Cognitive Performance, Depression, and Anxiety One-Year After Traumatic Brain Injury

Abstract

Objectives:

To evaluate associations between depression, anxiety, and cognitive impairment among individuals with complicated mild to severe TBI one year after injury.

Setting:

Multiple inpatient rehabilitation units across the United States.

Participants:

498 adults aged 16 and over who completed inpatient rehabilitation for complicated mild to severe TBI.

Design:

Secondary analysis of a prospective, multi-center, cross-sectional observational cohort study.

Main Measures:

Assessments of depression (TBI-QOL Depression) and anxiety (TBI-QOL Anxiety) as well as a telephone based brief screening measure of cognitive functioning (Brief Test of Adult Cognition by Telephone; BTACT).

Results:

We found an inverse relationship between self-reported depression symptoms and the BTACT composite score (β=−.18, p<.01) and anxiety symptoms and the BTACT composite score (β=−.20, p<.01). There was no evidence this relationship varied by injury severity. Exploratory analyses showed depression and anxiety were negatively correlated with both BTACT Executive Function factor score and BTACT Memory factor score.

Conclusion:

Both depression and anxiety have a small but significant negative association with cognitive performance in the context of complicated mild to severe TBI. These findings highlight the importance of considering depression and anxiety when interpreting TBI-related neuropsychological impairments, even among more severe TBI.

INTRODUCTION

Disability from complicated mild to severe traumatic brain injury (TBI) arises from a combination of cognitive, physical, and psychosocial impairments^1,2 that can negatively impact quality of life.^2–5 Depression and anxiety affect 20–30% of people with TBI ^6–12 and may contribute to poorer functional outcomes directly,^13–15 or indirectly via an adverse impact on cognitive functioning.¹⁴ Research on the relationship between psychological distress and cognitive test performance in TBI has grown substantially in recent years but remains limited by several methodological weaknesses, which this study attempts to address.⁷

In the general population, depression and anxiety are related to worse neuropsychological test performance.^16–19 Depression is associated with poorer attention, working memory, executive functions, and retrieval-based memory.^16,17 State anxiety is related to diminished working memory²⁰ whereas chronic stress has adverse impacts on brain structure and functioning and may increase the risk of developing major neurocognitive disorders in older age.^21,22 Anxiety and depression are highly comorbid conditions and have similar underlying mechanisms that are suspected to reduce cognition via dopaminergic, hypothalamic-pituitary-adrenal axis, frontal-subcortical, and genetic pathways. ²³

Some evidence suggests a relationship between cognition and depression or anxiety among people with TBI. A recent meta-analysis of 61 studies found a small but statistically significant correlation between depression and worse functioning in the domains of attention, memory, language, executive functioning, and processing speed.²⁴ Anxiety was significantly associated with poorer performance on measures of immediate and recent verbal memory, overall memory, executive functioning, and processing speed.²⁴ In the meta-analysis, injury severity appeared to moderate these relationships such that depressive and anxiety symptoms had a negative relationship with cognition among those with mild TBI but less so among those with severe TBI. The authors proposed severe TBI may cause worse cognitive deficits that mask any additional effects of emotional symptoms and/or results in poor insight so fewer self-reported mood symptoms. Importantly, the authors also found stronger relationships between mood symptoms and cognitive functioning among studies that excluded subjects who failed validity testing.²⁴

There were several limitations in the studies in which the meta-analysis was based.²⁴ Most studies were quite small, and few included substantial numbers of severely injured participants, precluding investigation of effect modifiers and making it difficult to document representativeness of the sample. In fact, of the 61 studies included in the meta-analysis, only 10 (16%) consisted of exclusively moderate to severe TBI samples pointing to the significant uneven representation of mild TBI in this literature base. With few exceptions, the scales or diagnostic criteria used to measure depression and anxiety contain questions about cognitive impairment, sleep, fatigue, and functional impairment that could confound the relationship between these psychological conditions and cognition. Lastly, many of the studies drew participants from clinical and rehabilitation settings where performance and symptom validity issues may be more prominent.

The objective of the current study was to investigate the associations of cognitive test performance with depression and anxiety in a large sample of persons with complicated mild to severe TBI. We hypothesized that more severe depression and anxiety symptoms would be associated with poorer cognitive functioning even when the anxiety and depression measures lacked confounding items. In addition, we hypothesized the negative effect of depression and anxiety would be greater among those with less severe injuries.

