Impact of Sociodemographic, Premorbid, and Injury-Related Factors on Patient-Reported Outcome Trajectories after Traumatic Brain Injury (TBI)

Abstract

Traumatic brain injury (TBI) remains one of the leading causes of death and disability worldwide. To better understand its impact on various outcome domains, this study pursues the following: (1) longitudinal outcome assessments at three, six, and twelve months post-injury; (2) an evaluation of sociodemographic, premorbid, and injury-related factors, and functional recovery contributing to worsening or improving outcomes after TBI. Using patient-reported outcome measures, recuperation trends after TBI were identified by applying Multivariate Latent Class Mixed Models (MLCMM). Instruments were grouped into TBI-specific and generic health-related quality of life (HRQoL; QOLIBRI-OS, SF-12v2), and psychological and post-concussion symptoms (GAD-7, PHQ-9, PCL-5, RPQ). Multinomial logistic regressions were carried out to identify contributing factors. For both outcome sets, the four-class solution provided the best match between goodness of fit indices and meaningful clinical interpretability. Both models revealed similar trajectory classes: stable good health status (HRQoL: n = 1944; symptoms: n = 1963), persistent health impairments (HRQoL: n = 442; symptoms: n = 179), improving health status (HRQoL: n = 83; symptoms: n = 243), and deteriorating health status (HRQoL: n = 86; symptoms: n = 170). Compared to individuals with stable good health status, the other groups were more likely to have a lower functional recovery status at three months after TBI (i.e., the GOSE), psychological problems, and a lower educational attainment. Outcome trajectories after TBI show clearly distinguishable patterns which are reproducible across different measures. Individuals characterized by persistent health impairments and deterioration require special attention and long-term clinical monitoring and therapy.

Keywords: Multivariate Latent Class Mixed Models (MLCMM); health-related quality of life (HRQoL); long-term outcomes; mental health; recuperation trajectories; traumatic brain injury.

Figure A1

Generic and disease-specific HRQoL trajectories (QOLIBRI-OS, SF-12 v2 MCS/PCS) for the best solutions of model 1 (random intercept, fixed slope) for subgroups with different TBI severity. Solid, colored lines indicate predicted trajectories, dashed lines the observed values. Shaded areas around lines indicate 95% confidence intervals for predicted values. Gray lines mark cut-off for unfavorable values for the QOLIBRI-OS (<53) as well as mean (=50) and standard deviation (+/−10) for the SF-12v2 MCS/PCS. Gray area marks values below the cut-off for unfavorable levels of quality of life (QOLIBRI-OS < 52, SF-12v2 MCS/PCS < 50).

Figure A2

Symptom trajectories (GAD-7, PHQ-9, PCL-5 and RPQ) for the best solutions of model 1 (random intercept, fixed slope) for subgroups with different TBI severity. Solid, colored lines indicate predicted trajectories, dashed lines the observed values. Shaded areas around lines indicate 95% confidence intervals for predicted values. Gray lines mark cut-off values for categorical classification of outcomes (GAD-7: 5 = mild, 10 = moderate, 15 = severe; PHQ-9: 5 = mild, 10 = moderate, 15 = severe; RPQ: 13 = mild, 25 = moderate, 33 = severe; PCL-5 ≥ 31: PTBS screening). Gray areas mark values above the cut-offs for clinical relevance (GAD-7 ≥ 10; PHQ-9 ≥ 10; PCL-5 ≥ 31; RPQ ≥ 13).

Figure A3

Association of HRQoL and symptom classes identified by Multivariate Latent Class Mixture Modelling (χ2(9) = 1152.96, p < 0.001; Cramer’s v = 0.39 [small: 0.1, medium: 0.2, large: 0.6]); percentages < 3% not displayed.

Figure 1

Flow chart of sample selection.

Figure 2

Generic and disease-specific HRQoL trajectories (QOLIBRI-OS, SF-12 v2 MCS/PCS) for the four-class solution. Solid, colored lines indicate predicted trajectories, dashed lines the observed values. Shaded areas around lines indicate 95% confidence intervals for predicted values. Gray lines mark cut-off for unfavorable values for the QOLIBRI-OS (<53) as well as mean (=50) and standard deviation (+/−10) for the SF-12v2 MCS/PCS. Gray area marks values below the cut-off for unfavorable levels of quality of life (QOLIBRI-OS < 52, SF-12v2 MCS/PCS < 50).

Figure 3

Symptom trajectories (GAD-7, PHQ-9, PCL-5 and RPQ) for the four-class solution. Solid, colored lines indicate predicted trajectories, dashed lines the observed values. Shaded areas around lines indicate 95% confidence intervals for predicted values. Gray lines mark cut-off values for categorical classification of outcomes (GAD-7: 5 = mild, 10 = moderate, 15 = severe; PHQ-9: 5 = mild, 10 = moderate, 15 = severe; RPQ: 13 = mild, 25 = moderate, 33 = severe; PCL-5 ≥ 31: PTBS screening). Gray areas mark values above the cut-offs for clinical relevance (GAD-7 ≥ 10; PHQ-9 ≥ 10; PCL-5 ≥ 31; RPQ ≥ 13).

Figure 4

GOSE-levels at three, six, and twelve months after TBI for each generic and disease-specific HRQoL trajectory class. Decreasing sample sizes due to death of participants. Percentages < 3% not displayed.

Figure 5

GOSE-levels at three, six, and twelve months after TBI for each symptom trajectory class. Decreasing sample sizes due to death of participants. Percentages < 3% not displayed.

Figure 6

HRQoL trajectories (QOLIBRI-OS, SF-12 MCS, SF-12 PCS) for the four-class solution of model 1 (random intercept, fixed slope). Stable good health status class is used as the reference group. Odds ratios and 95%-confidence intervals depicted. Reference group for each categorical variable listed after vertical line. SD = severe disability, MD = moderate disability, GR = good recovery. Values below 1 indicate lower probability of belonging to the non-reference group (i.e., persistent health impairment, deteriorating health, or improving health) compared to the stable good health group.

Figure 7

Symptom trajectories (PHQ-9, GAD-7, PCL-5, RPQ) for the four-class solution of model 1 (random intercept, fixed slope). Low stable health status class is used as the reference group. Odds ratios and 95%-confidence intervals depicted. Reference group for each categorical variable listed after vertical line. SD = severe disability, MD = moderate disability, GR = good recovery, ISS = Injury Severity Score, AIS = Brain Injury Score. Values below 1 indicate lower probability of belonging to the non-reference group (i.e., persistent health impairment, deteriorating health status, or improving health status) compared to the stable good health status group.