Emergency Stroke Calls: Obtaining Rapid Telephone Triage

Article history

The research reported in this issue of the journal was funded by PGfAR as project number RP-PG-0606-1066. The contractual start date was in August 2007. The final report began editorial review in April 2012 and was accepted for publication in May 2013. As the funder, the PGfAR programme agreed the research questions and study designs in advance with the investigators. The authors have been wholly responsible for all data collection, analysis and interpretation, and for writing up their work. The PGfAR editors and production house have tried to ensure the accuracy of the authors’ report and would like to thank the reviewers for their constructive comments on the final report document. However, they do not accept liability for damages or losses arising from material published in this report.

Declared competing interests of authors

Gary Ford has received honoraria for consultancy and educational activities (Lundbeck Ltd; Boehringer Ingelheim Ltd UK). Tom Quinn is a member of the steering group for the Strategic Reperfusion Early After Myocardial Infarction (STREAM) trial (Boehringer Ingelheim Ltd UK); Medicines Company for the European Ambulance Acute Coronary Syndrome Angiography (EUROMAX) trial and is a local collaborator/principal investigator for the 30 Day Study to Evaluate Efficacy and Safety of Pre-hospital vs. In-hospital Initiation of Ticagrelor Therapy in STEMI Patients Planned for Percutaneous Coronary Intervention (ATLANTIC) trial (AstraZeneca).

Copyright statement

© Queen’s Printer and Controller of HMSO 2014. This work was produced by Watkins et al. under the terms of a commissioning contract issued by the Secretary of State for Health. This issue may be freely reproduced for the purposes of private research and study and extracts (or indeed, the full report) may be included in professional journals provided that suitable acknowledgement is made and the reproduction is not associated with any form of advertising. Applications for commercial reproduction should be addressed to: NIHR Journals Library, National Institute for Health Research, Evaluation, Trials and Studies Coordinating Centre, Alpha House, University of Southampton Science Park, Southampton SO16 7NS, UK.

Background

Stroke is a leading cause of mortality and disability worldwide. 1 In England, 110,000 strokes and a further 20,000 transient ischaemic attacks (TIAs) occur every year. 4 Stroke is increasingly recognised as a medical emergency for which rapid access to specialist care reduces death and dependency4 through the delivery of thrombolysis and other medical, nursing and therapy interventions. 3

Rapid access to specialist care relies on a layperson’s recognition of symptoms and an understanding of the need for an emergency response. The time from symptom onset to the first point of contact for any help accounts for much of the pre-hospital delay. 25 Raising public awareness and encouraging people to contact the EMSs when they suspect a stroke is difficult. 26 A more modifiable aspect of pre-hospital delays may be to raise awareness of stroke among EMS staff, particularly as up to 70% of patients (or someone on their behalf) who seek help from a health professional make their first medical contact with the EMSs. 16–18

The interaction of EMDs with callers is likely to play a key role in the early identification of stroke and initiation of a rapid EMS response. Published reports on the ability of EMDs to recognise stroke from callers’ descriptions suggest that EMD sensitivity and PPV for identifying stroke are < 50%. 7–9 Therefore, improving the communication between EMDs and callers to increase the likelihood that stroke is recognised would promote the initiation of the pre-hospital pathways that are known to expedite patients’ access to specialist assessment and treatment. 27

There is evidence to suggest that the EMSs are more likely to be contacted by the public for suspected stroke if the person’s symptoms include speech problems, weakness, decreased level of consciousness and dizziness/loss of co-ordination. 28 There is less evidence of an association between the different types of stroke symptoms and the response of the EMSs.

We have undertaken a programme of work with the aim of increasing the ability of EMDs to identify stroke. The study described here is the first phase of this programme, in which we describe the associations between patient characteristics, stroke symptoms and diagnoses at different points on the stroke pathway.

Methods

Results

Seven hundred and thirty-five patients were included in this phase, of whom 359 (48.8%) were identified through hospital records ( Figure 1 ). The median age of the included patients was 78 years [interquartile range (IQR) 68–84 years]; 394 (53.6%) were male. Two hundred and forty (32.7%) patients had experienced a previous stroke and 178 (24.2%) patients had moderate to severe disability before the incident that led to the call to the EMSs ( Table 2 ). Six hundred and ninety-six (94.7%) patients had a final diagnosis of stroke and 166 (22.6%) patients died in hospital.

FIGURE 1.

Flow diagram of patients.

Discussion

This is the largest study reported to date to describe how the characteristics of suspected stroke patients are associated with a diagnosis of stroke at different points on the acute stroke pathway. We found that for people with a final diagnosis of stroke, the characteristics most consistently associated with a dispatch diagnosis of stroke were previous stroke, consciousness level, facial weakness, speech problems and lack of mental impairment. Therefore, if a caller describes any of these characteristics, there is an increased likelihood that the person has had a stroke. When patients are correctly dispatched as stroke, time from call to arrival at hospital is reduced.

The clear association between speech problems and a stroke diagnosis is consistent with other studies. Speech problems have been found to be one of the symptoms likely to prompt the public to contact the EMSs. 15,28 During calls to the EMSs for suspected stroke, speech problems have been the most frequently reported symptoms. 30 Similar to the findings of this phase, it has also been reported that speech problems were associated with a correct ambulance diagnosis of stroke. 20 Speech problems also affect the speed of response by the public: a rapid call (within 1 hour) to the EMSs was found to be more likely when speech problems were present. 20 The association between speech problems and stroke would clearly be important to include in any training package.

