Emergency Medicine

Is AVPU comparable to GCS in critical prehospital decisions? – A cross-sectional study

a b s t r a c t

Background: Advanced Trauma Life Support field triage utilizes the Glasgow Coma Scale (GCS) to assess the level of consciousness. However, prehospital care providers in Low- and middle-income countries (LMICs) often use the Alert, Verbal, Pain, and Unresponsive (AVPU) scale to assess the level of consciousness. This study aimed to determine whether prehospital AVPU categorization correlates with mortality rates in Trauma victims, similarly to GCS.

Methods: In this cross-sectional study conducted between November 2015 and January 2016, we enrolled a convenience sample of prehospital trauma-related field activations. The primary outcome measure was the prob- ability of death within 48 h for each category of AVPU. Results: In a convenience sample of 4514 activations, 1606 (35.6%) met exclusion criteria, four did not have AVPU, and four did not have GCS, leaving 2900 (64.2%) Trauma activations with both AVPU and GCS available for anal- ysis. Forty-eight-hour follow-up data were available for 2184 (75.3%) activations out of these 2900. The 48-h mortality rates for each category of AVPU were 1.1% (Alert), 4.3% (Verbal), 17.9% (Pain), 53.2% (Unresponsive); and, for each GCS-based injury severity category, they were 0.9% (Mild, GCS 13-15), 8.1% (Moderate, GCS 9-12), 43.5% (Severe, GCS <= 8). Overall, there was a statistically significant difference in GCS for each category of AVPU (p < 0.001) except between patients responding to verbal commands and those responding to pain (p = 0.18). The discriminative ability of AVPU (AUC 79.7% (95% CI 73.4-86.1)) and GCS (AUC 81.5% (95% CI 74.8-88.2)) for death within 48-h following hospital drop-off were comparable.

Conclusion: EMT assessments of AVPU and GCS relate to each other, and AVPU predicts mortality at 48 h. Future studies using AVPU to assess the level of consciousness in prehospital trauma protocols may simplify their global application without impacting the overall quality of care.

(C) 2022

  1. Introduction

Advanced Trauma Life Support (ATLS) emphasizes field triage to minimize the on-scene time [1]. Field Triage Decision Schemes are intended to rapidly identify critically-injured patients using the Glasgow Coma Scale (GCS) to assess the level of consciousness [2]. Most ambulance personnel in low and middle-income countries (LMICs) are basic emergency medical technicians , who typically do not utilize GCS. Instead, they often use the AVPU scale to assess the level of consciousness, assigning patients to one of 4 categories: A = Alert; V = responds to Verbal commands; P = responds to Pain; and,

* Corresponding author at: Stanford University School of Medicine, 900 Welch Road, Suite 350, Room 315, Stanford, CA 94305, USA.

E-mail address: [email protected] (S.R. Janagama).

U = Unresponsive. Differing scales to assess the level of consciousness at the ambulance-hospital interfaces can make signouts and care transi- tions complicated and inefficient. EMTs’ pre-arrival hospital notification is essential to the early preparation and mobilization of resources for trauma care in the emergency departments and hospitals. [1,3-5].. ATLS emphasizes Prehospital personnel obtaining and reporting infor- mation required for triage and care, including the patient’s history of present illness, physical exam, and other assessments like GCS [1]. An index that converts prehospital AVPU assessments to a GCS severity categorization (mild, moderate, severe) could potentially facilitate the use of field trauma triage and the flow of information between providers.

Minimizing prehospital delays, early initiation of hospital care, and adequate resuscitation can prevent a high proportion of trauma fatali- ties [6-8]. Sixty percent of Trauma deaths in LMICs are preventable


0735-6757/(C) 2022

compared to just 6.7% in high-income countries [6-8]. Annually, LMICs could potentially avert an additional 1.7 million injury-related deaths if they can reduce their case-fatality rate to the level of high-income countries [9]. This study aimed to assess whether prehospital AVPU cat- egorization correlates with GCS severity categorization and, similarly, with mortality rates in trauma victims.

