A randomised controlled trial of verbal guidance versus verbal guidance supplemented by a photographic aid for bystander identification of intramuscular tranexamic acid injection sites in a simulated road injury scenario

Abstract

Background

Non-compressible haemorrhage is a leading cause of preventable death following road injury. Tranexamic acid (TXA), when administered early, improves survival. Intramuscular (IM) administration offers a feasible route for early administration by lay bystanders. However, the ability of bystanders to correctly identify safe IM injection sites remains unclear. This study aimed to evaluate whether verbal guidance supplemented by a photographic aid improves the accuracy of site identification in a simulated road injury scenario.

Methods

In this randomised controlled trial, 64 lay participants were recruited on a university campus and randomised to receive either (1) verbal guidance alone or (2) verbal guidance plus photographic aid to locate the deltoid injection site on a simulated injured person. Site identification was assessed via sticker placement, and three expert raters with diverse medical backgrounds independently reviewed standardised photographs to determine site safety using a majority agreement rule. The primary outcome was safe site identification. A binomial generalised linear model assessed the association between intervention group and correct site identification. Inter-rater reliability was measured using Fleiss’ Kappa.

Results

Participants in the verbal guidance plus photographic aid group were significantly more likely to identify a safe injection site compared to those receiving verbal guidance alone (87.5% vs. 62.5%; OR 4.67, 95% CI 1.33, 19.92, p = 0.03). The presence of concerns regarding site safety was also significantly lower in the photo and verbal group (18.8% vs. 53.1%, p = 0.002). No significant associations were found between accuracy and participant age, gender, prior training, or confidence.

Conclusions

Supplementing verbal guidance with a photographic aid significantly improves bystander accuracy in identifying safe IM TXA injection sites in a simulated setting. This finding supports the potential integration of visual aids into emergency dispatch protocols to enhance early haemorrhage control in trauma care. Further research is needed to assess real-world application and impact.

Clinical trial number

ISRCTN Registry: ISRCTN41280918.

Trial registration

ISRCTN Registry: ISRCTN41280918, 08/08/2025 Retrospectively registered.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12873-025-01323-8.

Background

Non-compressible haemorrhage (NCH) refers to bleeding that cannot be controlled by applying direct pressure [1]. NCH is the leading cause of preventable death following road injury worldwide [2–5]. Rapid haemorrhage control is lifesaving, and tranexamic acid (TXA), an antifibrinolytic agent, has been shown to reduce mortality when administered within three hours of injury [6]. Traditionally given intravenously, recent evidence has demonstrated the effectiveness of TXA delivered intramuscularly (IM), offering a practical alternative when intravenous access is not feasible [7, 8]. This development expands opportunities for earlier administration, including by non-medical responders. With survival benefit decreasing by 10% for every 15-minute delay in its administration, enabling bystanders to deliver TXA as soon as feasible following road injury, particularly when Emergency Medical Services (EMS) response is delayed or not available, may save lives [9].

Realising the benefits of IM TXA for road injury depends on the ability of bystanders; individuals who are not part of a formal clinical response to an incident, to identify and use an appropriate injection site under guidance. Unlike professional responders, these bystanders may lack confidence and knowledge of anatomical landmarks and injection technique. Therefore, understanding how best to support bystanders in this task is essential to optimising early treatment for NCH.

Bystanders represent a diverse group, including those with no prior experience, those with basic first aid knowledge, and professional responders such as police or fire personnel. Their role in the road injury chain of survival (Fig. 1) can be critical, particularly in rural or resource-limited settings where EMS response may be delayed or unavailable [10]. While successful bystander interventions are well established in other settings, such as in delivering chest compressions in cardiac arrest, the feasibility of many layperson-administered trauma interventions remains unclear [11].

Fig. 1.

The road injury chain of survival

To inform system design and support scalable solutions, this study evaluates whether lay bystanders, under remote guidance, can accurately identify a safe deltoid IM injection site. We compare verbal guidance with verbal guidance supplemented by a photographic aid, during a simulated emergency services call, to determine whether visual support improves identification of an IM injection site in a simulated road injury scenario.

Methods

A randomised simulation study was conducted to evaluate the effectiveness of remote guidance with and without visual aid support in assisting laypeople to identify the correct site for IM injection of TXA.

Eligible participants were adults (aged 18 years and older) recruited on a University Campus (Exeter University, UK). A designated research stand was set up in a university communal area, allowing individuals to voluntarily approach and enrol in the study. Those with formal medical, nursing or clinical qualifications were excluded (basic first aid training was not excluded). Following consent, participants completed a brief demographic questionnaire and were then randomly assigned to one of the two instructional groups: (1) verbal guidance alone (Additional File 1) or (2) verbal guidance supplemented by a photographic aid (Additional Fig. 1). The simulation involved approaching a vehicle containing one of 4 study team members (two women and two men) portraying an injured patient sitting in the driver’s seat of a standard road car. Participants were guided to locate the deltoid injection site and indicate their selected location for injection delivery using a small circular sticker. Standardised photographs were taken from three different angles to document their choice (Supplemental materials).