One of the other stroke symptoms we found to be associated with a dispatch diagnosis of stroke was facial weakness. This is consistent with the symptoms included in the FAST29 and Recognition of Stroke in the Emergency Room (ROSIER)31 assessments. Facial weakness has also been associated with unprompted stroke recognition by EMDs. 20

We found only a weak association between lower limb weakness and a dispatch diagnosis of stroke. This was unexpected because unilateral weakness is one of the main symptoms of stroke and lower limb weakness is included in the ROSIER assessment. Moreover, in other studies, the onset of limb weakness has been found to precipitate a call to the EMSs. 28 This finding may be explained by the way in which callers described symptoms to the EMDs. We have found previously that leg and arm weakness are rarely reported (< 5% of cases) as such during a call to the EMSs. 24 Instead, motor problems are often reported by their consequences, such as a person collapsing, falling to the floor or being found on the floor, or dropping or not being able to hold an object. 24

The patient and stroke characteristics most often associated with an ambulance diagnosis of stroke were facial weakness, speech problems, no mental impairment and either leg/foot or arm/hand weakness. This fits in with previous research, which found that ambulance personnel are more likely to identify total anterior circulation strokes,32 whose symptoms include speech problems and motor and/or sensory deficit of at least two areas of face, arm or leg. This finding could suggest that speech problems, facial and arm/hand weakness are key stroke symptoms; these symptoms are usually assessed using the FAST. This does not mean that leg and foot weakness are not useful symptoms when diagnosing stroke. The FAST was not always undertaken in the ambulance, and the reasons for this lack of assessment were not documented in the EMS records. However, the high proportion of FAST-positive results suggests that ambulance staff tended to perform the FAST when they suspected stroke. A dispatch diagnosis of stroke was associated with facial weakness, and yet both dispatch diagnosis of stroke and facial weakness were independently associated with an ambulance diagnosis of stroke. This gives further support to the value of facial weakness when diagnosing stroke over the telephone or in the ambulance.

The modelling of the ambulance diagnosis initially included only patient and stroke characteristics as independent variables (as for the dispatch diagnosis). However, because ambulance staff would have been aware of the dispatch diagnosis, it was felt that modelling should be extended to include dispatch diagnosis in the modelling of ambulance diagnosis. When dispatch diagnosis was included as an explanatory variable in the modelling of ambulance diagnosis, the ORs for speech problems and mental impairment were attenuated, resulting in the terms no longer being significant, independent, explanatory variables for ambulance diagnosis. Both facial and leg/foot weakness remained in the models, but with differing effects on their ORs. The ORs for facial weakness were attenuated slightly, whereas that for leg/foot weakness increased slightly. The dispatch diagnosis demonstrated associations with both facial and leg/foot weakness as well as speech problems and mental impairment. The association between these last two variables and dispatch diagnosis may explain why they were not included in all of the ambulance diagnosis models when dispatch diagnosis was included. This may also reflect differences around the reporting of characteristics. Speech and mental impairment may be more likely or more clearly reported over the telephone. This contrasts with weakness, which may be described by its consequences. 24 For example, someone with leg/foot weakness, will often describe a fall, which is a consequence of leg weakness rather than reporting the actual leg/foot weakness. Facial weakness may be described over the telephone but it may be more observable in person by, for example, ambulance staff, which may explain some attenuation of the ORs.

The inclusion of missing data as a valid category in the models attenuated the ORs for the presence or absence of the characteristic. The characteristics were still present in the final models, meaning that the interpretation of the association between the characteristics and diagnosis was not greatly affected by missing data. The OR for the missing values of any of the characteristics tended towards unity (compared with presence of the characteristic), suggesting that the ‘true’ values of the missing data are likely to be a mixture of the presence or absence of that characteristic. There were some exceptions to this pattern. When previous stroke was modelled with dispatch diagnosis the OR was 0.98, suggesting that the ‘true’ value for missing data for this characteristic is likely to be absence of this characteristic. If someone has had a previous stroke it is highly likely that this would be known, whereas if there was no knowledge of a previous stroke, this is likely to mean that a person has not had stroke. There was a similar pattern with the association between arm/hand weakness and ambulance diagnosis. It is possible that if this characteristic is absent it is more likely to be not recorded than if it was present. For facial weakness, the OR for the missing value category was smaller than unity, albeit non-significantly smaller; this also suggests that those with missing data were typically those without facial weakness, and that the reason data were missing was non-assessment, probably because ambulance personnel did not suspect stroke.

The sensitivity of the dispatch diagnosis was just under 50%, which is similar to that reported in other studies. 7–9 If this is to be increased, it is necessary to identify those symptoms likely to have a high sensitivity in relation to a final diagnosis of stroke. The characteristics with the highest sensitivities were having no to only slight pre-stroke disability, and being alert. The latter characteristic is consistent with reduced consciousness being scored as a negative sign for stroke in the ROSIER assessment. In terms of positive signs, the sensitivity of unilateral weakness in the arm/hand or leg/foot was over 60% and these signs should be assessed. The sensitivity of the other signs consistent with a diagnosis of stroke, facial weakness and speech problems, was only just over 50%. This contrasts with the results of the modelling in which facial weakness and speech problems were consistently associated with stroke diagnoses. This may indicate the need to consider multiple symptoms when diagnosing stroke.

In our sample, 29.5% were identified as having a suspected stroke by ambulance personnel, despite having no record of a positive FAST. A possible explanation for this is the lower sensitivity of FAST to identify posterior circulation strokes. If non-FAST symptoms such as diplopia or visual field defect were present, ambulance personnel might reasonably have suspected stroke despite the absence of a positive FAST.

