Many conditions result in reduced vigilance and unconsciousness, including structural brain diseases such as intracranial hemorrhage or traumatic brain injury (TBI), and also diffuse neuronal dysfunction (e.g., due to non-convulsive status epilepticus, hypoxic encephalopathy after cardiac arrest or intoxications) and psychogenic unresponsiveness. While transient in some, loss of consciousness often requires management in an intensive care unit (ICU).

Disorders of consciousness are classically divided into coma, unresponsive wakefulness syndrome (UWS, previously referred to as vegetative state) and minimally conscious state (MCS), depending on the patient’s response to internal and external stimuli (1). When assessing coma, the level of consciousness, brainstem responses, motor responses and breathing patterns should be evaluated. Patients in coma have absence of arousal and awareness of one’s self and the environment lasting for more than 6 h by definition (2); thus they do not respond to internal (e.g., pain from injury, bladder fullness, thirst) or external stimuli (e.g., pain, touch, speech). Patients in UWS have a preserved arousal level, demonstrate sleep–wake cycles and 37–43% may show signs of awareness (3, 4). In MCS, cognitively mediated behavior occurs inconsistently, but is reproducible or sustained long enough to be differentiated from reflexive behavior (5). The level of consciousness is best expressed by the Glasgow Coma Scale (GCS) and the Full Outline of Unresponsiveness (FOUR) scale, which are the most standardized tools to assess the level of consciousnesses (6). The FOUR scale provides greater neurologic detail than the GCS and also incorporates brainstem function. It is especially useful in intubated patients (7).

For the diagnostic workup in those patients with prolonged and persistent unconsciousness, guidelines such as the one published 2020 by the European Academy of Neurology recommend a multimodal diagnostic approach including a standardized clinical evaluation, electroencephalography (EEG), and both structural and functional imaging (8). However, advanced diagnostic testing such as magnetic resonance imaging (MRI) may be limited by the patient’s poor general condition and by supportive devices such as ventilators. Thus, the bedside clinical examination in the unconscious patient is of great importance, allowing both a narrowing of the differential diagnosis, guided decisions on additional diagnostic testing and on prognosis (9). The neurologic exam is key in this diagnostic process [see, e.g., Christofi (7) for a review]. Here we will focus on the examination of the acutely unconscious patients though where relevant, we will discuss applications to the examination and evaluation of UVS and MCS. Amongst bedside examinations available, eye movement testing allows for an assessment of the integrity of both supratentorial (higher cortical) and infratentorial (brainstem, cerebellar) circuits. For example, conjugate eye movements rely on a functional network of neurons located in the brainstem (pons, midbrain) adjacent to the ascending arousal system that contributes to consciousness (10, 11). Ocular motor and vestibular testing in the unconscious patient therefore have a major role and can be considered as «a window to the function of the brain».

Here we review the range of eye movement abnormalities seen in the unconscious patient, with a focus on bedside and quantitative ocular motor and vestibular testing and the interpretation of findings and provide a practical approach how to work-up these patients. Furthermore, we will also discuss the prognostic value of eye movement abnormalities in the unconscious patient. Blinking and pupillary responses are not covered in this review, but may be found elsewhere (7, 9, 12, 13).

For this critical review a literature search (MEDLINE and Embase) was performed. The search strategy was designed by a clinical investigator with relevant domain expertise in neurology (AAT). We focused on English-language articles, using the following strategies with the following components: (1) defining the clinical state (i.e., coma, reduced vigilance), (2) examination techniques including bedside ocular motor / vestibular assessments or quantitative testing, and (3) defining the spectrum of eye movements studied. We also manually searched reference lists from eligible articles. We did not seek to identify research abstracts from meeting proceedings or unpublished studies. The literature search was last updated on September 9th 2025 and initially identified 4,241 unique citations. Based on a title and abstract review performed by AAT, a total of 110 papers were identified for full-text review and eventually 54 papers were considered suitable. Manually searching the reference list of selected papers resulted in another 57 papers that were reviewed. These papers provided the basis to determine currently reported practices, and to identify the range of ocular motor and vestibular findings seen in the unconscious patient. Thus, this paper reflects an extensive review of the current knowledge on eye movement testing in the unconscious patient and also identifies limitations and potential strategies to advance in the future.