  1. Method
    1. Overview and setting

This cross-sectional study relied on a convenience sample of trauma- related prehospital activations from a single prehospital agency: GVK Emergency Management and Research Institute (GVK EMRI). Data was collected between Nov 16, 2015, and Jan 30, 2016 (Monday to Saturday 8 AM to 5 PM for five weeks) across seven states in India (Andhra Pradesh, Assam, Gujarat, Himachal Pradesh, Karnataka, Me- ghalaya, and Telangana).

GVK EMRI is a prehospital ambulance service provider that partners with state governments to provide prehospital emergency medical services across India. They operate medical dispatch centers, train EMTs to provide patient care, manage prehospital ambulance activa- tions, and implement quality improvement activities. Their operations are centralized at the state level, and each state has its own operating unit. Across the 17 states they serve, GVK EMRI attends approximately 25,000 prehospital activations every day and has handled close to 90 million activations since beginning operations in 2005 [10].

GVK EMRI EMTs undergo a 52-day training program based on an educational curriculum comparable to that of the National Highway Traffic Safety Administration in the United States. All EMTs follow standard operating procedures delineated in a manual of prehospital care protocols, and they attend the annual refresher training program. While Basic EMTs do not learn about GCS during their initial training, they are introduced to the concept during their refresher training. How- ever, the individual components of GCS are available to all Basic EMTs on the Patient care report form to facilitate recording the patient’s GCS score.

Prehospital patients are cared for by EMTs in the field and are typically transported to the nearest appropriate government hospital but may be taken to private hospitals at the patient’s request. Govern- ment hospitals are categorized as primary, secondary, and tertiary. Primary hospitals function only during the day, do not provide intensive care services, and do not have Operating rooms. Secondary hospitals function around the clock, can provide some intensive care services, and all have operating rooms. Tertiary hospitals are teaching hospitals that operate 24 h/day, uniformly provide intensive care services, and all have operating rooms. At the time of this study, a handful of tertiary hospitals were designated as trauma care centers.

    1. Data collection

To capture data, we designed and used a secure form on REDCap [11]. Trained research assistants identified trauma-related activations in real-time, utilizing a software application connected to the ambu- lance dispatch center, and notified the EMTs involved in the trauma activation about potential enrollment into the study and data required for collection. Apart from excluding activations that met the defined criteria (Box 1), we only analyzed cases that had both AVPU and GCS data, and also 48-h follow-up data on mortality.

We collected data on AVPU and the individual components (eye, verbal, motor) of GCS at initial patient pickup and also determined the mortality at three distinct times: at initial patient pickup, at hospital drop-off, and at 48-h post-hospital drop-off. Further, we categorized GCS scores into three established ATLS brain injury severity categories (severe injury: GCS <= 8; Moderate Injury: GCS 9-12; and, Mild Injury: GCS 13-15). The Institutional Review Board at Stanford University

Box 1

Exclusion criteria.

  1. Hospital-to-hospital transfers, where the length of stay at the initial treating hospital is greater than 12 hours.
  2. Ambulance dispatch issues such as a dispatch cancellation, duplicate dispatch, or incorrect dispatch data.
  3. Research assistant could not contact the EMT.
  4. Injury primarily involved burns, electrocution, lightning, smoke inhalation, or an isolated animal bite.
  5. No evidence of trauma/injury on EMT arrival.
  6. EMT was caring for more than one trauma patient during a single activation.
  7. EMT transferred the patient to another ambulance provider.
  8. Patient died before the ambulance arrived.

deemed the study exempt as we used data that was collected for quality improvement purposes. The Ethics Committee at GVK EMRI approved the publication of the findings.

    1. Data analysis

The primary outcome measure was the death within 48-h following hospital drop-off for each category of AVPU. We calculated the probabil- ity of death and derived the odds of death within 48-h following hospital drop-off for both AVPU and GCS categories. The secondary out- comes were the differences in median GCS within each category of AVPU, and the discriminative ability of AVPU versus GCS for deaths within 48-h following hospital drop-off. We used the Kruskal-Wallis test followed by the Dunns Multiple Comparison to compare the median GCS for each category of AVPU. We used the area under the re- ceiving operator characteristic curve to test the discriminative ability and reported its 95% confidence interval. For data analysis, we used R (V.3.5.1) with R- studio (V.1.1.447, RStudio, Inc.).

