Volume 59, Issue 1 p. 34-37
Free Access

Simple bedside assessment of level of consciousness: comparison of two simple assessment scales with the Glasgow Coma scale*

A. F. McNarry

A. F. McNarry

Research Registrar

Search for more papers by this author
D. R. Goldhill

D. R. Goldhill

Senior Lecturer and Honorary Consultant in Anaesthesia and Critical Care Medicine, Department of Anaesthesia and Critical Care, Royal London Hospital, Barts and the London NHS Trust, Alexandra Wing, The Royal London Hospital, Whitechapel, London E1 1BB, UK

Search for more papers by this author
First published: 28 June 2008
Citations: 132
A. F. McNarry
[email protected]

Presented in abstract form at the 16th Congress of the European Society of Intensive Care Medicine in Amsterdam, October 2003.


Neurological assessment is an essential component of early warning scores used to identify seriously ill ward patients. We investigated how two simple scales (ACDU – Alert, Confused, Drowsy, Unresponsive; and AVPU – Alert, responds to Voice, responds to Pain, Unresponsive) compared to each other and also to the more complicated Glasgow Coma Scale (GCS). Neurosurgical nurses recorded patients' conscious level with each of the three scales. Over 7 months, 1020 analysable measurements were collected. Both simple scales identified distinct GCS ranges, although some overlap occurred (p < 0.001). Median GCS scores associated with AVPU were 15, 13, 8 and 6 and for ACDU were 15, 13, 10 and 6. The median values of ACDU were more evenly distributed than AVPU and may therefore be better at identifying early deteriorations in conscious level when they occur in critically ill ward patients.

The Audit Commission's 1999 Report ‘Critical to Success’ stated that a ‘highest priority recommendation’ was agreeing ‘danger signs’ to help identify ward patients at risk of deterioration [1]. Early warning scores (EWSs) are a method of identifying such patients [2]. The Intensive Care Society provides five different examples of an EWS in this report. Each of the five examples includes an assessment of the central nervous system or level of consciousness, although the methods used are not identical [3–7]. These scores have yet to be scientifically validated; however, we have shown the importance of neurological assessment in identifying patients who might require escalation to a critical care area and in the prediction of 30-day outcome [8, 9]. The Glasgow Coma Scale (GCS) is a 13-point scale described in 1974 by Teasdale & Jennet [10]. The GCS evaluates consciousness by scoring a response in three areas: eye opening, motor response and verbal performance. Some EWSs do use the GCS, although the application of the GCS requires skill to achieve consistency in scoring. Its use also complicates the calculation of the EWS, as a value obtained from calculation of the GCS must then be awarded a separate score to allow its incorporation within an EWS. Other EWSs use the AVPU scale (Alert, responds to Voice, responds to Pain and Unresponsive) to assess neurological status. This scale is widely used for the assessment of trauma patients as originally taught in Advanced Trauma Life Support (ATLS) [11]. Mackay et al.[12] have compared AVPU with the GCS. Their study population included patients aged 5 years or older being transferred by ambulance to an Accident and Emergency department. They suggested that a GCS of 13 was the division between Alert and responds to Voice, while a GCS of 9 was the division between responds to Voice and responds to Pain. Anecdotal evidence suggests that AVPU is simpler to use than the GCS but it may not identify the subtle changes sometimes seen in ward patients where consciousness may be altered by metabolic derangements, hypoxia or hypotension rather than by a direct traumatic insult.

To overcome this perceived difficulty the scale ACDU (Alert, Confused, Drowsy, Unresponsive) has been suggested. We therefore undertook to compare ACDU to AVPU, and to compare both scales to the GCS.


Following ethics committee approval, neurosurgical nurses on one ward, who were experienced in the application and recording of the GCS, were asked to record the conscious level of their patients using the AVPU, ACDU and GCS scales. All assessments were performed in that order. All patients in whom a GCS assessment was clinically required were eligible for inclusion in the study. Each group of three observations was recorded on a separate data sheet so that each assessment was treated as an isolated observation. We did not collect data on the number of assessments performed on each patient and no demographic or diagnostic details were recorded for any of the patients.