It is known that EMD dispatch classification and prioritisation directly impact the speed of ambulance response and the level of medical care sent. If the EMD suspects a time-critical condition such as stroke, an ambulance can be dispatched as a high priority; this is currently category A and up to a 19-minute response. The results of this phase showed that the times from the EMS call through to arrival of the ambulance at hospital were significantly shorter (except for time from departure at scene to arrival at hospital) for those patients with a final diagnosis of stroke who were dispatched as stroke. The median time from call to arrival at hospital was 3 minutes shorter for those dispatched as stroke than for those dispatched as not stroke. This gain appeared to be mainly attributable to reduced time at the scene, which could be because of ambulance staff expecting to deal with a stroke. The CI indicated that this difference was likely to be between 1 and 5 minutes. This potential reduction of up to 5 minutes in the time for a stroke patient to reach hospital if correctly dispatched highlights the potential importance of an accurate dispatch diagnosis so that an ambulance is dispatched with an appropriate level of priority. This would therefore facilitate early presentation and rapid specialist treatment, particularly thrombolysis, the benefits of which are highly time dependent. 3

Our findings are therefore consistent with those of other studies, which have shown that activation of the EMSs is the single most important factor in the rapid triage and treatment of acute stroke patients. Stroke patients (or those who act on their behalf) who activate the EMSs arrive at EDs earlier and are more rapidly evaluated. 11–15

Limitations

The retrospective nature of this phase has given rise to some limitations. In terms of the sampling, it would have been ideal if we could have prospectively studied every person making a call to the EMSs and determined their subsequent diagnosis. Clearly this was impractical, so instead we designed a sampling strategy utilising data from EMSs, EDs, hospital records and stroke registers. This was intended to maximise capture of records of patients going to hospital via the EMSs and in whom there was a suspected or confirmed diagnosis of acute stroke. One disadvantage of this strategy is that our sample was biased towards confirmed stroke patients. Also, the diagnosis of stroke was based on information recorded in the case notes. Although there is no reason to suspect that a high proportion of cases had been misdiagnosed, the findings would have been more robust if there had been a mechanism in place for verification of a sample of the diagnoses by an independent examination. The retrospective audit of hospital and EMS records included patients arriving at hospital by ambulance during a 12-month period (1 October 2006 to 30 September 2007). During this time AMPDS version 11.3 was used by EMDs across England. Therefore, the findings from this phase may apply only to EMDs using this version of AMPDS. Finally, the symptoms used in the analyses of dispatch diagnoses were not necessarily what the callers had reported.

Logistic regression is one of a number of techniques that could have been applied to investigate how patient characteristics and symptoms were related to diagnosis. The approach used did not consider how these clustered together, so the resulting model might not have been optimal for predictive purposes. Consideration of the use of alternative classification approaches, such as classification and regression trees and neural networks, would be recommended if the aim were to develop an algorithm predicting diagnosis. However, the purpose of this phase was simply to investigate which patient and stroke characteristics are associated with dispatch code, accuracy of dispatch code and ambulance diagnosis, for which logistic regression was the most appropriate and simple approach.

This study is the first phase of a programme of work with the ultimate aim of improving recognition of stroke by EMDs. This phase has contributed to the programme by identifying those symptoms that are strongly associated with a diagnosis of stroke.

Conclusion

Of all the symptoms of stroke that may be described to EMDs, facial weakness and speech problems are highly likely to indicate that the caller is describing a stroke. Although individual symptoms can indicate a stroke, the likelihood of a stroke is higher when more than one symptom is present. These findings will be synthesised with the findings from subsequent phases and included in training described later in this report aimed at developing EMDs’ awareness and recognition of stroke symptoms.

Summary

The findings from phase 1 have suggested that facial weakness and speech problems are highly likely to indicate that the caller is describing a stroke. Although these problems were present, it is unknown whether or not they were communicated to the EMD. The communication between the caller and the EMD is crucial in identifying suspected stroke and minimising delays. 33 Although problems with communication have been previously identified,23 no studies have explored callers’ experience of making a 999 call at the onset of stroke symptoms. Therefore, the next phase (phase 2) will aim to identify the features that expedited or delayed people’s initial decision to contact EMSs at the onset of acute stroke, and to explore callers’ experiences of the call.

Background

In the UK, EMDs currently use the AMPDS to categorise ambulance response and decide on the level of medical care sent. The system is effective at ruling out acute stroke in people with other conditions. However, it is poor at correctly identifying acute stroke, with > 50% of people subsequently confirmed as having had a stroke being misclassified as presenting with another condition by the EMD. 9 During a 999 call, the communication that takes place between the caller and the emergency call handler is crucial in the identification of suspected stroke. Correct identification of acute stroke by the EMD minimises delays and improves patient outcomes. Problems with communication have been identified previously. 23 However, callers’ experiences of making a 999 call at the onset of stroke symptoms have not been previously explored. The purpose of this phase was to explore callers’ experiences of the 999 call including the factors that influence their decisions to contact EMSs at the onset of stroke.

Methods

Results

Table 9 outlines the characteristics of the callers who participated, their relationship to the patient, and action taken.

Over half of the participants called the EMSs immediately (see Table 9 ). Of the 20 who did not seek help from the EMSs immediately, initial help was sought from family or friends (n = 9), GP (n = 8), NHS Direct (n = 1) or by calling EMSs but after a delay of > 2 hours (n = 2).

Four central themes were identified. Two of these related to the initial decision to contact EMSs at the onset of stroke: perceived seriousness, and seeking and receiving lay or professional advice. Two themes related to communication between the caller and the call handler: description of stroke symptoms by the caller and emotional response to acute stroke symptoms. Verbatim quotes in the following section are followed by the caller’s relationship to the patient, and the caller’s gender and age.

Discussion

Thirty participants sought first medical contact from the EMSs, with the remainder initially contacting primary medical services or a relative or friend. Only one patient (2%) called for an ambulance themselves, consistent with previous studies. 21,22 Patients who were alone often contacted a family member, who then sought further help. 36 Nine participants (18%) contacted a family member or friend for help.

People’s emotional response to the onset of stroke symptoms was a theme that emerged during data collection, with 52% of callers describing feelings such as panic, nervousness and fear. Participants contacted primary care, family or friends in order to receive advice or support about the course of action that should be taken. This was a common cause of delay in contacting EMSs.

The efficiency of the call handler was found to be reassuring. Emotional support and practical advice were both important to callers. However, some callers were unsure whether or not an ambulance had actually been dispatched during the call.

