AVPU vs GCS: how the two consciousness scales compare

What AVPU and the GCS each record

AVPU and the Glasgow Coma Scale both describe level of consciousness, but at very different resolution. AVPU is a four-level responsiveness scale: the patient is Alert, responds to Voice, responds to Pain, or is Unresponsive. It takes seconds, which is why it appears in ward early warning scores. The Glasgow Coma Scale, described by Teasdale and Jennett in The Lancet in 1974, rates three separate responses: eye opening from 1 to 4, verbal response from 1 to 5 and best motor response from 1 to 6, for a total of 3 to 15. A patient therefore sits on one of 4 AVPU levels but one of 13 GCS totals, and many more E-V-M combinations. AVPU says roughly how much stimulus it takes to get a response; the GCS records the best response in each of three domains, so it can show a change within a single AVPU level, such as speech becoming confused while the eyes stay open.

How AVPU levels line up with GCS scores

Each AVPU level corresponds to a range of GCS scores rather than a single number, so there is no exact AVPU to GCS conversion. Kelly, Upex and Bateman assessed both scales in 1,384 patients admitted to hospital with poisoning (Annals of Emergency Medicine, 2004). Of these, 1,138 were alert, 114 responded to voice, 87 responded to pain and 15 were unresponsive. The median GCS was 15 for alert, 13 (interquartile range 12 to 14) for voice, 8 (7 to 9) for pain and 3 for unresponsive, and the ranges for neighbouring levels overlapped. McNarry and Goldhill collected 1,020 measurements by neurosurgical nurses scoring each patient on AVPU, the GCS and a third scale (Anaesthesia, 2004). They found median GCS scores of 15, 13, 8 and 6. The two studies agree on the first three levels: a patient who responds only to pain has a median GCS of about 8, the top of the severe band.

Kelly's nursing staff recorded more difficulty using the GCS than AVPU, and alcohol-intoxicated patients were the hardest to assess on either scale. Every unresponsive patient needed intubation, and no patient with a GCS above 6 was intubated. McNarry and Goldhill's third scale, ACDU (alert, confused, drowsy, unresponsive), had medians of 15, 13, 10 and 6, more evenly spread than AVPU, which they suggested may make it better at picking up early deterioration.

Which scale predicts deterioration better?

On general wards, the Glasgow Coma Scale predicted death more accurately than AVPU in the largest comparison so far. Zadravecz and colleagues studied adult general ward patients at a single urban academic medical center, where nurses recorded the GCS and the Richmond Agitation-Sedation Scale (RASS) (Journal of Hospital Medicine, 2015). AVPU was taken from the eye subscale of each GCS. Across 295,974 paired observations from 26,873 admissions, 417 admissions (1.6%) ended in death in hospital. For death within 24 hours of an observation, the area under the ROC curve was 0.80 for the GCS, 0.82 for RASS and 0.73 for AVPU, and using the GCS and RASS together raised it to 0.85. The authors concluded that routine tracking of the GCS or RASS on general wards may improve detection of deterioration. Two cautions apply: AVPU was inferred from eye opening rather than assessed at the bedside, and the data came from one hospital.

Where each scale is used

The two scales sit inside different tools, with different needs for detail. The Royal College of Physicians' NEWS2 (2017) uses ACVPU, which adds new Confusion to AVPU. Alert scores 0, while new confusion, response to voice, response to pain or unresponsiveness all score 3, a red score on its own. NEWS2 therefore treats the step from alert to newly confused as seriously as unresponsiveness, and grades no further. The Sepsis-3 quick SOFA (qSOFA) counts altered mentation, described by the task force as any GCS below 15, as one of its three criteria. For traumatic brain injury the GCS is the standard grading, and the 2022 Lancet Neurology Commission gives mild as 13 to 15, moderate as 9 to 12 and severe as 8 or less. In practice, AVPU or ACVPU suits a quick screen and an early warning score, and the GCS suits neurological observations, head injury and any setting where a change of one or two points matters.

Worked example: three patients on both scales

Three patients show what AVPU keeps and what it loses. A man who has fallen opens his eyes when spoken to, is confused, and localises to pressure: E3V4M5, a GCS of 12, moderate on the head injury bands. On AVPU he responds to voice, and in NEWS2 his consciousness item scores 3. A woman with pneumonia has her eyes open and obeys commands but is newly disorientated: E4V4M6, a GCS of 14. On plain AVPU she is alert, but on ACVPU she has new confusion, which scores 3 in NEWS2, and her GCS below 15 meets the qSOFA mentation criterion. A third patient opens her eyes only to fingertip pressure, moans, and flexes normally: E2V2M4, a GCS of 8, severe on those bands and in line with the median of 8 for responds to pain in both 2004 studies. If her motor response falls from flexion to extension, her GCS drops to 6 while AVPU stays at P.

Enter each patient's eye, verbal and motor responses in the Glasgow Coma Scale calculator to see the total and the E-V-M notation.