,953 ($26,865)	$67,126 ($24, 249)

Note. TBI=traumatic brain injury, OI=orthopedic injury

Participants

Cincinnati Children’s Hospital Medical Center assumed institutional review board responsibility (20090128) for the research, and informed consent was obtained prior to participation. Children who sustained TBI or OI between the ages of 3 years, 0 months and 6 years, 11 months were recruited from 2003 to 2006 from three tertiary care children’s hospitals and a general hospital, all with Level 1 trauma centers located in Cincinnati, Columbus, and Cleveland, Ohio, United States. Eligibility criteria included English as the primary language spoken at home and no documentation of child abuse as a cause of injury. Children with a history of autism, mental retardation, or neurological disorder prior to injury were excluded. Eligibility for the TBI group included a blunt trauma to the head requiring overnight admission to the hospital and either a Glasgow Coma Scale²⁷ score <13 or a GCS of 13–15 with evidence for TBI-related brain abnormalities from computed tomography (CT) or magnetic resonance imaging (MRI). Consistent with previous investigations, ^{9, 28} severe TBI was defined as resulting in a GCS score of 8 or less, and moderate TBI was defined as a GCS score of 9–12 or a higher GCS score in the presence of abnormal neuroimaging. The GCS score assigned to the child was the lowest one recorded post-resuscitation. To estimate the effects of TBI while accounting for both pre-injury risk exposure and the experience of hospitalization, children admitted to hospitals for OI but without TBI were included as a comparison group. Inclusion criteria for the OI group were a documented bone fracture in an area of the body other than the head that required an overnight hospital stay and no evidence of loss of consciousness or other findings suggestive of TBI.

Procedures and Measures

Data were collected as part of a comprehensive evaluation of the child and family an average of 6.8 years post-injury as the child transitioned to early adolescence. As part of the larger assessment, children came to the medical center and completed clinical measures and the experimental narrative discourse measure as part of a 3–4 hour assessment battery. The entire assessment battery, including experimental narrative retell task, was independently scored by two individuals. The two individuals who scored the narrative task data were unaware of participant identity or injury group. Discrepancies were discussed and reconciled as necessary. The intra-class correlations for the scores for each of the discourse measure indices ranged from .82 to .96, reflecting good to excellent agreement.^29-30The intra-class correlations for gist was .96, which is excellent.

Discourse

As described in a previous report,²² the discourse measure used was a narrative story retell, ‘The Crow and the Peacock’ (see supplementary material). The task required children to listen to a conceptually complex, abstract story involving perspective taking and a moral. The children were provided with the following instructions: ‘I am going to read you a story called the Crow and the Peacock. Listen to the story carefully and try to remember as much as you can. When I am done, I am going to ask you to tell the story back to me. I am also going to ask you a few questions about the story’. After the story was read to the child, the examiner said, ‘Now you tell me the story. Tell me everything you can remember’. Once the children responded, the examiner commented, ‘Sometimes there are important lessons about life that we can learn from a story. What lesson can we learn from this story?’ Next, the children were asked four questions about the explicit content of the story (i.e., facts) and four questions about the implicit (i.e., inferential) content of the story. Finally, they were asked to rate 10 statements from the story as important or unimportant to the main message or lesson (five of each). The retelling was audio-recorded and later transcribed verbatim for analysis of language and information structures.³¹⁻³⁴The language structure domain focuses on lexical and sentential level of language production. ‘The Crow and the Peacock’ contains 287 words and 20 thematic units, with a thematic unit defined as an independent clause and all the dependent clauses that modify it (see supplementary material). Two language structure variables were analyzed: (1) the total number of words in the child’s story production and (2) the total number of thematic units. The overall score for language structure was the sum of the number of words and the number of thematic units. The information structure domain focuses on the information and meaning conveyed in the language production. The story consists of 34 informational units or core propositions and six gist propositions that were established a priori. The retelling of the story was coded with respect to whether each of these 40 propositions was present or not in the retell. Two information structure measures were analyzed: (1) the number of core propositions (out of 34) and (2) the number of gist propositions (out of six) in the child’s production. The overall score for information structure was the sum of the number of core propositions and the number of gist propositions. Gist propositions were used as a measure of gist reasoning. The primary outcome measures were the overall scores for language and information structure and gist propositions.