Although 68% of the callers suspected stroke, only 54% of the sample reported this to the call handler, consistent with previous findings that stroke is reported as the presenting problem in 44% of ambulance calls leading to a final diagnosis of acute stroke. 20 Currently, AMPDS algorithms require the call handler to ask questions about ‘what has happened’ to the patient, rather than asking what the caller suspects may be the diagnosis. This may make it more difficult for callers to report their suspicion of stroke.

Previous research into decision-making processes in people with symptoms of myocardial infarction has identified that prior knowledge of symptoms alone is not enough to initiate prompt action. 37 A recent review of the public’s awareness of stroke found that although between 27% and 100% of participants stated that they would call the EMSs, only 18% had actually done so. 26 Emotional response and context appear to be influential in deciding to seek emergency help. 38

Limitations

This design of this phase does have some limitations. This phase did not set out to identify and measure time delays between onset of symptoms and making an EMS call, as it was intended instead to study the caller’s experience of the process of seeking emergency help for acute stroke. Although participants were interviewed soon after the event, the effects of recall bias may have influenced the findings. Participants were representative in terms of the age and sex distribution of people who call 999 for suspected stroke. However, it was particularly difficult to contact and recruit callers who were unrelated to the patient, and this group is likely to be under-represented. Furthermore, relatives or carers of critically ill or deceased patients were not approached to take part. Therefore, the most serious strokes were not included and are under-represented. This phase was concerned only with patients who accessed EMSs via a 999 call in response to stroke symptoms. Further work would be needed to examine the experiences of patients who accessed health care by other routes such as direct presentation to ED, accessing other primary care services, or who did not seek health advice.

Conclusion

This is the first study to examine in detail the factors that influence the initial decision to contact EMSs at the onset of stroke symptoms, and to explore the stroke-specific factors that facilitate or misdirect effective communication between the caller and EMSs. Many callers seek lay or professional advice prior to contacting EMSs, and some believe that the onset of acute stroke symptoms does not warrant an immediate 999 call. The findings also underline the need for ongoing public education to raise public awareness of stroke symptoms and the appropriate response, in order to reduce delays in accessing emergency treatment.

Summary

At the onset of stroke, the callers’ ability to convey their suspected diagnosis is limited by a lack of stroke symptom recognition, often as a result of the diverse presentation of stroke symptoms. Many callers believe that the onset of acute stroke symptoms does not necessarily warrant an immediate call to EMSs. Delays often arise because the caller feels it necessary to seek other lay or professional advice prior to calling EMSs.

Even when callers recognise stroke and call EMSs, often they do not convey this impression to the call handler. The form of structured questioning used by the call handler does not always enable the caller to convey their full impression of what the problem is, or allow the call handler to assure callers that an ambulance has been arranged. In order to improve outcomes in hyperacute stroke care, more public education is needed to improve awareness of a wider range of stroke symptoms and the need for an urgent response.

If members of the public can recognise stroke symptoms but do not always convey this to call handlers, how do they describe their symptoms? The next phase (phase 3) aims to identify ‘key indicator’ words used by people making emergency calls for suspected stroke and to compare these with the final diagnosis in hospital. It entails a retrospective review of digitally recorded EMS calls for patients who had a diagnosis of suspected stroke by the EMS call handler and/or a final diagnosis of stroke.

Background

Stroke is a time-dependent medical emergency in which rapid access to specialist care reduces death and dependency. 1 The interaction of EMDs with callers is potentially important in early identification of symptoms suggestive of stroke and the initiation of a rapid EMS response. However, published reports on EMDs’ accuracy in recognition of stroke from callers’ descriptions are very few. Identifying ‘true stroke’ from an EMS call is challenging. This can be partly attributed to callers’ lack of use of the term ‘stroke’ when stroke symptoms are present. A recent study reported that stroke was spontaneously identified by the caller in only 44% of EMS patients in whom the final ED diagnosis was stroke. 4 Similarly, EMD sensitivity and PPV for identifying stroke using AMPDS software have been estimated at < 50%. 7–9 Detailed investigation of the terms used during EMS calls for suspected or subsequently confirmed stroke has not been previously undertaken. We wished to explore the terms which were used by callers and what response these prompted from the EMD, in order to identify the stroke- and non-specific terms used and EMDs’ response to them. We planned to use these findings in order to better inform training and service development for emergency services, for suspected stroke patients.

The aim of this phase was to identify ‘key indicator’ words used by people making emergency calls for suspected stroke.

Methods

Results

Six hundred and forty-three calls were received by the EMSs where patients were either dispatched as stroke and/or had a final diagnosis of stroke, of which 592 (92.1%) had complete EMS and hospital data. From the 592 calls, 473 (79.9%) patients had a final medical diagnosis of stroke or TIA and 119 (20.1%) patients had non-stroke diagnoses.

Three hundred and sixty-six (62%) of the callers were female. The majority 323 (55%) of the callers were family members ( Table 10 ). Of these, daughters (30%) and wives (28%) were the most likely to call EMSs on behalf of the patient. Median duration of calls was 2 minutes 40 seconds.

Most patients (n = 425; 72%) were in their own home at the time of the call to the EMSs. The next prevalent locations were public places (n = 76; 13%), and nursing and residential care settings (n = 82; 14%).

The patients’ characteristics can be seen in Table 11 . The average age, gender distribution and levels of consciousness were similar in those with and without a final diagnosis of stroke.

Of the participants with a final diagnosis of stroke and where scan results were available, 85.9% had a cerebral infarction.