The standardised photographs were independently reviewed by a panel of three clinicians with a collectively diverse clinical background including nursing, emergency medicine and pre-hospital critical care, and each with greater than 5 years experience administering IM injections. Each practitioner assessed the photographs autonomously and answered two questions; “Is the location safe” and “Would you have concerns about receiving an injection at this site?” The primary outcome of the study was determined using a two-out-of-three concordance rule, where an injection site was considered correctly identified if at least two of the three practitioners agreed on its classification based on the first question. Additionally clinicians were asked to provide narrative feedback in relation to their decision making; this was not mandatory.

The sample size was determined based on a clinically meaningful absolute difference of 35% between the two groups. As there was no prior literature to guide this estimate, the threshold was agreed upon following discussion within the authorship team, considering what would represent a meaningful difference in effectiveness between training approaches. A significance level (\alpha) of 0.05 and a statistical power of 80% (1 - \beta = 0.8) were used to ensure an adequate likelihood of detecting a true difference between groups, should one exist. Sixty-four participants were allocated in a 1:1 ratio to either training method using block randomisation in blocks of eight.

Descriptive statistics were used to summarise baseline characteristics, percentage or mean (SD) are reported. Inter-rater reliability was assessed using Fleiss’ Kappa for multiple raters. The binary primary outcome, ‘is the location safe’ was analysed using a binomial generalised linear model with a logit link. Other comparisons were analysed using 𝜒² test or Fisher’s exact test where cell counts were too small. All statistical analyses were performed using R (Version 4.4.2).

Results

The study was performed on 22/11/2024 at the University of Exeter (UK) campus. Sixty-four participants were successfully recruited to the study (recruited to power). The age range of the participants was as expected for a university campus, with a mean (standard deviation, SD) 22.7 (7.4) years in the verbal guidance only group and 25.5 (10.6) years in the verbal guidance plus photographic aid group (Table 1). The majority of participants had never called the emergency services 71.9% (n = 23) and 62.5% (n = 20) (for verbal and verbal plus photographic aid groups respectively) and only half in each group had received any prior first aid training.

Table 1.

Baseline characteristics of study population

Variable		Verbal guidance only (N = 32)	Verbal guidance supplemented by a photographic aid (N = 32)
Gender	Men	31.2% (10)	56.2% (18)
	Women	68.8% (22)	43.8% (14)
Age (years)		22.7 (7.4)	25.5 (10.6)
Have you ever called the emergency services	Yes	28.1% (9)	28.1% (9)
	No	71.9% (23)	62.5% (20)
	Missing	-	9.4% (3)
Prior First aid training	None	46.9% (15)	46.9% (15)
	Basic	53.1% (17)	50.0% (16)
	Biker Down	-	3.1% (1)
Anticipated training requirements*	No training	25.0% (8)	50.0% (16)
	Half day first aid	25.0% (8)	15.6% (5)
	First aid certification	3.1% (1)	15.6% (5)
	Basic injection	43.8% (14)	12.5% (4)
	Missing	3.1% (1)	6.2% (2)

N = 64, % (N) or Mean (SD)

*In the pre-simulation questionnaire, participants were asked ‘What level of training do you think would be necessary before giving an injection at an accident scene if you knew it would benefit the patient and had minimal risks?’

Interrater reliability among the three raters to assess if the location was safe yielded a kappa value of 0.13 indicating poor agreement (Table 2). The success rates were 87.5% in the verbal instructions plus photographic aid group and 62.5% in the verbal only group. For the main outcomes, ‘is the location safe?’ and ‘would you have concerns about this injection site?’, a binomial generalised linear model was fitted adjusting for the variables intervention group, age, gender, actor gender, and prior first aid training (coded as yes or no; Table 3). The odds that the location is safe are 4.7 times higher for the verbal instructions plus photographic aid group compared to the verbal instructions only group (p-value 0.03). There were no significant associations found between the safety rating of the location and age, gender (bystander or actor) and prior first aid training.

Table 2.