Pragmatics

The pragmatic judgment subtest of the Comprehensive Assessment of Spoken Language (CASL)³⁵ was administered as part of a larger test battery. The CASL was included in the data analyses as a standardized measure of language pragmatics and to measure cognitive skills similar to those assessed by the experimental narrative retell task. Scores are expressed as standard scores (M = 100, SD = 15).

Social Cognition

The Dodge Social Information Processing (SIP) Self-Report Stories³⁶ provides composite measures of 3 attributional styles of SIP: Assertive, Passive, and Aggressive. Each of the SIP composites consists of the total of responses from 6 stories. Each story is followed by 14 questions designed to elicit attributional style. Thirteen of the 14 questions were rated on a 5-point Likert scale (1 = Yes, definitely, 5 = No, definitely not). The Aggressive composite included items reflecting hostile attribution, situational evocation of anger, aggressive goal orientation, the likelihood there would be an aggressive response, and instrumental and interpersonal value attributed to the aggressive response. The Passive composite also included items assessing hostile attribution as well as those reflecting situational evocation of sadness, passive goal orientation, the likelihood that there would be a passive response, and instrumental and interpersonal value attributed to the nonassertive response. The Assertive composite included items reflecting a nonhostile attribution, an assertive/prosocial goal orientation, the likelihood that there would be a prosocial response, and the instrumental and interpersonal value attributed to an assertive/prosocial response. The Passive SIP score had unacceptably low internal consistency.³⁷Consequently, only the Assertive and Aggressive attributional styles were examined as predictors of discourse skills.

Home Environment

The quality of the home environment was assessed using the early adolescent version of the Home Observation for Measures of the Environment (EA HOME).³⁸ The EA HOME was designed for use with children between the ages of 10 and 15 years. It was administered in the child’s home at the long-term follow-up visit. The EA HOME integrates parent interviews with direct observations of characteristics of the home (e.g., space and availability of books) and the quality of the parent-child interactions. The EA HOME has been shown to predict longer-term developmental outcomes with satisfactory validity and reliability.^39–40 The total score was used in the current study, with higher scores indicative of a better quality home environment.

Psychosocial Outcomes

Parents rated their child’s social competence using the Home and Community Social and Behavior Scales (HCSBS),⁴¹ which has demonstrated satisfactory reliability and validity. ⁴²Analyses were based on the social competence total, expressed as a T-score, which by definition has M = 50, SD = 10.

Everyday Functioning

The Child and Adolescent Functional Assessment Scale (CAFAS)⁴³ was used to evaluate everyday functioning. The CAFAS is administered via structured interview with caregivers and yields standardized ratings for functioning across domains and settings. The CAFAS provides ratings for functioning in 8 domains: school, home, community, behavior toward others, moods/emotions, self-harmful behaviors, substance abuse, and thinking. Functioning in each domain is rated on an ordinal scale, ranging from 0 (unimpaired) to 30 (severe impairment) in 10-point increments. A total score is created by summing domain scores (range: 0–240), with scores less than or equal to 50 considered to be “unimpaired” and those greater than 50 as “impaired.” The CAFAS has established validity and excellent interrater reliability, ranging from 0.74 to 0.99.⁴⁴

Analysis

Group comparisons (OI, severe TBI, moderate TBI) were conducted using analysis of variance for the continuous measures of clinical outcome (pragmatic language, SIP, HSCBC, and CAFAS), with post hoc comparisons using Tukey HSD test. Discourse indices (language structure, information structure, gist, memory for story content, recognition of story meaning, and moral of the story) were correlated with CASL Pragmatic Language, SIP composites, and the behavioral measures. Analyses of covariance were conducted to compare groups on discourse measures controlling for age at injury. Two planned single degree-of-freedom Helmert contrasts (Severe TBI vs. OI and Severe TBI vs. Moderate TBI) were used for each analysis of covariance. Hierarchical linear regressions were conducted to examine associations of discourse measures (information structure, language structure, and gist) controlling for age at injury, group, and measures of the clinical outcomes and the home environment. Interactions of group with the latter measures were not significant and were thus excluded from the final models. An alpha level of .05 was used for statistical significance. IBM SPSS 24 was used for all analyses.