Discussion

There is evidence to suggest that the public are more likely to contact the EMSs for suspected stroke if their symptoms include speech problems, weakness, decreased level of consciousness, and dizziness/loss of co-ordination. 28

In this phase, the problem reported most frequently by callers to the EMSs was collapse or fall (39.9%). Motor problems were often reported by their consequences, which resulted in a person collapsing, falling to the floor, or being found on the floor. The terms fall or collapse have previously been reported by 17% and 21% of callers respectively. 20,22 The term stroke was used with similar frequency to collapse or fall, but is, at 37.2%, lower than the 45–51% reported in smaller series. 20,23 However, this phase is much larger than previous studies and focuses on the first reported problems between the caller and EMD when callers are most likely to report their initial interpretation of the problem. Following these first reported problems, the conversation becomes a series of closed questions as the EMD follows a specific AMPDS algorithm. Stroke was not mentioned in around 60% of the calls when the final diagnosis was stroke. This figure may not be too surprising given that the public is asked to describe what is wrong, which may lead them to state symptoms such as speech problems or limb weakness, rather than to provide a diagnosis.

We have found that the use of any of the terms ‘collapse or fall’, or ‘stroke’ by the caller tend to be associated with a final diagnosis of stroke, and have moderate sensitivity. These terms were also used by callers when the final diagnosis was not stroke. Stroke was mentioned nearly half as often in calls where the final diagnosis was not stroke compared with calls where the final diagnosis was stroke. For fall or collapse, a similar proportion of calls included mention of this term whether or not the final diagnosis was stroke. However, these data only provide part of the story around how terms mentioned in the call are indicative of final diagnosis. The sample in this phase had a diagnosis of stroke at either dispatch or as the final diagnosis. This means that we cannot draw any conclusions about the specificity of the terms. To properly gain an understanding of the specificity of these terms we needed to identify a sample of patients that have a non-stroke final diagnosis: this will be reported in Chapter 6 .

An important finding of this phase was that the common warning signs and symptoms of stroke (face weakness, limb weakness and speech disturbance) were stated in around 30% of cases. One of the FAST symptoms most frequently reported was speech problems. This finding is reflected in a recent Australian study, which found that speech problems were the most commonly reported FAST symptom by bystanders. 39 The same study39 found that facial weakness was reported in around 1% of cases, which is similar, although lower, to our finding of 4%. They found that arm weakness was reported in as few as 1% of cases in contrast to our findings where 14.5% stated arm weakness. However, this was still less than half of the number of cases that used the word stroke. The low numbers stating limb weakness may be due to callers having a tendency to describe the consequences of the weakness, such as being unable to grip or hold objects, rather than using such terms as arm weakness itself. These calls were made to the EMSs prior to the public FAST campaign. 40

Of calls where the EMS response was to dispatch as a stroke, the proportion of patients with a final medical diagnosis of stroke or TIA was 83%. EMDs used an AMPDS code of stroke in only 45% of all calls. Generic codes such as sick person and fall were often used, with the potential to delay appropriate assessment and treatment for stroke.

The callers to EMSs in our study were predominantly family members, the patient’s daughter being most likely, consistent with previously published Australian data. 20 The patient was rarely the caller (2% of all calls), consistent with other series12,20,41 and phase 2 of our research (reported earlier).

Patients with a final non-stroke-specific medical diagnosis were found to have both neurological and non-neurological conditions such as seizure, respiratory conditions and infections. Common stroke mimics such as seizure, syncope, space-occupying lesions, hypoglycaemia and sepsis have been reported elsewhere. 31

Pre-hospital notification of a patient with suspected stroke is known to be strongly associated with rapid admission to hospital,42 and is perhaps especially important in expediting assessment and delivery of thrombolytic therapy for eligible patients. It is possible that this opportunity is missed when non-stroke categories are used by EMDs.

Limitations

Detailed analysis of recorded calls to the EMSs has enabled us to study what was actually said by callers, rather than relying on abstracting data from patient records alone. However, this limited the number of data that we were able to collect about the caller, as we were only able to record those details discussed during the call, which were generally confined to the caller’s relationship to the patient. We also had to rely on the accuracy of the final medical diagnosis as recorded in the case notes and did not conduct any independent verification of this.

Conclusion

This is the largest study to look at the content of calls to the EMSs for suspected stroke. Within our sample, when callers contacted the EMSs and reported stroke as part of the first story they were correct in 85.5% of cases. This suggests that members of the public who use the word ‘stroke’ during the call are playing a vital role in initiating an immediate and appropriate response as outlined in the ‘stroke chain of survival’. 43

Summary

Detailed analysis of actual calls to the EMSs has enabled us to study what was said by callers during 999 calls for acute stroke. The problems reported most frequently by callers to the EMSs were collapse or fall, and stroke. Collapse or fall was most likely used by callers to describe motor problems, which resulted in a person collapsing, falling to the floor, or being found on the floor. When the term ‘stroke’ was used by the caller, the proportion of patients with a confirmed diagnosis of stroke or TIA was high. If stroke is suspected, members of the public should be encouraged to say the word stroke. However, findings of phase 2 of this programme suggest that even when callers suspect stroke, they may not always convey this to the EMD.

Within the sample selected, of those who said stroke, the majority of patients had a final diagnosis of stroke. However, it is not known if the word stroke is used by callers contacting the EMSs with other conditions. By listening to further calls in phase 4 we will identify any similarities or distinguishing key words that are said for other (non-stroke-related) conditions.

During analysis of the phase 3 EMS calls, it became apparent that misunderstanding of consciousness level was an important theme. Impaired consciousness level is an important, adverse prognostic sign in acute stroke, but is often also associated with stroke mimics such as epilepsy. Although not an intended phase of this programme of research, further investigation was thought to be necessary in order to explore this unexpected finding. We therefore undertook a secondary analysis of EMS recordings relating to discussion of consciousness level in patients with symptoms suggestive of acute stroke, which is reported in the next phase (phase 3a).