  1. Results

In a convenience sample of 4514 activations, 1606 (35.6%) met exclusion criteria. Among 2908 (64.4%) activations considered for en- rollment, 4 (0.1%) did not have AVPU recorded, and 4 (0.1%) did not have GCS data recorded. The remaining 2900 (64.2%) cases had both AVPU and GCS data for analysis. One-hundred and sixty-six patients were transported to primary hospitals (5.7%), 1753 to secondary hospi- tals (60.4%), 393 to tertiary hospitals (13.6%), and 539 to private hospi- tals (18.6%). Destination hospital data was unavailable for 49 cases (1.7%).

    1. AVPU and GCS

Overall, each individual categorization of AVPU mapped to a significantly different median GCS score (Kruskal-Wallis test, H(3) = 936, p < 0.001) (Fig. 1). However, on multiple comparisons, the differ- ence between the median GCS of patients responding to verbal commands (GCS 12) and patients responding to pain (GCS 10) was not statistically insignificant (p = 0.18) (Table 1). The GCS severity categorization seemed to correlate with the AVPU categories as these variables lacked independence on chi-squared testing (?2 = 2517.8, df = 6, p < 0.001).

    1. Death at 48 hours

Among the 2900 prehospital activations with available AVPU and GCS data, 2184 (75.3%) were followed-up with at 48-h post-hospital arrival, and 61 (2.8%) of these patients died. The Probability of death

Image of Fig. 1Table 2

Probability and odds of death at 48 h post drop-off. AVPU

Patients Deaths Odds ratio




95% CI


95% CI
































Mild (GCS13-15)





Moderate (GCS 9-12)







Severe (GCS 3-8)











Fig. 1. Median and interquartile range of GCS for each category of AVPU.

at 48-h post-drop-off were comparable at the different types of destina- tion hospitals – primary (0% (0/129) 95% CI 0.0-3.6), secondary (2.9%

(38/1324) 95% CI 2.1-4.0), tertiary (3.5% (10/286) 95% CI 1.8-6.5) and

private (2.8% (12/423) 95% CI 1.5-5.0). The probability of death did not vary even among the prehospital activations that did not have destination hospital data (4.5% (1/22) 95% CI 0.2-24.9).

Compared to Alert patients, the probability of death within 48-h post-drop-off was 50-fold higher for Unresponsive patients, nearly 18-fold higher for patients responding to Pain, and four times higher for patients responding to Verbal commands. There is a three to four- fold increase in the probability of death and a four to five-fold increase in odds of death as severity increases from A to V to P to U (Table 2). Likewise, there is a five to nine-fold increase in the probability of death and nearly nine-times higher odds of death for each successive GCS severity categorization from mild to moderate to severe (Table 1).

    1. AVPU and GCS discriminative ability

The discriminative ability of AVPU (AUC 79.7% (95% CI 73.4-86.1)) and GCS (AUC 81.5% (95% CI 74.8-88.2)) for death within 48-h follow- ing hospital drop-off were comparable (Fig. 2).

  1. Discussion

For prehospital trauma activations, our findings reveal a multifold increase in the probability of death within 48-h as the severity increases on both the GCS and AVPU assessment scales. We found a statistically significant difference in the median GCS that correlated with each AVPU categorization, although the GCS ranges for the Verbal and Pain categories did overlap. Prior literature has mixed evidence for the differ- ence in median GCS for verbal and pain categories, with a few studies reporting an overlap and others reporting a significant difference [12-16]. Given the multifold increase in the probability of death for each category of both AVPU and GCS, the comparable discriminative ability of AVPU and GCS, and the significant difference in median GCS for each category of AVPU, the idea of mapping AVPU to a GCS range may be supported by this evidence.

Table 1

Dunn’s multiple comparison.?





