Triaging head injury and stroke without a CT scanner: a district-hospital playbook

The literature on traumatic brain injury, stroke, and altered sensorium is overwhelmingly written from inside tertiary centres. The CT scanner is assumed. The neurosurgeon is reachable in minutes. The neuro-ICU has a bed. None of this is true at most of the 36,000+ Indian hospitals with emergency services. In district hospitals, sub-divisional hospitals, and CHCs across the country, the casualty medical officer often faces a head-injury or stroke patient with no CT, no MRI, no neurosurgeon on call, and a transfer decision that has to be made in minutes — not hours.

The question that decision actually answers is not "what is the diagnosis?" It is "do I transfer this patient now, and how fast?" The pupil exam is one of the very few bedside tools that survives the no-imaging gap. The trouble is that the way we usually do the exam — penlight + impression — does not.

This is a playbook for the district-hospital MO and the transport team: what the published evidence says about pupillary signs in head injury and stroke, what the penlight exam misses, and what a quantitative bilateral pupil reading can add when there is no scanner to fall back on.

The no-CT reality, briefly

CT is concentrated in tertiary and large urban private hospitals. Most district and sub-divisional hospitals either lack CT entirely or have a single scanner that is not staffed 24×7. The Brain Trauma Foundation, AANS, and Indian neurotrauma society guidelines all assume rapid imaging — they offer little for the casualty MO at 2 a.m. on a rural highway who has just received a road-traffic head injury and a stroke arriving simultaneously, with the nearest CT three hours away.

In that setting, the realistic toolkit is: a focused history, the GCS, a vitals chart, a fundus check, a motor exam, and the pupil exam. Of these, the pupil exam is the one with the highest published prognostic value per second spent — if the measurement is reliable.

The pupil as a triage primitive

The pupillary light reflex traces a long arc through the brainstem: retina → optic nerve → pretectal nucleus → bilateral Edinger-Westphal nuclei → CN III → ciliary ganglion → iris sphincter, with a separate sympathetic loop driving redilation. Two consequences flow from that anatomy. First, the reflex is anatomically near the structures that fail catastrophically in raised intracranial pressure — the dorsal midbrain and the tentorial edge where CN III runs. Second, a new pupillary asymmetry, a sluggish reaction, or a unilateral fixed dilated pupil points at exactly that anatomy, often before other localising signs appear. For a casualty MO without imaging, this is the closest thing to a clinical proxy for early herniation.

The published mortality data make the case sharply. According to PubMed, Brennan, Murray and Teasdale formalised pupillary reactivity into the GCS-Pupils score: combined GCS + pupil reactivity outperforms either alone, with mortality rising from 51 % at GCS 3 to 74 % at GCS-Pupils 1 (DOI, PMID 29631516). Martins et al., in 748 severe-TBI patients, reported an adjusted odds ratio of 11.52 for death with bilateral mydriasis, and 2.65 with anisocoria, versus isocoric reactive pupils (DOI, PMID 19590314). Tien et al. found that 100 % of GCS-3 trauma patients with bilateral fixed dilated pupils died, versus 42 % mortality in those whose pupils remained reactive (DOI, PMID 16508482). The 2024 NINDS TBI Classification Initiative now formally recommends that "pupillary reactivity should be documented in all patients" with TBI, recorded separately from the GCS — explicit acknowledgement that the pupil number is not subordinate to the GCS number, it is independently load-bearing (DOI, PMID 40393504).

The same finding holds in stroke. Kobata et al., in 1,638 WFNS grade V SAH patients, found 3-month favourable outcomes of 4.5 % with bilateral unreactive pupils versus 21.4 % with reactive pupils — a fivefold difference that measurably influenced surgical decision-making (DOI, PMID 36825904). For the district MO triaging a hemorrhagic-stroke arrival, the pupil reading is one of the few prognostic indicators they can collect at the bedside.

The under-triage signal in apparent "mild" injury is just as important. Bossers et al., in 808 patients triaged as mild TBI, found that absence of equal and reactive pupils was an independent risk factor (OR 2.1) for being more severely injured than the triage GCS suggested, and 12.9 % were eventually reclassified upward (DOI, PMID 29032474). For a district MO making transfer decisions on volume, that finding flags exactly the patient who looks fine but is not.