Background

Acute stroke is a medical emergency, in which timely access to emergency medical treatment, notably thrombolytic therapy, is vital to reduce mortality and morbidity. 3 However, symptoms of acute stroke are not always easy for bystanders to recognise. 26 Public information campaigns about common symptoms, such as the FAST, have been implemented in many countries including the UK, USA and Australia to expedite help-seeking behaviour. 39,44,45 Prompt access to emergency stroke services also depends on a bystander’s (usually not the patient themselves) identification and interpretation of the symptoms as requiring EMS response, callers’ ability to communicate with the EMS call handler about the patient’s symptoms, and the call handler’s questioning and listening skills. 33,46

Clinical presentation of acute stroke often includes an altered state of consciousness. This is an important sign of stroke severity and a prognostic indicator. Consciousness is defined as ‘the state of being aware of physical events or mental concepts. Conscious patients are awake and responsive to their surroundings’. 47

The question ‘Is he/she conscious?’ is routinely asked by EMS call handlers to determine the presence of life-threatening conditions, and is included as a standard question in computer-aided systems such as the AMPDS (v11.3) used by over 3000 agencies worldwide. Currently in the UK, EMS calls are prioritised as ‘red’ (attendance at the scene within 9 minutes), ‘amber’ (up to 19 minutes) or ‘green’ (up to 60 minutes). Assessment of the patient’s consciousness level is an important factor in correctly coding and prioritising calls to ensure the appropriate level of response. However, lay persons’ understanding of the term ‘conscious’ is known to be poor. 48 Although health-care professionals utilise tools such as the Glasgow Coma Scale (GCS), the Alert, Voice, Pain, Unconscious (AVPU) score and the National Institutes of Health Stroke Scale (NIHSS) stroke assessment framework, to assess patients’ responses to stimuli and thus identify patients with reduced or fluctuating levels of consciousness, these are not generally available to untrained bystanders and do not feature in media campaigns.

During analysis of the phase 3 EMS calls, the theme of misunderstanding of consciousness level emerged. Impaired consciousness level is an important adverse prognostic sign in acute stroke, but is often also associated with stroke mimics such as epilepsy and metabolic causes. Patients with an altered level of consciousness need to receive the highest priority of ambulance dispatch to enable urgent medical assessment, irrespective of the cause. We therefore undertook a secondary analysis of EMS recordings relating to discussion of consciousness level in patients with symptoms suggestive of acute stroke. The aim of this phase was to identify how patients’ consciousness level was questioned, described and interpreted by callers and EMDs during calls to the EMSs for suspected stroke.

Methods

Results

There were 592 calls in the data set, of which 109 (18%) patients had been noted to have an altered level of consciousness documented by attending EMS personnel. Of these, 44 (40%) calls had required further clarification of the patient’s level of consciousness by the call handler. Of these calls, 30 (68%) were dispatched as red, 5 (11%) as amber, 2 (5%) as green and no category of dispatch was recorded for 7 (16%). All 44 calls were made by a bystander (e.g. a family member), not by the patient themselves. Demographic information for patients and callers is summarised in Table 15 .

In the 44 calls included in this analysis, responses to the initial standard question ‘Is he/she conscious?’ were as follows: 14 callers replied that the patient was conscious, 5 callers replied that the patient was unconscious, 9 stated that they were ‘semi-conscious’ or used a similar term, and 16 callers were unable to state whether or not the patient was conscious.

In many calls there was apparent misunderstanding or clarification needed about the patient’s level of consciousness. This entailed unscripted or protracted dialogue between the caller and call handler. In response to an equivocal statement by the caller, the call handler would repeat, paraphrase or reverse the standard questions, usually until a ‘yes’ or ‘no’ response was received. Even then, the final response itself might be qualified by the caller with some additional information. An example is shown in Table 16 . In this case, the call handler repeatedly asks if the patient is ‘conscious’ or ‘awake’, but receives equivocal responses. The question is then reversed to ask ‘Is he unconscious?’, yet still fails to elicit a clear ‘yes/no’ response from the caller.

Questions asked by call handlers to clarify patients’ consciousness level included variations on the standard questions ‘Is he/she conscious?’ and ‘Is he/she completely awake?’ There were also variations on a further question, ‘Is he/she able to talk normally?’ which was prompted by the AMPDS algorithm if the call had been identified as suspected stroke by the call handler. This question was not prompted, for calls that were coded as another presenting problem such as fall.

Clarifying questions used by call handlers fell into four categories: consciousness level; level of awakeness, awareness or alertness; patient’s ability to talk and patient’s response to stimuli (usually verbal) ( Table 17 ).

In contrast to the semi-scripted line of questioning from the call handler, callers engaged in protracted and discursive descriptions of the patient’s consciousness level. Three themes were identified within the dialogues:

1. caller’s difficulty in determining consciousness level

2. miscommunication and need for clarification of stated consciousness level

3. association of consciousness level with breathing.

Discussion

This is the first study that has explored the public’s understanding and communication of consciousness level during EMS calls for patients with symptoms suggestive of acute stroke.

Emergency medical services calls are one example of ‘institutional talk’. 49 Such interactions are partly or wholly scripted, or follow additional rules to those in everyday conversation, in order to achieve a specific outcome. In the case of EMS calls, the intended outcome is the classification of the urgency and nature of the patient’s presenting problem in order to facilitate the provision of appropriate and timely emergency medical care. However, the highly structured format of the caller–call handler dialogue, and the nature of the medical emergency situation, may make it difficult for callers to ask for clarification of medical terms used by the call handler, and to convey uncertainty about their interpretation of the patients’ symptoms.

Despite poor public understanding, the term ‘conscious’ is used routinely in both the AMPDS system, and in the NHS Pathways programme which is being implemented across EMSs in the UK. Further to this, the call handler’s standard question ‘Is he/she conscious?’ implies that a dichotomous ‘yes/no’ answer is expected from the caller. This is unhelpful and potentially misleading in stroke, where more subtle changes in consciousness level may be present. Therefore, it is unsurprising that some callers make partial, equivocal and sometimes contradictory statements about the patient’s consciousness level. Call handlers are then obliged to resort to the use of rephrased questions and other terms such as awake, alert, responsive and speech/communication, in order to more fully elicit the patient’s level of consciousness. This can lead to protracted, time-consuming and often somewhat confusing dialogue as the caller attempts to convey the nuances of the patient’s level of consciousness.