16.018 (<0.001)

Unresponsive Verbal

20.289 (<0.001)

18.207 (<0.001)

3.302 (<0.001)

-0.911 (0.181)

-4.522 (<0.001)

* H (p-value).

The existing literature already supports the use of early warning scores that rely on the AVPU scale, as opposed to the GCS, in trauma [17-19]. The modified early warning score and the National Early Warning Score are examples of such scores. Since the introduction of the initial variant in 1997, numerous studies and reviews have shown their utility in predicting mortality [17-19]. Therefore, the increased use of AVPU in scales and protocols may be reasonable and effective given this prior evidence.

The perceived need for an AVPU-to-GCS conversion stems from the fact that ambulance medical personnel may be more comfortable using AVPU scoring when applying trauma-care guidelines that traditionally rely on GCS scores. For example, the ATLS Field Triage De- cision Scheme, which only utilizes the GCS for assessment of the level of consciousness, helps prehospital personnel choose the most appropri- ate destination for the transport of trauma victims. In addition, GCS findings are integral to several other trauma-care protocols, including those for trauma airway management.

pre-arrival notification of the patient’s current condition contributes to the early initiation of appropriate trauma care and mobilization of resources at the receiving facility [1]. Early initiation of care is one of the recommended opportunities for intervention (OFI) to reduce the burden of Preventable deaths in LMICs [6-8]. In one study involving

Image of Fig. 2

Fig. 2. AVPU and GCS ROC curves compared.

2523 trauma deaths, traumatic brain injury-related (14.9%[307]), airway-related (14.3%[293]), and prehospital delay-related (10.3% [212]) OFIs topped the list [7].

Established EMS systems in high-income countries have advanced EMTs on ambulances who routinely provide pre-arrival notification, thereby enabling a smoother transition at the ambulance-hospital inter- face. Advanced EMTs also have a broader skillset compared to basic EMTs. However, advanced EMTs are more costly and take longer to train. In India, the basic EMT is trained over two months, whereas an Advanced EMT requires two years [20].

LMICs, like India, rely primarily on basic EMTs, who are less expen- sive and easier to recruit, train and deploy. India has expanded its prehospital care services across the country in part by having one EMS entity provide for both ambulance operations as well as basic EMT train- ing [21]. Launched in 2005, this organization, an Indian public-private partnership, has trained over 100,000 Basic EMTs to work on their 10,000 plus ambulances. However, it is estimated that they will need to train an additional 65,000 additional EMTs to meet the prehospital personnel needs of the entire population. Given the conflicting evidence for the utility of advanced prehospital care providers [22], a high attrition rate among current prehospital care providers, and the ever- increasing demand for ambulance staffing, the most likely approach to meet the growing needs of India’s EMS system is to train and deploy basic EMTs.

It may be unreasonable to expect these basic EMTs, who are more comfortable with the AVPU scale, to reliably assess the GCS in trauma victims as part of their routine clinical practice. Gaining proficiency in assessing the GCS requires time and experience, and the current litera- ture points to high interrater variability of GCS assessments [23], even among physicians who have more extensive training and experience than basic EMTs.

A disconnect between protocol requirements (assessing GCS) and prehospital personnel capabilities (comfort with AVPU) can adversely impact decision-making, patient care, and outcomes. Two options seem appropriate to address this disconnect: a model that converts AVPU into most likely GCS-severity categorization or directly substitut- ing AVPU for GCS in standardized trauma protocols. Further studies can test these hypotheses and determine if such an approach can provide comparable outcomes to current-day trauma care.

    1. Limitations

Aside from a patient’s level of consciousness, many factors can influ- ence mortality rates at 48 h. For patients with the same GCS or AVPU categorization, the type of the destination hospital, availability of inten- sive care services and operation rooms, time to treatment, and the clinical nature of the injury might have influenced their overall out- comes, thereby impacting the external validity of our findings. Moreover, the use of a convenient sample limits the strength of our con- clusions. A quarter of prehospital activations could not be followed up, and though unlikely, significant differences in mortality rates among this group could have altered our findings.