Nurses making the observations were given information about the various scales and time was taken to ensure that all staff were familiar with the scales. However, no specific training was given in the application of either of the simpler scales. Staff were encouraged to use their own judgement, as this is how the AVPU and ACDU scales would be likely to be applied in an EWS by general ward staff. The participating nurses were also asked to comment on which of the simpler scales they found easier to use in each situation (AVPU, ACDU or neither). Statistical analysis was performed using GraphPad Prism Version 4.00 (GraphPad, San Diego, CA). The data sheet is included as Appendix A.


In all, 1062 data sheets were collected over a 7-month period. Of these, 1000 records compared AVPU, ACDU and the GCS, while a further 20 compared ACDU and GCS alone. Forty-two records (4.0%) were excluded from analysis as they were inadequately completed. The question concerning which of the simpler scales was easier to use was answered in 736 data sets, 389 expressing a preference between AVPU and ACDU.

The relationship between AVPU, ACDU and the GCS is shown in Fig. 1. The relationship between the median values and ranges of the GCS described by each category of the AVPU and ACDU scales is shown in Table 1. Table 2 summarises the preferences for the scales expressed by the staff.

Table 1. Median GCS and degree of dispersion associated with components of AVPU and ACDU.
Number of observations 404 (40.4%) 441 (44.1%) 88 (8.8%) 67 (6.7%) 493 (48.3%) 266 (26.1%) 140 (13.7%) 121 (11.9%)
Median GCS 15 13 8 6 15 13 10 6
50% Range
15–15 10–14 7–9 5–7 14–15 13–15 9–11 6–9
12–15 9–14 5–12 3–10 11–15 10–14 8–14 3–10
7–15 5–15 4–14 3–13 7–15 6–15 6–14 3–14
Table 2. Summary of views of those who recorded a preference with associated GCS (ACDU easier to use, AVPU easier to use, Neither scale easier to use).
Preferred ACDU Preferred AVPU Preferred neither
Number 173 (23.5 %) 216 (29.3 %) 347 (47.1 %)
Median GCS of
14 10 14
13–14 8–14 11–15
Range 6–15 3–15 3–15

When the data sets containing both AVPU and ACDU observations were compared (1000 pairs), a significant difference between the two scales was found (Wilcoxon matched pairs, p < 0.0001). Each component of AVPU and ACDU described a statistically distinct range of GCS values (Kruskal–Wallis p < 0.0001, with Dunn's Multiple Comparisons post test p < 0.001 all comparisons). While the median GCS of the Confused group and the responds to Voice group are the same, the interquartile range (25–75% of values) is narrower and higher in the Confused group. The median values of ACDU were more evenly distributed than AVPU. The nurses preferred to describe conscious level using AVPU when the GCS was lower, while ACDU was preferred when the GCS was higher (Kruskal–Wallis p < 0.0001, Dunn's Multiple Comparisons post test p < 0.001 for AVPU compared to ACDU and ‘no preference’ to AVPU; ACDU compared to ‘neither preferred’ was not significant).

The same ranked position was used to describe the GCS (Alert and Alert 388, responds to Voice and Confused 225, responds to Pain and Drowsy 34, Unresponsive and Unresponsive 67) in 71% (714/1000) of the observations where both AVPU and ACDU were recorded.


Our findings describe the relationship between AVPU, ACDU and the GCS scoring systems. They provide evidence of the median GCS values and variation associated with the components of the simple scores. Both AVPU and ACDU were able to stratify patients by GCS ranges. Our results in the ‘responds to Voice’ and ‘responds to Pain’ groups are not dissimilar to the results of Mackay et al.[12] who described a median value of 13 for ‘responds to Voice’ (interquartile range 12–14) and a median of 8 (interquartile range 6–10) for ‘responds to Pain’. Our results suggest that ‘Confused’ and ‘Drowsy’ distinguish distinct groups in the midrange of GCS assessments better than ‘responds to Voice’ and ‘responds to Pain’.