The widely used GCS50 comprises three responses (motor, eye opening and speech) to assess consciousness level. Its use is problematic in stroke patients as those with aphasia may be unable to speak and yet may be fully conscious and alert. The NIHSS utilises a four-point scale which enables a more accurate delineation of the patient’s consciousness level in stroke. In contrast, the dichotomous nature of questioning about consciousness level in EMS protocols does not allow for recognition of gradations of consciousness level, other than a simple ‘yes/no’ response.

Many acute stroke patients experience symptoms of aphasia or cognitive impairment. 51 Such presentations may make it more difficult for a bystander to distinguish ‘true’ changes in consciousness level,52 and adds to the challenge of conveying the patient’s actual level of consciousness to the call handler. It can therefore be difficult for callers to swiftly and accurately assess the consciousness level of a patient with acute stroke symptoms and to communicate this information with the call handler.

In this phase, attending EMS personnel documented that 17% (109/643) of patients had an altered consciousness level, within the range of previously reported prevalence rates of 16–41% in stroke. 52 Although some patients’ consciousness level may have deteriorated or improved between the initial call and ambulance arrival, it is also possible that impaired consciousness level was under-reported during the emergency calls.

Although GP-initiated calls were excluded from this phase, we included calls from nursing and residential care home settings. Some of these did not provide an unequivocal description of the patient’s consciousness level, but tended to describe altered consciousness level using subjective terms such as ‘drowsy’ and ‘unresponsive’, rather than giving a ‘yes/no’ answer to the call handler.

Our findings suggest that, despite its wide use in medical practice and specifically in EMS dispatch protocols, the term ‘conscious’ is not widely understood and is poorly communicated by bystanders when making EMS calls about patients with symptoms suggestive of acute stroke. This phenomenon may lead to imprecise, contradictory and lengthy information exchange which may be time wasting, misleading and potentially dangerous. Our findings suggest that a structured assessment by ambulance dispatchers of consciousness level, with categories encompassing drowsiness and the separate assessment of speech and breathing problems might assist in obtaining a more accurate assessment of consciousness level in patients with suspected stroke.

Limitations

We were unable to verify patients’ actual level of consciousness at the time the EMS call was made. We used data recorded by attending EMS personnel on or shortly after arrival at the scene as a proxy for this. However, the patient’s consciousness level could have improved or deteriorated since the time the call was made. We were limited in the number of data we were able to collect about caller characteristics, as such information is not routinely obtained during EMS calls. Our findings were restricted to calls relating to patients with suspected or confirmed stroke, but there are, of course, many other emergency conditions that affect level of consciousness.

Conclusion

Routine questioning of EMS callers about a patient’s consciousness level is important to help prioritise the EMS response, but is characterised by ambiguities or contradictions in statements made by callers. When there is a reduced or fluctuating level of consciousness, or cognitive or speech impairment, a dichotomous ‘yes/no’ answer does not capture nuances of the patient’s condition. The closed nature of routine questioning does not necessarily elicit the necessary information and is therefore supplemented by both call handlers and callers with additional unscripted questions and colloquial responses. Although our study relates only to patients with acute stroke, our findings have implications for other EMS calls. It is possible that the issues described in this phase are also found in emergency calls about patients with other conditions associated with acute changes in consciousness level. Further research is needed to explore this, and to inform EMS call handler education and future public awareness campaigns. Further work is also needed with members of the public to identify which term or terms are best understood in conveying altered levels of consciousness in emergency situations.

Summary

This secondary analysis of EMS recordings relating to discussion of consciousness level was not an intended phase of this programme of research. This analysis revealed that although routine questioning of EMS callers about a patient’s consciousness level is important to help prioritise the EMS response, the discussion is characterised by ambiguities or contradictions in statements made by callers. Establishing a patient’s consciousness level during an EMS call is potentially more difficult in stroke patients with cognitive or speech impairment, where a dichotomous ‘yes/no’ answer does not capture the nuances of the patient’s condition. Further research would be required to explore if these findings have implications for patients with conditions other than stroke. The training package to be developed later in this programme of research will include issues around establishing consciousness level during EMS calls for patients with stroke.

In the previous phase (phase 3) the majority of callers to the EMSs who said ‘stroke’ had a confirmed final diagnosis of stroke in hospital. However, it is not known if the word stroke is used by callers contacting the EMSs with other conditions. By listening to further calls in the next phase (phase 4) we will identify any similarities or distinguishing key words that are said for other non-stroke conditions, illnesses and injuries.

Background

Stroke is increasingly recognised as a medical emergency for which rapid access to specialist care can reduce death and dependency. 4 Rapid access therefore has the potential to reduce severity of stroke, health services input and length of stay. In the longer term, the burden of stroke could be reduced for individuals, carers and society as a whole. It is therefore vitally important that the public and front-line staff such as EMSs are able to recognise the symptoms of suspected stroke and initiate a rapid response. People with suspected stroke should be taken immediately to hospital. Early presentation at hospital provides greater opportunity for time-dependent stroke treatment, such as thrombolysis. 53 Subsequently, patients will have more immediate access to organised stroke care, which is known to have a positive impact on survival and dependency. 54 Furthermore, early neurological attention is related to improved functional outcome and shorter hospitalisation. 55

One way of facilitating a patient’s transfer to hospital and therefore improving the chance for early presentation is through enhancing communication between the general public and the EMSs. Although problems with communication have been identified,23 no studies to date have explored the content of these verbal interactions. If we had a better understanding of how patients or those acting on their behalf interact with emergency services, this may lead to ways of improving emergency service utilisation among suspected stroke patients. 23 Furthermore, if we explore the interaction between people and the EMSs and identify how this communication can be enhanced, we can use this information to inform the content of the stroke-specific training provided for EMS personnel.