Results

Group differences in demographic and family characteristics and clinical outcomes

Demographic and family characteristics

Group comparison (OI, Severe TBI, Moderate TBI) were conducted using chi-square analyses for dichotomous demographic variables (sex, race, maternal education) and analysis of variance for continuous variables (age at injury, age at assessment, time from injury to assessment, current grade in US school system, census median family income). None of the group comparisons were significant (see Table 1 for details).

Clinical outcomes

As indicated in Table 2, the severe TBI group had significantly lower scores than the OI group on all measures except for the SIP composites. The severe TBI group also performed significantly worse than the moderate TBI group on all measures other than the SIP composites, EA-HOME total, and CAFAS total. The moderate TBI group differed significantly from the OI group on the EA-HOME total and the CAFAS Total.

Table 2.

Group differences on measures of clinical outcome

Group	Severe TBI n = 14	Moderate TBI n = 40	OI n = 64	F	p
CASL Pragmatic Judgment a,b	93.93 (19.44)	104.32 (15.13)	105.86 (12.50)	3.99	.021
SIP Aggressive Composite	2.81 (.72)	2.65 (.66)	2.53 (.59)	1.22	.299
SIP Assertive Composite	3.80 (.36)	3.55 (.71)	3.65 (.43)	1.03	.359
HCSBS Social Competence- Social Competence Total a,b	43.79 (11.53)	50.95 (10.60)	54.48 (8.03)	7.74	.001
CAFAS Total a,c	49.29 (38.52)	33.75 (35.35)	14.84 (18.68)	11.43	.000
EA-HOME Total a,c	45.5 (6.21)	45.23 (9.20)	49.45 (7.37)	4.02	.02

Note. Significant (p < .05) post hoc group differences using Tukey’s HSD test indicated by superscript letters.

^aSevere TBI and OI group,

^bSevere and Moderate TBI group, or

^cModerate TBI and OI group difference significant. TBI=traumatic brain injury, OI=orthopedic injury, CASL= Comprehensive Assessment of Spoken Language, SIP= Social information processing, HCSBS=Home & Community Social Behavior Scales, CAFAS= Child and Adolescent Functional Assessment Scale, EA-HOME= Early Adolescent Home Observation for Measures of the Environment.

Associations of discourse performance with pragmatic language, SIP composites, and psychosocial outcomes

As shown in Table 3, discourse indices were primarily correlated with each other and with CASL Pragmatic Judgment. Discourse gist was correlated with all other discourse measures and all clinical measures except for the SIP Assertive Composite. The SIP Assertive Composite was not correlated with any of the measures.

Table 3.

Bivariate correlations between measures for all groups collapsed

	1	2	3	4	5	6	7	8	9	10	11	12
1. Language	-
2. Information	.85**	-
3. Gist	.53**	.75**	-
4. Explicit	.33**	.49**	.59**	-
5. Implicit	.23*	.31**	.44**	.54**	-
6. Important	.23*	.34**	.37**	.26**	.22*	-
7. Unimportant	.26*	.33**	.32**	.16	.21*	.47**	-
8. Pragmatic Judgment	.40*	.50**	.51**	.50**	.47**	.37**	.35**	-
9. SIP Aggressive	−.10	−.26**	−.27**	−.32**	−.14	−.13	−.09	−.24**	-
10. SIP Competent	−.12	−.09	.04	−.005	−.01	.10	.08	.13	−.06	-
11. HCSBS	.14	.23*	.38**	.21*	.14	.26**	.16	.32**	−.23	.15	-
12. CAFAS	−.18*	−.25	−.25**	−.17	−.11	−.14	−.16	−.32**	.25**	.03	−.64**	-

^*p < .05,

^**p < .01, SIP= Social information processing, HCSBS= Home & Community Social Behavior Scales, CAFAS= Child and Adolescent Functional Assessment Scale

Group differences in discourse performance

Table 4 presents group means and standard errors for the discourse measures including language structure (T-units + total words), information structure (core + gist propositions), memory for story content, recognition of story meaning, and moral of the story. As reported in Table 5, two planned single degree-of-freedom Helmert contrasts (Severe TBI vs. OI and Severe TBI vs. Moderate TBI) revealed that the Severe TBI group performed significantly worse than both the OI group and the Moderate TBI group on all discourse measures.

Table 4.

