The research, published in Frontiers in Neurology, argues that a favorable GOSE score should not be mistaken for a favorable life. The GOSE has long been the cornerstone of outcome measurement in brain injury research. Administered as a structured interview, it assigns survivors to one of eight levels ranging from death and vegetative state through severe and moderate disability to upper good recovery.

The National Institute of Neurological Disorders and Stroke has recognized it as a core outcome metric precisely because it emphasizes daily functioning, is straightforward to interpret, and can be delivered in person, by phone, or by mail. But the scale has known weaknesses: patients can be misclassified, there is no true gold-standard method of administration, and borderline cases are difficult to assign. Most importantly for the new study, GOSE captures disability and independence, not the lived experience of recovery as patients perceive it.

To probe that gap, researchers at the Edward Via College of Osteopathic Medicine, Virginia Tech-Carilion School of Medicine, High Point University, and the Carilion Clinic Brain Injury Center paired the GOSE with the Quality of Life after Brain Injury (QOLIBRI) questionnaire, a disease-specific, patient-reported instrument covering six domains: Emotions, Feelings, Independence, Physical problems, Social relationships, and Thinking abilities. Each domain probes a different facet of recovery, from depression, anxiety, and agitation to memory and concentration, from headaches, dizziness, and fatigue to interpersonal relationships and community participation. The team enrolled 138 adults aged 18 to 65 who were at least one year post-injury, with eligibility defined by a best Glasgow Coma Scale score of 3 to 12 within the first 24 hours, loss of consciousness lasting more than 30 minutes, or post-traumatic amnesia exceeding 24 hours.

The cohort skewed heavily toward the severe end of the spectrum: 112 of the 138 participants had severe TBI, while 26 had moderate injuries. Motor vehicle collisions were the most common mechanism of injury, accounting for 72 cases, followed by falls and blunt traumatic injury with 26 each. The average age was 35 years, and 110 of the participants were male, reflecting the skewed incidence of TBI in men.

A single trained investigator administered the GOSE face-to-face to eliminate interrater variability, and all participants completed the QOLIBRI, with only one person requiring a staff member to read items aloud because of physical limitations. The mean GOSE score was 6.3, with a median of 7, indicating that most participants had reached the upper moderate disability or lower good recovery range. Statistically, the challenge was to model the relationship between a single functional score and six correlated quality-of-life domains without inflating error.

The researchers fit a mixed-effects model by maximum likelihood, treating participant identity as a random effect because each of the 138 participants contributed six domain scores. GOSE score, age, sex, injury severity, QOLIBRI domain, and the interaction between GOSE and each domain entered as fixed effects. Because GOSE is ordinal, the team also tested an ordinal model, but an ANOVA comparison showed no statistical benefit to the ordinal transformation (AIC: linear = 7361.9 versus ordinal = 7368.7; chi-squared = 1.19, p = 0.88), and the ordinal version would have added many parameters to a modest sample, raising overfitting concerns.

They therefore treated GOSE as a continuous variable, with cautious interpretation. The headline result is striking in its negative form: GOSE score alone was not significantly associated with quality of life. In the mixed-effects model, the main effect of GOSE carried an estimate of −0.79 with a 95% confidence interval spanning −3.2 to 1.5 and a p-value of 0.57.

Age, sex, and injury severity also failed to reach significance. The action lay in the interactions. The estimated marginal trend of GOSE within the Independence domain was 10.4 (95% CI 8.1 to 12.7, p < 0.001), meaning each step up the functional scale corresponded to a substantial gain in perceived autonomy.

The Thinking abilities domain showed a trend of 5.7 (95% CI 3.3 to 8.0, p < 0.001), and the Physical problems domain, after reverse coding so that higher scores mean fewer symptoms, showed a trend of 2.8 (95% CI 0.44 to 5.1, p = 0.019). By contrast, the social and emotional picture was far dimmer. The GOSE trend within the Feelings domain, which captures changes in self-perception and identity, was 2.2 with a p-value of 0.061, a weak and non-significant association.