  1. Conclusion

AVPU recorded by Basic EMTs seems to be a dependable surrogate for GCS in prehospital care settings. AVPU recorded by Basic EMTs pre- dicts the probability of death at 48 h; AVPU seems to correlate closely with the recorded GCS severity categorization. Future studies using AVPU to assess the level of consciousness in prehospital trauma proto- cols may simplify their global application without impacting the overall quality of care.

Prior presentations


Financial support

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Author contributions

SAM, MCS, JAN, GVR, MAK, and SRJ contributed to the study design and manuscript production. SRJ, JAN, and MAK contributed to data analysis.

Credit authorship contribution statement

Srinivasa R. Janagama: Writing – review & editing, Writing – original draft, Supervision, Software, Resources, Project administration, Methodology, Investigation, Formal analysis, Data curation, Conceptual- ization. Jennifer A. Newberry: Writing – review & editing, Supervision, Resources, Project administration, Methodology, Investigation, Formal analysis, Conceptualization. Michael A. Kohn: Methodology, Formal analysis, Conceptualization. G.V. Ramana Rao: Writing – review & editing, Supervision, Resources, Project administration, Methodology, Conceptualization. Matthew C. Strehlow: Writing – review & editing, Supervision, Resources, Methodology, Investigation, Conceptualization. Swaminatha V. Mahadevan: Writing – review & editing, Writing – original draft, Supervision, Methodology, Investigation, Conceptualization.

Declaration of Competing Interest

SAM, SRJ, MCS, MAK, and JAN do not have any conflict of interest to disclose. GVR is a full-time employee of the GVK Emergency Manage- ment Research Institute, which operates the ambulance services studied here.


We would like to thank the team of research assistants at GVK EMRI who helped to administer this study: Aruna Gimkala, Aparna Chakraborty, Steffy Christian, Kabyashree Gogoi, Roshidul Islam, Chetana Jani, Paiabaskhem Mukhim, Hanamesh Nagappa, Dharmesh Prajapati, Dilip Pate, Divya Patel, Komala PJ, Hem Prakash Thakur, Isberth Tham, Om Raj, Marada Lakshmana Rao, and Munirathnamma Venkateshappa.


  1. Advanced Trauma Life Support(R)10th ed.. Chicago: American College of Surgeons; 2018..
  2. McCoy E, Chakravarthy B, Lotfpour S. Guidelines for field triage of injuried patients. West J Emerg Med. 2013;14(1):69-76.
  3. Handolin LE, Jaaskelainen J. Pre-notification of arriving trauma patient at trauma center: a retrospective analysis of the information in 700 consecutive cases. Scand J Trauma Resusc Emerg Med. 2008;16(1):1-5. https://doi.org/10.1186/1757-7241- 16-15.
  4. Lillebo B, Seim A, Vinjevoll OP, Uleberg O. What is optimal timing for trauma team alerts? A retrospective observational study of alert timing effects on the initial man- agement of trauma patients. J Multidiscip Healthc. 2012;5:207-13. https://doi.org/ 10.2147/JMDH.S33740.
  5. Synnot A, Karlsson A, Brichko L, et al. Prehospital notification for major trauma pa- tients requiring emergency hospital transport: a systematic review. J Evid Based Med. 2017;10(3):212-21. https://doi.org/10.1111/jebm.12256.
  6. Yeboah D, Mock C, Karikari P, Agyei-Baffour P, Donkor P, Ebel B. Minimizing prevent- able trauma deaths in a limited-resource setting: a test-case of a multidisciplinary panel review approach at the Komfo Anokye teaching Hospital in Ghana. World J Surg. 2014;38(7):1707-12. https://doi.org/10.1007/s00268-014-2452-z.
  7. Roy N, Kizhakke Veetil D, Khajanchi MU, et al. Learning from 2523 trauma deaths in India- opportunities to prevent in-hospital deaths. BMC Health Serv Res. 2017;17

(1):1-8. https://doi.org/10.1186/s12913-017-2085-7.