Means and standard error on discourse measures co-varying for age at injury

	Severe TBI n = 14		Moderate TBI n = 40		OI n = 64
Measure	Mean	SE	M	SE	M	SE
Language structure score (T –units + words) a,b	74.64	11.55	114.80	6.85	110.17	5.41
Number of thematic units a,b	6.79	1.00	9.78	.59	9.20	.47
Word count a,b	67.84	10.64	105.01	6.31	100.96	4.98
Information structure score (core + gist, out of 40) a,b	7.86	1.68	15.48	.99	14.61	.78
Total number of core propositions (out of 34) a,b	5.86	1.42	11.82	.84	11.07	.66
Total number of gist propositions (out of six) a,b	2.00	.39	3.66	.23	3.53	.18
Memory for story content
Total score on explicit questions (out of eight) a,b	5.22	.47	6.67	.28	6.50	.23
Total score on implicit questions (out of eight) a,b	3.77	.65	5.39	.38	5.21	.32
Recognition of story meaning
Correct identification of important statements (out of five) a,b	3.57	.22	4.46	.13	4.46	.10
Correct identification of unimportant statements (out of five) a,b	3.06	.26	4.45	.15	4.49	.12
Moral of story a,b	4.07	.88	6.74	.52	6.45	.41

Note. Significant (p < .05) group differences indicated by superscript letters.

^aSevere TBI and OI group,

^bSevere and Moderate TBI group. TBI=traumatic brain injury, OI=orthopedic injury, T-unit= Thematic unit

Table 5.

Group differences (Severe TBI vs. OI and Severe TBI vs. Moderate TBI)

	Severe TBI vs. OI					Severe TBI vs. Moderate TBI
				95% CI					95% CI
Measure	F	p	η2	Lower	Upper	F	p	η2	Lower	Upper
Language overall score (T –units + words)	7.05	.01	.08	79.8	105.8	8.65	.005	.14	81.53	107.90
Number of thematic units	4.19	.04	.05	6.87	9.20	6.92	.01	.12	7.09	9.37
Word count	7.25	.009	.08	72.85	96.75	8.67	.005	.14	74.32	98.63
Information overall score (core + gist, out of 40)	12.22	.001	.14	9.40	13.18	16.74	.000	.24	9.83	13.50
Core propositions (out of 34)	11.01	.001	.12	6.97	10.04	13.04	.001	.20	7.23	10.46
Gist propositions (out of six)	10.08	.002	.11	2.30	3.25	16.26	.000	.24	2.40	3.23
Memory for story content
Explicit questions (out of eight)	4.75	.03	.06	5.28	6.45	8.57	.005	.14	5.44	6.42
Implicit questions (out of eight)	3.87	.05	.06	3.76	5.20	4.65	.03	.08	3.81	5.36
Recognition of story meaning
Important statements (out of five)	12.07	.001	.14	3.75	4.27	8.40	.006	.14	3.70	4.32
Unimportant statements (out of five)	21.54	.000	.23	3.48	4.08	12.57	.001	.20	3.37	4.15
Moral of story	5.31	.02	.06	4.30	6.27	5.65	.02	.10	4.48	6.48

Note. TBI=traumatic brain injury, OI=orthopedic injury

Association of clinical outcomes with discourse performance

Six hierarchical linear regressions were conducted to examine associations of the discourse performance (language structure, information structure, and gist) with child and injury characteristics. For the first three regressions, age at injury was entered in the first step. Dummy variables representing planned independent contrasts that compared the severe TBI group to the OI group and the moderate TBI group to the OI group were entered as predictors in the second step. CASL pragmatic judgment and the two SIP composites were entered in the third and fourth steps, respectively, to determine if these measures predicted discourse language skills independent of group.

Table 6 reports findings from regression analysis examining CASL Pragmatic Language and the two SIP composites as predictors of discourse performance. For the model predicting the language composite, age at injury entered in step 1 was not significant, F change (1, 113) = 1.07, p = .30. The addition of the group contrast in step 2 accounted for 6% of the variance, F change (2, 113) = 3.47, p = .03, indicating that the Severe TBI group had significantly lower scores than the OI group. The addition of the CASL pragmatic judgment in step 3 was not significant, F change (1, 112) = .025, p = .08. The Assertive and Aggressive SIP scores entered in step 4, and were also not significant, F change (2, 110) = 1.26, p = .28.

Table 6.