The penlight problem — even more severe in the district setting

The reliability data on the penlight exam come overwhelmingly from tertiary centres with experienced critical-care nurses and consistent training. Couret et al. (2016), in a prospective double-blinded neuro-ICU study, reported 18 % global discordance with quantitative measurement, a 39 % error rate for pupils under 2 mm, and missed 50 % of true anisocoria (DOI, PMID 27072310). Olson et al. measured interrater κ of 0.54 for size and 0.40 for reactivity across 2,329 paired assessments; only 33 % of pupils called "non-reactive" were truly non-reactive (DOI, PMID 26381281). Kerr et al. showed nurses systematically underestimate pupil size, with the error growing as size grows (DOI, PMID 27134226).

If trained tertiary-care nurses miss half of true anisocoria with a penlight, the realistic miss rate at a district hospital — variable lighting, rotating MOs, language differences with the family historian, no neuro consult to fall back on — is almost certainly higher. A 2019 systematic review concluded that automated pupillometry is more precise, more reliable, and detects pupillary change before clinical deterioration is apparent (DOI, PMID 30484008). The earlier-detection signal matters most where the next step — transfer — takes hours.

A practical no-CT triage frame

The decision the district MO is actually making, several times a shift, looks like this. Do I keep this patient and observe? Do I transfer non-urgently? Do I transfer urgently? Do I transfer with intubation and a bag-valve attendant? The pupil reading is one of the inputs that reshape that decision in seconds.

A useful, evidence-anchored frame for the no-CT setting:

• Bilateral fixed dilated pupils in a head-injured patient is a near-mortal finding that does not get better with time at your facility (Tien, Martins, Brennan). It is an immediate-transfer signal, with airway control if available.
• A new, persistent anisocoria of ≥ 1 mm in a patient with depressed sensorium is the single most useful localising sign of evolving mass effect, uncal herniation, or CN III compromise (Martins; AANN guideline). The AANN Clinical Practice Guideline names ≥ 1.0 mm as the threshold for clinically significant new-onset anisocoria. Document, escalate, transfer.
• Sluggish or asymmetric reactivity in an apparent "mild" head injury (intact GCS, walking-talking, vomiting once) flags the under-triage patient (Bossers, OR 2.1). Lower the threshold to image, observe, or transfer.
• Bilaterally reactive pupils in a deep coma is a finding compatible with metabolic, toxic, post-ictal, or hypoxic causes — useful for ruling out primary catastrophic structural injury, especially in a setting where you cannot rule it out with imaging.
• Pinpoint pupils with respiratory depression is the opioid pattern; reactivity preserved on quantitative measurement can support a trial of naloxone in the absence of a tox screen.
• Asymmetric reactivity in a stroke arrival (Marshall et al. demonstrated automated pupillometry detects PLR alterations in acute stroke; Maxin et al. detected acute large-vessel occlusion via smartphone pupillometry — DOI, PMID 37857150) supports rapid transfer to a stroke-ready centre even when the diagnosis cannot be locally confirmed.

None of this replaces imaging. It triages who needs the imaging, in what order, and with what urgency the transfer happens.

The documentation argument

A district-hospital case sheet that reads "PERRL" is not a defensible record — clinically, administratively, or medicolegally. A time-stamped, bilateral, two-page pupil report attached to the file is. It documents that the exam was done, when, by whom, with what objective values for both eyes, and against what reference ranges. If the patient deteriorates en route or after transfer, the receiving team has a baseline; if questions are asked later, there is a record. This is the same shift quantitative blood pressure measurement made over palpation — not a replacement for clinical judgement, but a defensible artefact that backs the judgement up.

A shareable record opens a second door — remote opinion before transfer

The two-page PupiLUX Pro Report is a PDF. That is a quietly important fact for the no-CT setting, because it means the same artefact that documents the exam can also travel — over WhatsApp, over email, over whatever channel the casualty MO already uses to reach colleagues at the referral centre.

The decision the district MO is wrestling with is rarely "what is the diagnosis." It is "do I transfer, how fast, with what stabilisation, and to whom." A short call to a tertiary-centre neurosurgeon or ER consultant is often the most useful single input into that decision — but the call is hard to make compelling without something concrete to put in front of the receiving clinician. "The pupils look sluggish" is hard to act on. A two-page PDF showing 6.2 mm vs 4.1 mm anisocoria, baseline pupil diameters, constriction percentage, latency, and the recording timestamp is a different conversation entirely. Send it ahead, ring the consultant, ask their opinion. If they say "transfer urgently and start mannitol now," the MO has acted on a remote-consultant view backed by an objective bedside measurement, not a one-line WhatsApp impression. If they say "observe locally and re-image at four hours if you can," the MO has documentation to support keeping the patient.