In the examination of the unconscious patient the structured and detailed neurologic exam plays a key role as outlined in the seminal works of Fisher (12) and of Plum and Posner (9); thus a neurology consult is often ordered (14). The neurologic assessment of the unconscious patient covers a broad range, including the level of responsiveness by applying external stimuli, pupillary responses, spontaneous body movements, eye movements (both spontaneous and reflexive), corneal responses, muscle tone, autonomic functions, fundoscopy and otoscopy (7, 12, 15, 16) (see Table 1). Intubation, sedation and muscle relaxation can limit the neurologic examination, impairing the assessment of higher cortical functions including cognition, speech, language, motor function and the sensory system. Thus, the assessment of brainstem function [see, e.g., Benghanem et al. (17)] with a focus on eye movement testing usually offers the best opportunity to assess brainstem-cerebellar networks including the vestibulo-ocular reflex (VOR) and the gaze-holding circuits (9, 12, 18, 19).

Quantitative ocular motor and vestibular testing offers the advantage of collecting standardized and reproducible eye movement responses and thus facilitating diagnosis, and the assessment of disease progression and their response to treatment (see also (11) for a review of the diagnostic and prognostic value of eye movement recordings in the unconscious patient). However, recording eye movements in the unresponsive patient in the ICU setting may be challenging or even not feasible (89). Both mobile equipment and better recording paradigms adapted to ICU setting are needed. Furthermore, with the ophthalmoscope one can detect pathological intrusive eye movements as small as a few tenths of a degree, emphasizing the importance of the clinical bedside examination.

For testing the integrity of the VOR both caloric irrigation (18) and (recently) galvanic stimulation have been used. Caloric irrigation applies a low-frequency stimulus to the vestibular organ and thus may elicit different responses compared to those from high-frequency head rotations. In deeply comatose patients combining head rotation with a caloric irrigation may be needed to elicit a vestibular response (12). Caloric irrigation also offers another way to stimulate the vestibular organs when head turns (as used for testing the VOR at the bedside) are contraindicated, e.g., in patients with spinal trauma.

The integrity of the VOR has been quantified in both comatose patients and patients in UWS (48). Both comatose patients and those in UWS showed an initial oppositely directed nearly compensatory eye movement when a fast head turn was applied. In both patient groups the head was then kept in a position such that the eyes were brought to an eccentric position. With this paradigm eccentric gaze holding was impaired in both groups; the eyes drifted back toward the center position indicating the velocity-to-position integrator was impaired, which was likely due to loss of cerebellar Purkinje cells. Sinusoidal oscillations of the head produced compensatory slow phases in both patient groups. Intermittent quick phases, however, were seen only in patients with UWS and were not sufficient to keep the eyes close to the straight-ahead position in the orbit.

Developing predictive measures for recovery from coma and UWS is an unmet need and of great importance both from health, ethical and economic perspectives. Various potential predictors for poor outcome in comatose survivors after cardiac arrest have been systematically reviewed recently. However, no assessment of the potential role of eye movement analysis was provided and no use of machine learning / artificial intelligence was made in these studies (107). Thus, few studies have assessed the prognostic value of ocular motor and vestibular assessments in unconscious patients in a prospective way. This, however, is a prerequisite for the design of longitudinal studies in these patient populations. Feasibility studies have suggested that quantitative eye tracking in combination with machine-learning algorithms may serve as a non-invasive biomarker for recovery after TBI (106). Likewise, in a prospective study following up the natural history of patients with non-traumatic coma, the presence of orienting eye movements and of the VOR at the bedside were among the features predicting for reaching independent existence (108), while those patients with absent VOR or absent ocular following had poor outcomes. In comatose patients after cardiac arrest, assessing eye movements at the bedside could be used to help predict outcome (109). A lack of a VOR response or a lack of spontaneous orienting or conjugate roving eye movements within 2 weeks were poor prognostic markers.