Summary of hierarchical regression analysis examining age at injury, group, CASL Pragmatic Language, and the SIP composites as predictors of discourse performance (n = 118)

	Dependent variable (standardized β)
Predictors	Language composite	Information composite	Gist
Step 1
Age at injury	.106	.085	.057
R2 Change	.011	.007	.003
Total R2	.011	.007	.003
Step 2
Age at injury	.091	.064	.037
Severe TBI vs. OI	−.228*	−.302*	−.300*
Moderate TBI vs. OI	.037	.054	.014
R2 Change	.057	.102	.092
Total R2	.069	.109	.095
Step 3
Age at injury	.090	.063	.109
Severe TBI vs. OI	−1.03*	−.877*	−2.336*
Moderate TBI vs. OI	.037	.054	.559
Pragmatic judgment	.825	.586	5.07*
R2 Change	.025	.013	.171
Total R2	.094	.122	.267
Step 4
Age at injury	.085	.061	.377
Severe TBI vs. OI	−1.65	−.894	−2.171*
Moderate TBI vs. OI	.037	.054	.687
Pragmatic judgment	.848	.620	4.459*
SIP Aggression	.559	.153	−1.907
SIP Assertive	.055	−.165	.093
R2 Change	.020	.002	.024
Total R2	.114	.124	.291

^*p < .05, TBI=traumatic brain injury, OI=orthopedic injury CASL=Comprehensive Assessment of Spoken Language, SIP= Social information processing

In the model predicting information structure, age at injury entered in step 1 was not significant, F change (1, 115) = .83, p = .36. The addition of group contrast in step 2 accounted for 10% of the variance, F change (2, 113) = 6.46, p = .002, indicating the Severe TBI group had significantly lower scores than the OI group. Associations with discourse information structure were not evident for either the CASL Pragmatic Judgment entered in step 3, F change (1, 112) = 1.60, p = .20, nor for the Assertive and Aggressive SIP scores entered in step 4, F change (2, 110) = .111, p = .89.

For the model predicting gist, age at injury entered in step 1 was not significant, F change (1, 113) = .366, p = .54. The addition of the group contrast in step 2 accounted for 9% of the variance, F change (2, 111) = 5.64, p = .005, indicating that the Severe TBI group had significantly lower scores than the OI group. The addition of the CASL pragmatic judgment in step 3 accounted for 17% of the variance, F change (1, 110) = 25.70, p = .000, indicating that higher pragmatic judgment scores are associated with better gist scores. The Assertive and Aggressive SIP scores entered in step 4, were not significant, F change (2, 108) = 1.83, p = .16.

Table 7 reports findings from next three regression analysis examining SES and the EA-HOME as predictors of discourse performance. For the model predicting language structure performance, age entered in step 1 was not significant. The addition of group in step 2 was significant, indicating that the Severe TBI group had significantly lower language structure scores that the OI group, F change(2, 114) = 4.43, p = .01. The association of language structure performance with SES and EA-HOME scores in step 3 was not significant, F change(2, 112) = 1.99, p = .14.

Table 7.

Summary of hierarchical regression analysis examining age at injury, group, SES, and home environment (EA HOME) as predictors of discourse performance (n = 118)

	Dependent variable (standardized β)
Predictors	Language composite	Information composite	Gist
Step 1
Age at injury	.120	.099	.075
R2 Change	.014	.010	.006
Total R2	.014	.010	.006
Step 2
Age at injury	.097	.070	.048
Severe TBI vs. OI	−.256*	−.328*	−.323*
Moderate TBI vs. OI	.036	.053	.028
R2 Change	.071	.119	.109
Total R2	.086	.128	.115
Step 3
Age at injury	.104	.079	.060
Severe TBI vs. OI	−.224*	−.279*	−.258*
Moderate TBI vs. OI	.073	.092	.077
SES z-scores	.061	.179	.247*
EA-HOME	.138	.123	.149
R2 Change	.031	.071	.123
Total R2	.117	.199	.238

^*p < .05, TBI=traumatic brain injury, OI=orthopedic injury, SES=Socioeconomic status, EA-HOME= Early Adolescent Home Observation for Measures of the Environment.

For the model predicting information structure performance, age entered in step 1 was not significant. The addition of group in step 2 was significant, indicating that the Severe TBI group had significantly lower language composite scores than the OI group, F change(2, 114) = 7.75, p = .001. The association of discourse performance with SES and EA-HOME in step 3 was not significant, F change(2, 112) = 4.95, p = .10.