Outcomes from study

Variable		Verbal guidance only (N = 32)	Verbal guidance supplemented by a photographic aid (N = 32)
Is the location safe	Yes	62.5% (20)	87.5% (28)
	No	37.5% (12)	12.5% (4)
Concerns	Yes	53.1% (17)	18.8% (6)
	No	46.9% (15)	81.2% (26)
Actor gender	Man	46.9% (15)	56.2% (18)
	Woman	53.1% (17)	43.8% (14)
How confident	Very confident	18.8% (6)	25.0% (8)
	Somewhat confident	21.9% (7)	40.6% (13)
	Neutral	25.0% (8)	25.0% (8)
	Somewhat unconfident	28.1% (9)	9.4% (3)
	Very unconfident	6.25% (2)	-

Table 3.

Location safety and concerns

	Is the location safe?		Would you have concerns about receiving an injection at this site?
Variable	Odds Ratio (95%CI)	p-value	Odds Ratio (95%CI)	p-value
Intervention (ref verbal instructions only)	4.67 (1.33,19.92)	0.03	0.13 (0.03,0.44)	0.002
Age	0.99 (0.92,1.07)	0.71	1.07 (1.00, 1.16)	0.06
Bystander gender ref women	0.86 (0.24, 3.24)	0.82	1.60 (0.47, 5.74)	0.45
Actor gender ref man	1.52 (0.43,5.65)	0.52	0.64 (0.18, 2.11)	0.46
Had first aid training	1.71 (0.47, 6.69)	0.42	1.07 (0.30, 3.80)	0.91

After the simulation, there was no significant difference at the 5% level found in confidence in giving an injection between the intervention groups (p-value 0.08) and no significant difference between ‘is the location safe’ and confidence (p-value 0.21). There was no significant difference between previous first aid training and confidence (p-value 0.72).

The odds of concerns in the verbal guidance supplemented by a photographic aid group are 87% lower compared to the verbal guidance only group (p = 0.002). No significant associations were found at the 5% level between concerns and age, gender bystander/actor or previous first aid training.

The rater agreement for concerns over the site of the injection had a kappa of 0.34, indicating fair agreement. In the verbal instructions only group the consensus found concerns on 53.1% (n = 17) sites and in the verbal instructions and photographic aid group 18.8% (n = 6). Of the 17 concerns in the verbal instructions only group, concerns were; too high (n = 8, 47.1%), risk of hitting a nerve (n = 4, 23.5%), too low (n = 3, 17.6%), no comment/not enough information in comment to classify (n = 2, 11.8%). Of the 6 concerns in the verbal and photographic aid group, concerns were; risk of hitting a nerve (n = 3, 50.0%), no comment/not enough information in comment to classify (n = 2, 33.3%), too low (n = 1, 16.7%).

Prior to the simulation participants were asked what level of training they thought would be necessary to give an injection at the scene of a collision if they knew it would benefit the patient and had minimal risks. 71.9% (n = 23) of the verbal instructions only group and 50% (n = 16) of the verbal instructions plus photographic aid groups thought that prior training, either first aid or injection training, was necessary (Table 1). After the simulation however, 40.6% (n = 13) in the verbal instructions only and 65.6% (n = 21) in the verbal instructions plus photographic aid group rated themselves confident or very confident. Of participants who commented on what would make them feel more confident, more people in the verbal instructions only group (n = 8) wanted more precise measurements, for example distance in centimeters to find the location, than in the verbal instructions plus photographic aid group (n = 4). Modified instructions, such as slower delivery or more guidance, were requested by n = 6 in the verbal instructions only group compared to n = 4 in the verbal instructions plus photographic aid group. More training and practice was recommended (n = 4 and n = 5 in the verbal instructions only and verbal instructions plus photographic aid groups respectively). Five participants in the verbal instructions only group would have liked a diagram or photo.

Discussion

This study demonstrates that when providing bystanders with telemedicine guidance, supplementing this with a photographic aid significantly improves their ability to accurately identify a safe site for IM TXA injection.

A strength of this study is its randomised design, which reduces the risk of bias and increases the reliability of the findings. The use of a standardised assessment process, in which injection sites were independently evaluated by three experienced practitioners, further strengthens the validity of the outcome measures. Additionally, the study used a quasi-realistic environment (different simulated patients sat in a car), enhancing external validity by mimicking some of the real-world conditions in which bystanders may be required to assist.

However, several limitations should be acknowledged. This was a simulated study, and as such participants were not under the same psychological pressures they might experience in an actual emergency [12]. Younger adults in higher education were overrepresented in our participant group and this threatens the external validity and generalisability of our results outside of a university campus environment. Participants had no prior medical training, but were willing to engage with the study which could risk self-selection bias, and this baseline level of confidence or interest in emergency care may not be generalisable to the wider population. While the use of expert panel review to determine correct site identification is robust there was poor concordance between these experts. This may reflect the diversity in the clinical backgrounds and potentially either different approaches or baseline training, therefore future studies may benefit from an initial training package for panel reviewers to improve concordance in the evaluation process. Furthermore, the study does not evaluate whether bystanders could safely administer an injection in a real-world scenario. Confidence intervals for the main outcome were wide; greater sample sizes and increasingly heterogenous participants may enhance the generalisability of findings in future studies.