Within the Social relationships domain, the trend was a negligible 0.74 (95% CI −1.6 to 3.1, p = 0.53), statistically indistinguishable from zero. In other words, two survivors with identical GOSE scores could report radically different levels of social participation and emotional well-being, and the functional scale would have no way of telling them apart. The authors interpret this as evidence that psychosocial recovery follows a trajectory of its own, one that may depend on social support networks, employment opportunities, cultural expectations, and psychiatric comorbidities rather than on functional independence alone.

An unusual statistical detail deserves attention. The participant random effect returned a variance and standard deviation of zero, a so-called singular fit, indicating that after accounting for the fixed effects, the model detected no between-participant variability in quality-of-life scores. To confirm this was not an artifact of the fitting algorithm, the team refit the model with restricted maximum likelihood and with the bound optimization by quadratic approximation, both of which reproduced the singular result.

They also ran a generalized estimating equation (GEE) analysis, which handles repeated within-participant observations without estimating a variance, and found near-identical coefficients and significance patterns, with an estimated within-participant correlation near zero. The residuals were roughly symmetrically centered around zero, with standardized values ranging from −2.9 to 2.7, supporting the model's assumptions of normality and homoscedasticity. The findings align with, and extend, earlier work.

Large prospective studies such as TRACK-TBI have shown that many survivors carry a substantial symptom burden and reduced quality of life years after injury, often irrespective of their functional recovery status. Prior studies linking GOSE to health-related quality of life had generally treated quality of life as a single global construct; the domain-specific approach here reveals that the association is concentrated in independence, cognition, and physical symptoms. The authors also note that while functional outcomes show a clear dose-response relationship with injury severity, persistent post-traumatic symptoms such as headaches, chronic pain, medication side effects, and psychosomatic interactions do not always follow severity in the same way, which may explain why the physical domain, though significant, showed a weaker trend than independence or thinking.

Several limitations frame the conclusions. The cross-sectional design cannot establish whether improvements in functional status cause gains in quality of life or whether the two evolve independently over time; longitudinal studies tracking both measures at multiple time points are needed. Self-reported quality of life is vulnerable to recall bias, mood, cultural attitudes toward illness, and social desirability.

Excluding individuals with GOSE scores of 1 or 2, appropriate for self-report measures, restricts generalizability to the highest-severity cases and may bias findings toward higher-functioning survivors. The 79.7% male composition warrants cautious interpretation, and the absence of demographic effects may reflect sample-specific factors, including a possible survival effect in which only relatively higher-functioning older survivors remained in the cohort. Notably, injury severity itself did not predict quality of life, consistent with prior evidence that initial clinical severity indices such as the Glasgow Coma Scale have limited predictive value for long-term patient-reported outcomes.

The clinical message is nonetheless clear. A favorable GOSE score does not necessarily equate to a favorable life, and relying solely on functional scales risks overlooking persistent deficits in social participation, emotional well-being, and self-identity. The authors argue for routine incorporation of patient-reported outcomes such as the QOLIBRI into clinical care, so that rehabilitation planning can address the full spectrum of patient needs and recovery can be measured from the patient's perspective as well as the clinician's.

Future work, they suggest, should combine longitudinal designs with psychosocial and biological variables to build prognostic models that capture both functional and meaningful recovery.

Subject of Research: The relationship between functional recovery measured by the GOSE and domain-specific quality of life after moderate-to-severe traumatic brain injury

Article Title: Functional recovery and domain-specific quality of life after moderate-to-severe traumatic brain injury: a mixed-effects analysis

Article References: Badra, R., Scardina, B., Nelson, M., Mishra, V., van Doren, T. P., Martinez, M., & Weppner, J. L. (2026). Functional recovery and domain-specific quality of life after moderate-to-severe traumatic brain injury: a mixed-effects analysis. Frontiers in Neurology, 17, 1871168. https://doi.org/10.3389/fneur.2026.1871168

Image Credits: AI Generated

DOI: 10.3389/fneur.2026.1871168

Keywords: traumatic brain injury, Glasgow Outcome Scale-Extended, QOLIBRI, quality of life, functional recovery, patient-reported outcomes, neurotrauma, rehabilitation, mixed-effects model, cognitive function, social relationships, independence