For the model predicting gist, age entered in step 1 was not significant. The addition of group in step 2 was significant, indicating that the Severe TBI group had significantly lower gist scores than the OI group, F change(2, 114) = 7.03, p = .001. The association of gist reasoning with SES and EA-HOME scores in step 3 was significant, F change(2, 112) = 9.05, p = .000, indicating that higher SES was associated with better gist scores.

Associations of discourse performance with psychosocial outcomes

Six hierarchical linear regressions were conducted to examine whether discourse performance was associated with psychosocial outcomes (HBSBS: Social Competence Total and CAFAS Total). Age and dummy variables representing contrasts of the Severe TBI group to the OI group and the Moderate TBI group to the OI group were entered as predictors in the first step for all models. As indicated in Tables 8–10, TBI was a significant predictor of psychosocial outcomes in all three models. Severe TBI was associated with poorer outcomes on both the HCSBS and CAFAS in all three models, and moderate TBI with poorer functioning on the CAFAS in all three models and with the HCSBS in the gist model.

Table 8.

Summary of hierarchical regression analysis with age, group, and discourse information structure as predictors of psychosocial outcomes (n = 118)

	Dependent variable (standardized β)
Predictors	HCSBS Total	CAFAS Total
Step 1
Age at injury	.015	.038
Severe TBI vs. OI	−.349*	.370*
Moderate TBI vs. OI	−.170	.293*
R2 Change	.120	.167
Total R2	.120	.167
Step 2
Age at injury	.004	.052
Severe TBI vs. OI	−.297*	.306*
Moderate TBI vs. OI	−.180	.304*
Information	.161	−.196*
R2 Change	.023	.033
Total R2	.142	.201

^*p < .05, TBI=traumatic brain injury, OI=orthopedic injury, HCSBS=Home & Community Social Behavior Scales, CAFAS= Child and Adolescent Functional Assessment Scale

Table 10.

Summary of hierarchical regression analysis with age, group, and discourse gist as predictors of psychosocial outcomes (n = 118)

	Dependent variable (standardized β)
Predictors	HCSBS Total	CAFAS Total
Step 1
Age at injury	.015	.038
Severe TBI vs. OI	−.349*	.370*
Moderate TBI vs. OI	−.170	.293*
R2 Change	.120	.167
Total R2	.120	.167
Step 2
Age at injury	.001	.047
Severe TBI vs. OI	−.247*	.307*
Moderate TBI vs. OI	−.181*	.299*
Gist	.321*	−.195*
R2 Change	.091	.034
Total R2	.211	.201

^*p < .05, TBI=traumatic brain injury, OI=orthopedic injury, HCSBS=Home & Community Social Behavior Scales, CAFAS= Child and Adolescent Functional Assessment Scale

In the first two models with age at injury and group contrasts in the model, the entry of the information structure in step 2 accounted for an additional 3% of the variance in the CAFAS, F change(1, 113) = 4.73, p = .03, but was not significantly associated with the HCSBS. For the second two models, with age at injury and group contrasts in step 1, the addition of language structure in step 2 was not significantly associated with the HCSBS, F change(1, 112) = .611, p = .43, or the CAFAS, F change(1, 113) = 2.383, p = .12. In the last two models with age at injury and group contrasts in the model, the entry of the gist in step 2 accounted for an additional 18% of the variance in the HCSBS, F change(1, 113) = 4.73, p = .03, and 17% of the variance in the CAFAS, F change(1, 113) = 4.77, p = .031, indicating that higher gist scores were associated with better psychosocial functioning.