The potential value of eye movements in predicting recovery was investigated in patients with UWS as well. In one study, EEG features (gamma power modulation) in association with what were described as “conjugate slow ballistic eye movements” have been quantified to develop a potential method to detect signs of cortical responsiveness (110). In this study, changes in the time course of gamma power in relation to conjugate eye movements were observed only in less severely affected UWS patients. However, the question of whether this pattern is supportive of recovery in UWS patients remains to be addressed.

The lack of standardized, quantitative ocular motor and vestibular testing in the unconscious patient is a major deficiency in the research of unconscious patients. Likewise, no clear and consensus-driven semiological classification for various eye movement abnormalities in the unconscious patient is available, making a uniform description of eye movement patterns seem challenging.

Furthermore, the spectrum of underlying causes of unconsciousness is broad and strongly affects the prognosis. Incomplete descriptions of the extent of brain damage both structurally and functionally and the use of sedative and other drugs that affect brain function further add variability when assessing patient cohorts for outcome and features with predictive power. For bedside assessments of spontaneous and reflexive eye movements, only moderate inter-rater agreement was noted (with Kappa-values between 0.46 and 0.49) (111), emphasizing the need for appropriate training in assessing eye movements in the comatose patient and also for quantitative measurements. The availability of mobile VOG devices with virtual reality (VR) environments allows the application of a broad range of visual and vestibular stimuli in the ICU patient in a standardized way and thus presents a promising technological development. However, the difficulty of accurate calibration remains a challenge to obtaining reliable quantitative eye movement recordings in the unconscious patient. Finally artificial intelligence (AI) offers a great opportunity to analyze large, continuous data streams from sensors including electrocardiography, EEG and others as collected in ICU patients in a more meaningful and efficient way. Such an example is training a machine learning algorithm to detect eye movements from continuous EEG traces in comatose patients after cardiac arrest for optimizing prognostication. The feasibility of this approach was demonstrated, with high accuracy of automated eye movement detection from electro-oculography (EOG) in post cardiac-arrest patients in one study (112). EEG has the advantage of its broad availability in ICUs, standardized recording setups and data collection. However, the value of such novel approaches must be evaluated in the clinical setting. Big data analytics will facilitate the identification of clusters, thus, future multicenter studies with large sample sizes will allow a better description of patterns of eye movement abnormalities in unconscious patients Furthermore, our understand of disease mechanisms causing eye movement abnormalities has grown substantially in the past decades and new testing paradigms (such as positional testing and quantitative assessment of ocular torsion) have been added and allow for a more detailed interpretation of findings.

Eye movement testing in the unresponsive patient, especially caloric and “dolls head” VOR responses to head rotations, has been a bedrock of the neurological approach to the comatose patient for more than half a century. It offers targeted bedside testing of several brainstem reflexes and the evaluation of the integrity of neuronal networks connecting the cortex, brainstem and the cerebellum. Modern ways to record eye movements at the bedside with video-oculographic techniques and virtual reality, refinements of the evaluation and better understanding of vestibular responses, and new testing techniques such as galvanic stimulation, detection of torsional eye movements, and vestibular evoked potentials, all provide new ways to probe the function of the brain and correlate findings with cause, treatment and outcome. Analysis of patterns of eye movements plays a growing role in determining the prognosis of unconscious patients. Specifically, the presence of reflexive eye movements such as the VOR or visual following has been linked to better outcome and therefore should be included when assessing outcome. Quantitative, non-invasive continuous eye tracking, combined with AI-based analysis of large, multimodal datasets collected in a modern ICU may be used as a tool for prognosis in the future.