  1. Sandal TL, Esposito TJ, Whitney JR, et al. Analysis of preventable trauma deaths and opportunities for trauma care improvement in Utah. J Trauma Inj Infect Crit Care. 2011;70(4):970-7. https://doi.org/10.1097/TA.0b013e3181fec9ba.
  2. Mock C, Joshipura M, Arreola-Risa C, Quansah R. An estimate of the number of lives that could be saved through improvements in trauma care globally. World J Surg. 2012;36(5):959-63. https://doi.org/10.1007/s00268-012-1459-6.
  3. GVK EMRI. lives saved. [Accessed Mar 01, 2022]. http://www.emri.in; 2022.
  4. Harris PA, Thielke R, Gonzalez N, Conde JG, Taylor R, Payne J. Research electronic data capture (REDCap)–a metadata-driven methodology and workflow process for providing translational research informatics support. J Biomed Inform. 2008;42 (2):377-81. https://doi.org/10.1016/j.jbi.2008.08.010.
  5. Kelly CA, Upex A, Bateman DN. Comparison of consciousness level assessment in the poisoned patient using the alert/verbal/painful/unresponsive scale and the Glasgow coma scale. Ann Emerg Med. 2004;44(2):108-13. [ISSN: 1097-6760].
  6. McNarry AF, Goldhill DR. Simple bedside assessment of level of consciousness: com- parison of two simple assessment scales with the Glasgow coma scale. Anaesthesia. 2004 Jan;59(1):34-7. https://doi.org/10.1111/j.1365-2044.2004.03526.x14687096.
  7. Hoffmann F, Schmalhofer M, Lehner M, Zimatschek S, Grote V, Reiter K. Comparison of the AVPU scale and the pediatric GCS in prehospital setting. Prehosp Emerg Care. 2016 Jul-Aug;20(4):493-8. https://doi.org/10.3109/10903127.2016.1139216. Epub

2016 Mar 8. 26954262.

  1. Nuttall AG, Paton KM, Kemp AM. To what extent are GCS and AVPU equivalent to each other when assessing the level of consciousness of children with head injury? A cross-sectional study of UK hospital admissions. BMJ Open. 2018;8(11):e023216 Published 2018 Nov 28. https://doi.org/10.1136/bmjopen-2018-023216.
  2. Zadravecz FJ, Tien L, Robertson-Dick BJ, et al. Comparison of mental-status scales for predicting mortality on the general wards. J Hosp Med. 2015;10(10):658-63. https://doi.org/10.1002/jhm.2415.
  3. Morgan RJM, Williams F, Wright M. An early warning scoring system for detecting developing critical illness. Clin Intensive Care. 1997;8:100.14.
  4. Nannan Panday RS, Minderhoud TC, Alam N, Nanayakkara PWB. Prognostic value of early warning scores in the emergency department (ED) and acute medical unit (AMU): a narrative review. Eur J Intern Med. 2017;45:20-31. https://doi.org/10. 1016/j.ejim.2017.09.027.
  5. Alam N, Hobbelink EL, van Tienhoven AJ, van de Ven PM, Jansma EP, Nanayakkara PWB. The impact of the use of the early warning score (EWS) on patient outcomes: a systematic review. Resuscitation. 2014;85(5):587-94. https://doi.org/10.1016/j.re- suscitation.2014.01.013.
  6. GVK EMRI. Training programs. [Accessed Mar 01, 2022]. http://www.emri.in/ training-program-2/; 2022.
  7. GVK EMRI. Who we are. [Accessed Mar 01, 2022]. http://www.emri.in/about-us/; 2022.
  8. Ryynanen OP, Iirola T, Reitala J, Palve H, Malmivaara A. Is advanced life support bet- ter than basic life support in prehospital care? A systematic review. Scand J Trauma Resusc Emerg Med. 2010;18(1):62. https://doi.org/10.1186/1757-7241-18-62.
  9. Gill M, Martens K, Lynch EL, Salih A, Green SM. Interrater reliability of 3 simplified neurologic scales applied to adults presenting to the emergency department with al- tered levels of consciousness. Ann Emerg Med. 2007;49(4):403-7. 407.e1. https:// doi.org/10.1016/j.annemergmed.2006.03.031.

Leave a Reply

Your email address will not be published. Required fields are marked *