Discussion

The goal of this study was to examine the effects of TBI on cognitive communication skills using a narrative discourse task an average of 7 years after early childhood injury. Relative to children with moderate TBI and OI, children with severe TBI produced narratives with less language (i.e., number of words and T-units), as well as reduced story information (core propositions) and central semantic meaning (gist propositions). They also had poorer memory for story content (explicit and implicit questions), more difficulty recognizing important and unimportant story details, and difficulty inferring the underlying moral of the story. These findings are consistent with previous reports documenting impaired discourse in children with severe TBI up to three years post-injury and suggest that discourse skills are sensitive markers of the long term effects of severe TBI in early childhood.^{22–25, 31–34, 45–46} Children with severe TBI also had lower scores on a test of pragmatic judgment compared to both the moderate TBI and OI groups, and performance on this measure was associated with gist reasoning. SES was also shown to be associated with gist, which is consistent with previous research showing that social and environmental factors significantly contribute to children’s cognitive, behavioral, and language outcomes following TBI.^10,12 Children with severe TBI demonstrated worse psychosocial functioning compared to both moderate TBI and OI groups and lower scores on a measure of discourse gist were associated with poorer social competence on the HCSBS and daily functioning on the CAFAS. In sum, children with severe TBI appear to have deficits in language, information, and gist reasoning many years after injury that are associated with psychosocial functioning in everyday settings. Importantly, discourse gist was the most sensitive discourse measure in predicting both social competence and impairments in every day functioning. These findings support previous research, while extending the findings to the longer-term effects of severe injury in early childhood.²³⁻²⁴Additionally, they underscore the centrality of gist reasoning to long-term functioning post early childhood TBI.

MacDonald’s model of cognitive communication competency posits that cognitive, linguistic, emotional, individual, and contextual factors contribute to discourse skills.¹⁷ Several of the findings provide information on the nature of these contributions. First, with regard to discourse functioning, the absence of an association of age at injury with discourse skills at this longer post-injury follow-up contrasts with previous evidence of poorer discourse performance in children injured at younger ages assessed up to 18 months and 3 years post injury.^22,34 The absence of significant effects of age at injury may be due to the narrow age range of our cohort as well as the fact that the cohort was nearly seven years post injury on average. Second, the results confirm associations of injury severity with long-term impairments in cognitive, linguistic, and psychosocial functioning and suggest that some of these impairments are related to poorer discourse skills, specifically gist reasoning. These results are consistent with MacDonald’s model and previous research that characterizes gist reasoning as a higher-order cognitive function that relies on a complex interplay of linguistic, cognitive, personal, and contextual factors.^7,17 Furthermore, gist reasoning entails a constructive/integrative process, and while children with TBI are able to process information in a bottom-up fashion, they continue to demonstrate difficulty engaging in more integrated top-down processing to abstract gist meaning.^23–25 Third, children who sustained a moderate TBI in early childhood did not show the same long-term impairments as children with severe TBI but had outcomes more similar to children in the OI group. These results are consistent with other research indicating that children with severe TBI perform significantly worse than children with mild/moderate TBI on narrative discourse tasks.^25,34 Finally, although children with moderate TBI preformed like children in the OI group on cognitive and linguistic measures, including the discourse task, they were more impaired than the OI group on the CAFAS, indicating persistent deficits in behavioral and everyday functioning. Discourse information structure and gist were also associated with CAFAS scores, suggesting that cognitive communication skills may contribute to poorer psychosocial functioning even in some children with moderate TBI. Consequently, both moderate and severe TBI in early childhood can put children at an increased risk for long-term psychosocial impairments.

We found discourse skills, specifically gist, were significantly associated with children’s social communication skills (pragmatics) as measured on standardized testing. These findings are similar to the literature relating discourse to other cognitive and neuropsychological skills.^22,24,46 Our results also found an association between children’s ability to abstract gist meaning from the narrative discourse task and psychosocial functioning including social competence and daily functioning. These results provide additional evidence for an association between discourse skills and psychosocial functioning up to 7 years after injury. Similar relationships between discourse skills and psychosocial functioning have been reported for adults with TBI indicating that impaired discourse skills were significantly related to poorer levels of social integration and quality of life.²¹ For children with TBI, Chapman and colleagues reported a significant relationship between narrative discourse skills and perceived communicative competence on a well-validated psychosocial measure 1 year after injury.²⁰ Our results provide additional support for relationships among linguistic, cognitive, and psychosocial abilities that are commonly postulated in children with TBI.^19,46–47 Given the extended length of time since injury in our sample, cognitive-communication impairments that contribute to poor psychosocial functioning after TBI may present long-term challenges for children who sustain severe injuries in childhood. Unfortunately, these deficits can have debilitating effects on new learning and peer interactions, resulting in increasing gaps between injured children and their peers. Therefore, future studies should continue to examine the relationship between discourse gist and academic and social outcomes after TBI in longitudinal studies.