METHODS

Participants

This study is the secondary analysis of a prospective, multi-center, cross-sectional observational cohort study. Data for this project were gathered through the TBIMS National Database, a multicenter longitudinal prospective cohort study. Detailed information about the database including enrollment, data collection techniques, variables recorded, and participating centers can be found at the TBIMS website (www.tbindsc.org). In brief, individuals 16 years or older may be included in the multisite database if they sustained a moderate to severe TBI as measured by posttraumatic amnesia (PTA) greater than 24 hours, loss of consciousness (LOC) greater than 30 minutes, Glasgow Coma Scale (GCS) score of less than 13, or intracranial neuroimaging abnormalities. Participants must be 16 years or older at time of injury, admitted to the system’s hospital emergency department within 72 hours of injury, and received acute care and comprehensive inpatient rehabilitation within the model system hospitals. In the United States, patients are typically admitted to inpatient rehabilitation because severity of symptoms and functional decline prevent a safe return home; however, there are a number of social, economic, and environmental factors that may guide admission decisions for inpatient rehabilitation.

The sample included in this dataset were 498 individuals enrolled in the TBI Model Systems (TBIMS) and had completed the anxiety and depression subscales of the Traumatic Brain Injury Quality of Life (TBI-QOL) measure, the Brief Test of Adult Cognition by Telephone (BTACT) at one-year post-injury, and had data regarding duration of PTA, which was used as indicator of TBI severity. Our analytic sample was derived from a modular study consisting of a subset of the TBIMS National Database and as well as data from the core TBIMS National Database (Figure 1). Both samples underwent the same recruitment and research procedures. All participants or their proxies completed an informed consent process and each participating center obtained local Institution Review Board approval including Mt. Sinai Icahn School of Medicine and the University of Washington.

Figure 1.

Flow chart of participant inclusion and analysis.

Procedure

Trained TBIMS research assistants collected information regarding injury severity and medical course from hospital and emergency medical service records. Demographic information such as date of birth, education, and pre-morbid functioning were collected in interviews with the subjects or family/significant others. Cognitive functioning, depression, and anxiety symptom data used in this study were collected by telephone at one-year follow-up (within a two-month window, i.e., 10 to 14 months post-injury).

Measures

Demographic and Injury-Related Variables.

Demographic information and injury-related characteristics were obtained during standard data collection for TBIMS using a semi-structured interview during acute rehabilitation and medical record abstraction. Participants self-reported age, sex (male or female), ethnicity (Hispanic/Latino/Spanish), and race (white, black or African American, Asian, American Indian or Alaskan Native, or Native Hawaiian or Other Pacific Islander). Years of education was collected and used as a continuous variable.

Injury Severity.

Trained TBIMS research assistants collected information regarding injury severity. Although multiple sources of injury severity were collected (GCS, time to follow commands, PTA), the PTA had the most complete data and was most consistently recorded. Emergence from PTA was assessed prospectively by repeated administration of the GOAT or O-Log. The Goat is a 10-item measure that assesses orientation as well as memory for events preceding and following TBI and a score of ≥76 at two consecutive administrations is consistent with emergence from PTA.²⁵ The Orientation log is a brief measure used to assess orientation, which includes questions related to place, time, and situational factors and a score of ≥25 on two consecutive occasions over 72 hours is consistent with emergence from PTA.²⁶ Using PTA as a measure of TBI severity is common in TBI research.²⁷ and is a frequent practice in the analysis of TBIMS given the robustness of this variable in this large multi-site dataset.^27–29 There is also substantial evidence that PTA is predictive of functional outcomes post-TBI.^29–31 While cut-offs scores have been suggested by research showing PTA greater than seven days is associated with worse outcomes,³² research has shown that prognostic accuracy is best achieved using PTA as a continuous variable.³³

Depression and Anxiety.

The TBI-QOL is as a patient-reported outcome assessment of the domains of physical, emotional, cognitive, and social health.³⁴ For the present study, two short-form subscales measuring symptoms of depression (10 items) and anxiety (10 items) were administered at each time point. The TBI-QOL measures affective aspects of depression and anxiety while omitting somatic and cognitive symptoms that are common to TBI to reduce risk of confounding TBI symptoms with mood. Participants rated each item on a 5-point Likert-like scale ranging from 1 (never) to 5 (always). All TBI-QOL scores were transformed to a T score, with a mean of 50 (SD = 10). Participants score on the TBI-QOL reflect their standing in the general population.³⁴ In this study, T scores exceeding 60 were considered indication of mood disorder.

Cognitive Functioning.