Bystander interventions have been widely studied in other areas of emergency medicine, particularly in the context of out-of-hospital cardiac arrest, where early Cardiopulmonary Resuscitation (CPR) and defibrillation significantly improve survival rates [13]. Our findings align with this broader evidence base, indicating that with structured support, bystanders can perform critical emergency interventions with a high degree of accuracy. The observed improvement with photographic aids is consistent with research showing that multimodal instruction enhances procedural learning and execution [14].

The findings of this study have important implications for emergency medical response systems and policy development. As IM TXA emerges as a viable option for optimising outcomes in prehospital settings, integrating bystander-assisted administration into trauma response protocols prior to EMS arrival could significantly expand access to this life-saving intervention and maximise its efficacy by reducing the time from injury to administration; however, practical considerations remain, including the current lack of an auto-injector formulation and the need to ensure safe storage, handling, and training [9, 15, 16]. The demonstrated effectiveness of photographic aids suggests that emergency dispatch services could incorporate visual guidance into their protocols, potentially through mobile applications or SMS-based image delivery systems [17]. Furthermore, the study highlights the potential for expanding community-based first-aid training programmes to include haemorrhage control measures, reinforcing the role of the public in optimising the road injury chain of survival. While layperson administration of TXA remains exploratory, there are practical considerations that would need to be addressed before implementation. These include how TXA would be supplied, stored, and accessed by members of the public, as well as whether it would be carried by individuals or placed in publicly accessible locations (e.g. alongside AEDs). Although TXA is not currently available in an auto-injector format, future product development could help support this. The COVID-19 vaccination programme and naloxone distribution schemes demonstrate that with appropriate training and oversight, non-clinicians can safely deliver intramuscular injections in emergency contexts [18].

While this study is indicative that the use of photographic guidance in bystander interventions is beneficial to IM drug administration, several questions remain. Future research could explore whether these findings translate to real-world emergencies, where stress, time pressure, and environmental factors may impact performance. Additionally, studies could assess whether bystanders can safely and effectively administer TXA.

Another area for future investigation could be exploring the accuracy and usability of standalone visual and written instructions in scenarios where telephone or telemedicine guidance is unavailable (e.g. due to poor signal or remote location). Evaluating such approaches could help determine the feasibility of deploying TXA for use by untrained lay responders in diverse settings, including rural or remote areas within high-income countries where EMS response may be delayed, as well as in low-resource environments where formal prehospital systems are limited or absent. Further work may also be needed to assess the scalability of these interventions, including their integration into emergency dispatch systems and the practicality of wider public training programmes.

Conclusion

This study supports the integration of telemedicine-assisted bystander interventions for IM TXA administration following road injury. The addition of photographic guidance significantly improves the accuracy of IM injection site identification, providing a potential mechanism to enhance pre-hospital trauma care. Future work should focus on real-world application and implementation strategies to maximise the benefits of bystander-administered TXA in emergency settings.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

NCH

Non Compressible Haemorrhage

TXA

Tranexamic Acid

IM

Intramuscular

Author contributions

TN RF and EF contributed to the conception of the work. TN, EF, RF, and JTM contributed to the design and acquisition of data. All authors contributed to the analysis and interpretation of the data, participated in drafting the work, and approved the final submitted version. All authors agree to be personally accountable for their own contributions and to ensure the integrity and accuracy of the work.

Funding

IMPACT is funded by the Road Safety Trust (UK) and Vision Zero South West (UK). Funders did not have a role in the conceptualization, design, data collection, analysis, decision to publish, or preparation of the manuscript.

Data availability

The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

This study was conducted in accordance with the ethical principles of the Declaration of Helsinki. It involved a simulation-based design with no clinical interventions, no collection of personal health data, and no administration of treatment or collection of biological samples. The UK Health Research Authority’s Is This Research? decision tool (www.hra-decisiontools.org.uk), which is aligned with the UK Policy Framework for Health and Social Care Research (2017), was used to determine whether NHS Research Ethics Committee (REC) approval was required. Based on the information provided, the activity was deemed not to be research under the Framework and was therefore classified as a service evaluation and public engagement exercise. As such, it did not require NHS REC approval. All participants provided informed consent prior to participation. They were given written and verbal information about the study, advised that participation was voluntary, and informed they could withdraw at any time. No participants were under the age of 18.

Consent for publication

Not Applicable.

Competing interests

The authors declare no competing interests.