Our findings should be interpreted in light of several limitations. The small size of the severe TBI group reduced the statistical power of our analyses. However, our results are consistent with previous research that indicates that children with severe TBI experience chronic cognitive communication impairments and the gap between the moderate and severe TBI groups may widen as social and higher order cognitive skills fail to develop in children with a history of severe TBI.^48–50 On the other hand, the cognitive and linguistic measures used in our study may not have been sensitive enough to detect subtle but clinically significant impairments in the moderate TBI group. Although we anticipated that the narrative discourse task would be a more sensitive measure of cognitive communication competence, the task may not have been challenging enough to detect impairments in the moderate TBI group as they transitioned into adolescence. However, discourse gist, was significantly associated with social competence and everyday psychosocial functioning independent of group. Future studies should continue to examine gist reasoning and other discourse indices in more complex narratives (greater length and conceptual complexity) and different types of discourse tasks that reflect communication in real life settings (e.g., expository, persuasive). For example, expository discourse has been called the “language of curriculum” and persuasive discourse occurs in both academic (e.g., school debates, essays) and social situations (e.g., convincing a friend to try a new coffee shop, urging your parents to buy you a smart phone). Adolescents with TBI perform significantly worse than their typically developing peers in persuasive discourse, summarizing both compare-contrast and cause-effect expository lectures, and in summarizing procedural discourse passages, which all require “getting the gist”.^19,51–52 Therefore, it appears that discourse gist may be the most sensitive discourse measure for detecting cognitive communication competence and psychosocial functioning after TBI.

The current findings have potential implications for the assessment, follow-up, and treatment of cognitive communication disorders after pediatric TBI. Clinically, this study provides a novel and multifaceted paradigm of cognitive communication competence that assess the long-term consequences of pediatric TBI. Particularly, discourse gist, as measured by 6 items of central semantic meaning, may serve to further elucidate the long-term cognitive communication and psychosocial functioning impairments after TBI. Our results suggest that discourse skills, especially gist reasoning, should be considered as a target for speech and language therapy and cognitive rehabilitation. Studies that have examined gist reasoning training (versus information based training) for adolescents with chronic stage TBI have reported significant gains in their ability to abstract meaning from complex information, as well as significant improvements in executive functioning and participation in daily activities.^53–54 Therefore, interventions targeting gist reasoning may have ecological value for improving the long-term consequences of TBI, and future studies should focus on developing metrics for assessing gist reasoning.

In conclusion, our findings suggest that children who sustained a severe TBI in early childhood are at risk for long-term cognitive communication impairments that may be related to a complex interplay of injury, individual, and social factors. Additionally, cognitive communication impairments after severe TBI may negatively influence children’s psychosocial functioning. We found that children who sustained a moderate TBI in early childhood did not show the same long-term cognitive communication impairments as the severe TBI group. However, children with severe TBI and even those with moderate TBI who have deficits in discourse gist may continue to experience poorer psychosocial functioning several years after injury. Given these finding, gist reasoning may be a promising and sensitive clinical measure to characterize functionally relevant complex cognitive communication impairments after pediatric TBI. Future research is needed to determine if cognitive communication impairments resulting from childhood TBI persist into adulthood and if they are associated with poorer psychosocial outcomes, including lower educational and occupational attainments.

Table 9.

Summary of hierarchical regression analysis with age, group, and discourse language structure as predictors of psychosocial outcomes (n = 118)

	Dependent variable (standardized β)
Predictors	HCSBS Total	CAFAS Total
Step 1
Age at injury	.015	.038
Severe TBI vs. OI	−.349*	.370*
Moderate TBI vs. OI	−.170	.293*
R2 Change	.120	1.67
Total R2	.120	.167
Step 2
Age at injury	.008	.051
Severe TBI vs. OI	−.331*	.335*
Moderate TBI vs. OI	−.172	.298*
Language	.072	−.137
R2 Change	.005	.017
Total R2	.125	.185

^*p < .05, TBI=traumatic brain injury, OI=orthopedic injury, HCSBS=Home & Community Social Behavior Scales, CAFAS= Child and Adolescent Functional Assessment Scale

Implications for Rehabilitation.

• The assessment and treatment of cognitive communication impairments after traumatic brain injury are complex and require consideration of individual, contextual, and psychosocial factors.

• Discourse, in general, and gist reasoning specifically, is a sensitive measure for detecting cognitive communication competence several years after pediatric traumatic brain injury.

• Treatments to address gist reasoning deficits may support improvements in global and social functioning