The Brief Test of Adult Cognition by Telephone (BTACT) is a short cognitive battery is administered in 20 minutes or less.³⁵ The test, which is administered by telephone or in-person, consists of five core subtests assessing episodic verbal memory, working memory span, verbal fluency, inductive reasoning, and speed of processing. The test yields a total composite score and two domain scores for Memory and Executive Functioning (EF). BTACT scores are standardized relative to the general population on age (by decade), gender, and education (less than bachelors, bachelor’s degree or higher).³⁶

Data Analyses

All statistical analyses were performed using R studio. Descriptive statistics were used to characterize the sample demographics and injury-characteristics. Two multiple linear regression models with the BTACT Composite z-score as the dependent variable were used to examine the relationships between injury severity (duration of PTA), TBI-QOL Depression and TBI-QOL Anxiety. An interaction variable was calculated to examine how injury severity moderated the relationship between mood symptoms and cognition. TBI-QOL scores were converted to z-scores to aid in interpretation when comparing to BTACT scores, which are calculated as z-scores. Demographic variables were not included in the model since BTACT z-scores are corrected for age, sex, and education. The assumptions of a multiple linear regression model were met. There were no concerns with multicollinearity among the variables or homoscedasticity of errors. No outliers were removed for these analyses and cases with missing PTA data were excluded from the analyses. Post-hoc analyses consisted of calculating correlation coefficients to examine the relationships between depression, anxiety, verbal memory, and executive functions and size of these correlations were compared the using Steigler’s z test for comparing dependent elements of a correlation matrix.

RESULTS

The sample was primarily male, White, and married,. Detailed characteristics are presented in Table 1. Indicators of TBI severity are presented in Table 2. Average BTACT z-scores and TBI-QOL T-scores are presented in Table 3. Of those with TBI-QOL Depression scores, 72 (14.45%) had T-scores at or above one standard deviation from the mean (T≥60). Of those with TBI-QOL Anxiety scores, 82 (16.47%) had T-scores at or above one standard deviation above the mean.

Table 1.

Participant Characteristics

	Mean (N=498)	SD
Age	44.04	19.53
Sex (Male) n, %	365	73.29%
Years of Education	13.77	2.67
Race n, %
White/Caucasian	374	75.10%
Black/African American	74	14.85%
Native American	16	3.21%
Asian/Pacific Islander	13	2.61%
Other	21	4.21%
Ethnicity n, %
Hispanic	70	14.05%
Non-Hispanic	428	85.94%
Marital Status n, %
Married/Significant other	219	43.98%
Single	179	35.94%
Divorced	61	12.25%
Separated	21	4.22%
Widowed	18	3.61%
Prior psychiatric hospitalization	32	6.43%
Prior suicide attempt	27	5.42%
Prior mental health treatment	123	24.70%

Table 2.

Summary of TBI severity among indicators.

	Mean	SD	Min	Max	Median
Days in PTA, m, sd	20.37	19.46	0	133	17
0–24 hours, n %	73	14.66%
1–7 days, n %	76	15.26%
>7 days, n %	349	70.08%
Days to follow commands, m, sd	5.33	8.89	0.5	74	1
Days spent in rehab	22.77	18.31	2	202	17

Table 3.

Cognitive and Affective Outcomes at Year 1.

	Mean	SD
TBI-QOL Depression T-Score	48.53	10.23
TBI-QOL Anxiety T-Score	48.89	10.94
BTACT Composite z-score	−.70	1.16
BTACT Memory z-score	−.52	1.03
BTACT Executive Function z-score	−.61	1.17

Results from multiple regression models using BTACT Composite score as the dependent variable are presented in Table 4. More days in PTA (β=−.18, p<.01) and higher TBI-QOL Depression (β=−.17, p<.01) were significantly associated with worse cognitive performance on the BTACT. There was no significant interaction between PTA and TBI-QOL depression. Similarly, more days in PTA (β=−.20, p<.01) and higher TBI-QOL Anxiety (β=−.20, p<.01) were significantly associated with worse cognitive performance on the BTACT and there was no significant interaction between PTA and TBI-QOL Anxiety.

Table 4.

Associations between BTACT composite z-score, injury severity, and mood symptoms.