This is not telemedicine in any formal sense — there is no platform, no integration, no liability shift. It is the same informal consult that already happens between district and tertiary clinicians a hundred times a day in India, but with a defensible measurement attached. The receiving consultant interprets, the on-site MO decides, the artefact in the file shows what was shared, when, and to whom. PupiLUX does not interpret; it just produces the document the conversation can be built on.

Note on regulatory framing. PupiLUX produces measurements; clinicians interpret. Sharing the report with a colleague for an opinion is the colleague's clinical judgement on the colleague's licence — exactly as it would be if the conversation were about a BP cuff reading or an ECG strip.

Where PupiLUX fits in a no-CT setting

PupiLUX is a smartphone-based bilateral pupillometer. It runs on a personal iPhone — no dedicated hardware to procure, no consumables, no IT integration — and produces a 7-second bilateral PLR reading with a two-page PupiLUX Pro Report carrying six quantitative parameters per eye. For the casualty MO at a district hospital, that combination is the operationally important one: the device they need is already in their pocket, the workflow is short enough to use on every depressed-sensorium arrival, and the output is a printable PDF that goes into the case file.

PupiLUX is a measurement and screening tool — not a diagnostic device. The report shows the numbers and the reference ranges; the clinician interprets and decides. We deliberately do not generate diagnoses, decision rules, or transfer recommendations. Those are clinical judgements that belong to the MO at the bedside.

Disclaimer. PupiLUX is a measurement tool, not a diagnostic device. For informational and screening purposes only.

Closing

The reliability gap in the bedside pupil exam is widest exactly where the consequences of missing a finding are largest — district and sub-divisional hospitals managing head injury and stroke without imaging. If you run, work in, or transport patients to such a facility and would like to evaluate PupiLUX, write to us at info@pupilux.ai. The fuller clinical write-ups for trauma and neurocritical care live at pupilux.ai/clinical/trauma and pupilux.ai/clinical/neuro-icu; the longer primer is at pupilux.ai/clinical/introduction.

References

All references verified against PubMed on 2026-04-27. DOI links included per PubMed attribution requirements.

1. Brennan PM, Murray GD, Teasdale GM. Simplifying the use of prognostic information in traumatic brain injury. Part 1: The GCS-Pupils score. J Neurosurg. 2018;128(6):1612–1620. DOI · PMID 29631516.
2. Martins ET, et al. Mortality in severe traumatic brain injury: a multivariated analysis of 748 Brazilian patients. J Trauma. 2009;67(1):85–90. DOI · PMID 19590314.
3. Tien HC, et al. Do trauma patients with a GCS of 3 and bilateral fixed and dilated pupils have any chance of survival? J Trauma. 2006;60(2):274–278. DOI · PMID 16508482.
4. Menon DK, et al. Clinical Assessment on Days 1–14 for the Characterization of Traumatic Brain Injury: Recommendations from the 2024 NINDS TBI Classification Initiative. J Neurotrauma. 2025;42(13–14):1038–1055. DOI · PMID 40393504.
5. Kobata H, et al. Significance of Pupillary Findings in Decision Making and Outcomes of WFNS Grade V Subarachnoid Hemorrhage. Neurosurgery. 2023;93(2):309–319. DOI · PMID 36825904.
6. Bossers SM, et al. Discrepancy between the initial assessment of injury severity and post hoc determination of injury severity in patients with apparently mild TBI. Eur J Trauma Emerg Surg. 2017;44(6):889–896. DOI · PMID 29032474.
7. Couret D, et al. Reliability of standard pupillometry practice in neurocritical care. Crit Care. 2016;20:99. DOI · PMID 27072310.
8. Olson DM, et al. Interrater Reliability of Pupillary Assessments. Neurocrit Care. 2016;24(2):251–257. DOI · PMID 26381281.
9. Kerr RG, et al. Underestimation of Pupil Size by Critical Care and Neurosurgical Nurses. Am J Crit Care. 2016;25(3):213–219. DOI · PMID 27134226.
10. Phillips SS, et al. A Systematic Review Assessing the Current State of Automated Pupillometry in the NeuroICU. Neurocrit Care. 2019;31(1):142–161. DOI · PMID 30484008.
11. Marshall M, Deo R, Childs C, Ali A. Feasibility and Variability of Automated Pupillometry Among Stroke Patients and Healthy Participants. J Neurosci Nurs. 2019;51(2):84–88. DOI · PMID 30489422.
12. Maxin AJ, et al. A smartphone pupillometry tool for detection of acute large vessel occlusion. J Stroke Cerebrovasc Dis. 2023;32(12):107430. DOI · PMID 37857150.