We used subjects who were on a neurosurgical ward and in whom measurement of the GCS was indicated. This selected group therefore included a relatively high percentage of those with a GCS < 15. The results were recorded by a small number of nurses on one ward who were skilled in the use of the GCS. The relatively short time scale for data collection reduced the number of new staff who had to be introduced to the study, with the potential for error. Neurosurgical nurses have experience in assessing level of consciousness and this distinguishes them from staff on a general ward assessing neurological state as part of an EWS. We do not feel that this is likely to alter the conclusions substantially, as, whilst the neurosurgical nurses use the GCS routinely in their clinical practice, they would have had little exposure to AVPU or ACDU. Most of the patients in this study would have a central nervous system cause for a decrease in their level of consciousness. On general wards there are likely to be a wider range of causes, including metabolic, hypoxic and perfusion related events.

To our knowledge, the ACDU subjective scale has not been previously evaluated. All assessments were made in the same order − AVPU, ACDU and then GCS. We felt it important to score GCS last to minimise the incidence of nurses assigning categories of the simpler scores to GCS values. In retrospect, it would have been desirable to randomise the order in which AVPU and ACDU were scored to minimise potential bias between these measurements. Similarly, we cannot distinguish in the results between those who marked just the same rank of description (Voice and Confused or Pain and Drowsy) for an assessment episode and those who genuinely assessed the patient's conscious level according to the descriptions available. We deliberately chose not to collect any other data in order to simplify data collection, accuracy and compliance. We therefore do not have information on how many data sets were collected for individual patients. As the study was made to compare measurement scales, we do not believe that this is relevant to the conduct of the study.

None of the simpler scores should replace GCS for the formal evaluation of a critically ill patient. Our intensive care outreach experience suggests that AVPU is inadequate for detecting early changes in conscious level that can occur in patients with impending or actual critical illness. It is unclear how to score confused or disorientated patients who clearly have an altered level of consciousness but may not respond to voice. ACDU is a four-point scale that has been suggested by others and was the tool we chose to assess against AVPU. When a preference was expressed, the nurses liked ACDU when there were relatively small decreases in the GCS and seemed to prefer AVPU for those with greater degrees of neurological dysfunction. Compared to AVPU, ACDU may be better for the simple ward assessment of seriously ill patients. We believe that the ACDU scale should be considered as a screening tool for deterioration in conscious level. It is simpler to use than the GCS and is easily incorporated into an EWS.


The authors are grateful for the participation of the nursing staff, Royal Ward, The Royal London Hospital, without whose help this study would not have been possible.


Appendix A

GCS/AVPU/ACDU comparison study:

Please complete a separate data sheet each time observations are carried out.

Please do not identify yourself or the patient on this form

Please carry out the following assessments, using your own clinical judgement for the first two, and the standard measuring techniques for the Glasgow coma scale.

Is the patient?  (tick one)  Scale 1
Alert and Orientated
Responds to Voice
Responds to Pain
Is the patient?  (tick one)  Scale 2
Alert and Orientated

Which scale do you find easier to use? Remember that each clinical scenario may make one or other scale better at that particular time − you don't have to stick by one answer every time!
AVPU (Scale 1)  ACDU (Scale 2)  Neither

Please circle your choice

Now, please record the Glasgow coma scale in the usual way:

Eyes open Spontaneously   4
To Speech   3
To Pain   2
None   1
Best verbal response Appropriate/Orientated   5
Confused   4
Inappropriate Words   3
Incomprehensible Sounds   2
None   1
Best motor response Obeys Commands   6
Localises to pain   5
Flexion (withdrawal)   4
Abnormal Flexion   3
Extension   2
None   1
Total GCS /15