	B	Standard Error	β	t	p	95%CI
						LB	UB
Depression
(Constant)	−.50	.07		−6.87	<.01	−.69	−.33
PTA	−.01	<.01	−.18	−4.62	<.01	−.01	−.01
TBI-QOL  Depression	−.27	.07	−.17	−3.78	<.01	−.30	−.10
PTA*Depression	<.01	<.01	.03	.51	.61	−.01	.01
Anxiety
(Constant)	−.47	.07		−6.53	<.01	−.67	−.30
PTA	−.01	<.01	−.20	−4.64	<.01	−.01	−.01
TBI-QOL Anxiety	−.21	.06	−.20	−3.29	<.01	−.22	−.04
PTA*Anxiety	<.01	<.01	.05	.85	.39	<−.01	<.01

Note. PTA=post-traumatic amnesia; This table represents two, independent multiple linear regression models using BTACT composite z score as the dependent variable. The first model includes independent variables PTA, TBI-QOL Depression, and PTA*TBI-QOL Depression (F(3,494)=15.61, p<.001, Adj. R² =.08). The second model includes PTA, TBI-QOL Anxiety, and PTA*TBI-QOL Anxiety (F(3,494)=12.45, p<.001, Adj. R² =.06).

An exploratory subgroup analysis was also conducted to examine the relationships between depression, anxiety, and the BTACT factor scores. Depression was statistically significantly correlated with BTACT Memory (r(496)=−.13, p<.01), BTACT EF (r(496)=−.21, p<.01), and BTACT Composite scores (r(496)=−.21, p<.01) and there was no significant difference in size of the correlation coefficients between the two BTACT domains (EF vs. Memory; z(498)=−1.29 p=.10). Anxiety was statistically significantly correlated with BTACT EF (r(496)=−.17, p<.01), BTACT Memory subscale (r(496)=−.11, p=.01), and BTACT Composite scores (r(496)=−.17, p<.01). There was no statistically significant difference in size of the correlation coefficients between the two BTACT domains (EF vs. Memory; z(498)=−.96, p=.17). Results also showed that the size of the correlation coefficients between TBI-QOL Depression and BTACT Composite and the TBI-QOL Anxiety and BTACT Composite were not significantly different (z(498) = .65, p = .26). Similarly, there were also no statistically significant differences when comparing the size of correlations between anxiety and depression and the BTACT Memory subscale (z(498) = −.32, p = .37) and BTACT EF subscale (z(498) = −.65, p = .26). Taken together, these findings show anxiety and depression had similar magnitude associations with the BTACT Total composite score, and BTACT factor scores of memory and executive functioning.

DISCUSSION

In this study, we examined the associations between depression, anxiety, and cognition one year after mild complicated to severe TBI. Findings indicated both depression and anxiety have small negative associations with cognitive performance in the first year after mild complicated to severe TBI. This study addressed important limitations in the existing body of literature by presenting data from a large cohort that included a large proportion of people with severe TBI. ²⁴ It also addressed limitations of common diagnostic measures of depression and anxiety with the use of unidimensional, item-response theory based measures that do not include cognitive or somatic symptoms that may be confounded by TBI sequelae.³⁴ As such, these findings provide strong evidence that depression and anxiety are associated with worse cognitive functioning across the spectrum of TBI, including among those with more severe injuries.^3,37,38

The rates of depression and anxiety symptoms in this sample were notable, with at least 15% of the sample reporting levels of depression and/or anxiety symptoms one standard deviation above the mean, which is often used as a marker of clinical significance. This rate is consistent with the findings of Scholten and colleagues (2016) based on a pooled estimate across 34 studies.³⁹ It is slightly lower than other reports focusing on the first year after injury,^8,9 which may reflect differences in assessment instruments. There was a significant correlation between the depression and anxiety scores and 70% of those with elevated symptoms of depression also reported elevated scores on anxiety, consistent with previous findings (Bombardier, et al, 2010 [60%]; Hart, et al, 2016 [72%]).^8,10 It is noteworthy that the rates of depression and anxiety are slightly lower than similar studies possibly due to the scales used to assess mood in this study (i.e. TBI-QOL) deliberately omitting cognitive and somatic symptoms of depression and anxiety that have overlap with TBI sequelae. As such, use of TBI-QOL measures for the assessment of mood disorders may increase specificity by reducing false positive findings caused by confounds from TBI-related symptoms (e.g., neurocognitive impairment, sleep disruption).⁴⁰

Post-hoc analyses showed that anxiety and depression were both associated, to a similar magnitude, with a total composite of cognitive performance, and with sub-composites of memory and executive functioning. Although the effect sizes (Pearson’s correlations) are small, these results suggest that symptoms of psychological distress are associated with poorer global cognitive performance after TBI. These results are consistent with findings from other studies associating psychological symptoms with cognitive disorder.^3,41–43 In addition, review of regression effect sizes (standardized betas) indicate these effects were small; for every increase in 1 standard deviation on the measures of psychological symptoms, cognitive performance z-scores declined by −.12 for anxiety and −.18 for depression.

Though effect sizes were small, the current findings may still be clinically significant. Successful treatment of depression and/or anxiety may have a beneficial impact on cognition. Fann and colleagues demonstrated improved cognitive performance among people with TBI and major depression who were treated with sertraline.⁴⁴ In a non-randomized, run-in sertraline trial among people with mild TBI, psychomotor speed, verbal memory, and visual memory improved significantly relative to baseline after 8 weeks. In a double-blind, randomized controlled trial in complicated mild to severe TBI, people treated with sertraline demonstrated improved information processing speed relative to placebo controls after 12 weeks.⁴⁴ Treatment of depression in non-TBI samples has also been associated with improved cognitive performance, particularly in memory, verbal fluency, and psychomotor speed.⁴⁵ While improvement in depression and/or anxiety may not result in a full recovery of cognitive abilities, the benefits would likely extend beyond cognitive performance to increase improved functioning overall and warrants further research.¹⁷

Contrary to expectations, there was no evidence that injury severity modified the relationship between depression or anxiety symptoms and cognition. In a recently published meta-analysis, results suggested that an inverse relationship between cognition and depression/anxiety was evident among those with mild TBI but was mostly unsupported among those with more severe TBI.²⁴ It was hypothesized that the cognitive impairment attributed to moderate-severe TBI masked the effects of mood on cognition and limited participant ability to accurately reflect and report on mood.²⁴ The current findings dispute this hypothesis and suggests mood symptoms must be considered and treated among those with more severe TBI injuries.

The use of PTA as a continuous variable to evaluate injury severity following TBI has been validated in the literature.^28–30,32 However, the literature base examining the relationships between cognition, depression or anxiety, and injury severity have used a range of methods to define injury severity including duration of loss of consciousness, PTA cut-offs,³ diagnostic criteria,⁴¹ and GCS scores.⁴⁶ As such, differences in findings between our study and others may be attributed to differences in these methods.

This study addresses some of the important weaknesses of the existing literature base by using a sample that is more than twice as large as the next largest study and is representative of people who undergo inpatient rehabilitation in the United States.⁴⁷ Our sample includes a large proportion of people with severe TBI. We utilized unidimensional, item-response theory-based measures of depression and anxiety that do not include cognitive or somatic items.³⁴ Finally, testing was performed purely for research and not used for clinical or medical-legal purposes reducing risk of symptom validity problems.⁴⁸

This study has some notable limitations. Due to the cross-sectional nature of the design, we cannot speculate about the temporal and causal direction of the relationships between these variables. In addition, lack of a control group and omission of confounds that can also contribute to mood and cognition (e.g., pain, pre-existing conditions, medication effects, etc.) limit conclusions regarding causality. Furthermore, data were collected at approximately one year after injury; therefore, we do not know whether these relationships are similar or different at various time points post TBI. Finally, the sample consists of individuals with mild complicated to severe TBI who were admitted to inpatient rehabilitation in the United States and are not representative of all TBI patients. As such, we can only speak to observations made in this nationally representative cohort.

The TBI-QOL scales are excellent measures of depression and anxiety severity, and T-scores are used to indicate probable clinically elevated depression and anxiety symptoms, but to date do not have established cutoff scores to indicate, for example, a diagnosis of Major Depressive Disorder or Generalized Anxiety Disorder. Consequently, these results cannot be used to determine the extent to which cognition is related to diagnosable mood and anxiety disorders following TBI. Lastly, the study was based on a relatively large sample and the analyses focused on statistical, not clinical significance.

While the BTACT has been validated as a screening measure for cognition in TBI, ^49,50 it is not a substitute for full neuropsychological battery and thus the data derived from it must be interpreted with caution. That said, this study presented a unique opportunity to determine whether the effects of depression and anxiety are robust enough to influence performance on telehealth delivered cognitive testing protocols and the findings further bolster the hypothesis that depression and anxiety are important variables to consider in all forms of cognitive testing post TBI.

Conclusions

Findings from this study add to the body of literature indicating anxiety and depression are negatively associated with global cognitive functioning following mild complicated to severe TBI. While most of the literature emphasizes depression, findings from this study suggest future research should examine anxiety symptoms as well depression symptoms, as they are highly comorbid and are similarly associated with global cognitive difficulties. It extends the research by demonstrating the relationship between psychological symptoms and cognition is similar regardless of injury severity (within mild complicated to severe TBI injuries) and points to the importance of addressing psychological symptoms in the neurorehabilitation of TBI. That said, longitudinal research is needed to clarify the temporal and causal relationships between depression, anxiety, and cognitive functioning as well as effects of depression/anxiety treatment in this population.