In a previous phase of this programme of research (phase 3) we listened to 999 calls and recorded the key indicator words said by people describing a suspected stroke. By listening to further calls we will identify any similarities or distinguishing key words that are said for non-stroke-related conditions, illnesses and injuries.

Methods

Sample and sampling

The calls between the public and the EMSs that led to an EMS vehicle being dispatched were identified over a 1-week period (8–14 March 2010). Figure 2 shows the flow diagram for the study.

FIGURE 2.

Flow diagram of inclusion and exclusion in study. Reproduced with permission from Leathley MJ, Jones SP, Gibson JME, Ford GA, McAdam JJ, Quinn T, et al. ‘Can you send an ambulance please?’: a comparison of callers’ requests for emergency medical dispatch in non-stroke and stroke calls [published online ahead of print 13 July 2013]. Emerg Med J 2014.

Inclusion criteria

The 999 call was made by a member of the public; patients arrived at hospital through EMSs; patients presented to the study hospital ED (whether admitted, or discharged after ED attendance).

Exclusion criteria

Calls relating to patients aged < 18 years. Calls which were likely to contain highly distressing or sensitive material were also excluded from the analysis; these included traumatic injury (i.e. laceration) (n = 70), poisoning (n = 40), physical or sexual assault-related injuries (n = 28), suicide (n = 27), road traffic accident (n = 24), choking (n = 4); women in labour (n = 3), and animal attack (n = 1). Calls were excluded if a medical practitioner had already seen the patient [i.e. interhospital transfers; patients referred by their primary care physician (GP) or GP staff on behalf of the patient) (n = 293)]. Sixty-six duplicate calls were excluded. For 22 people dispatched by the EMSs, no call data were available. Calls in which the final diagnosis was missing or the final diagnosis was stroke were also excluded.

Call data from phase 3 were also included in the analysis (a description of the sampling for these data can be found in Chapter 4 ).

Results

Analysis focused on the calls made to the EMSs for the 277 patients for whom we had complete call data and a final medical diagnosis. One hundred and seventy-three (62.5%) of the callers were female. Family members accounted for a large proportion of the callers (40.1%) ( Table 19 ). Of these, wives (10.8%), daughters (8.7%) and sons (8.3%) were the most likely to call EMSs on behalf of the patient.

Most patients (79.1%) were in their own home at the time of the call to the EMSs; others were in public places (10.8%) or in nursing and residential care settings (6.5%). The remainder were unknown (3.5%). Patient characteristics can be seen in Table 20 .

The frequency of the number of problems reported in the first stories can be seen in Table 21 . The modal number of problems was one, occurring in just under 30% of calls. Similarly, four or more problems were reported in around 30% of calls, with < 5% reporting six problems. In total, 738 problems were mentioned for the 277 patients.

Of the 277 calls that were identified, stroke or TIA were mentioned in eight calls (2.9%). One or more of the individual FAST items were mentioned in 24 calls (8.7%). The most often reported FAST items were limb weakness in 12 (4.3%) calls, speech problems in 11 (4.0%) calls, with only one (0.4%) call including facial weakness. None of the calls included mention more than one of the FAST items.

We explored the relationship between the most common problems mentioned in the first story from phase 3 and whether the final diagnosis in hospital was stroke or not stroke ( Table 22 ). Within the table, the data pertaining to calls with a final diagnosis of stroke come from phase 3, and the data pertaining to calls with a non-stroke final diagnosis come from this phase (phase 4).

Table 22 shows that when the final diagnosis was stroke, nearly 40% of the callers mentioned stroke. In contrast, when the final diagnosis was not stroke, only 2.9% of callers mentioned stroke. The mention of fall or collapse was common in calls where the final diagnosis was stroke (around 40%), and it was also used in one-quarter of calls where the final diagnosis was not stroke. Both speech problems and limb weakness were mentioned in around 15% of calls where the final diagnosis was stroke, whereas these two problems were each mentioned in < 5% of calls relating to a non-stroke diagnosis. Consciousness level was mentioned a similar number of times for final diagnosis stroke or not stroke, as was mention of being ‘generally unwell’. Facial weakness was mentioned in < 4% of calls relating to a final diagnosis of stroke, but was only mentioned once in calls where there was a non-stroke final diagnosis. Describing more than one of the FAST items occurred in only 3% of calls where the final diagnosis was stroke, but where the final diagnosis was not stroke, none of the callers mentioned more than one FAST item.

Because of the sampling strategy adopted in phases 3 and 4 we are only able to describe for the word ‘stroke’ the sensitivity from phase 3 and the specificity from phase 4 data; these data can be extracted from Table 22 . In phase 3, the sensitivity of ‘stroke’ was 39.7% and in phase 4 specificity is 97.1%. The sensitivity of the individual FAST items in identifying stroke was low for speech (15.2%), limb weakness (14.8%) and very low for facial weakness (3.6%). However, the specificity of the individual items was very high: 96%, 95.7% and 99.6% for speech problems, limb weakness and facial weakness respectively. If more than one item of the FAST was used, this had low sensitivity (3.0%) but 100% specificity.

Discussion

Detailed analysis of actual calls to the EMSs has enabled us to study what was said by callers during stroke and non-stroke calls, rather than relying on abstracting data from patient records alone. The callers to EMSs in this phase were more often bystanders, the police or through buzzer helplines, compared with the stroke calls, where family members were the most likely people to call the EMSs. 20,46 In the non-stroke calls the patient was also more likely to be the caller (12.6%) than in calls for patients with suspected stroke (2.1%). The non-stroke calls tended to be based on patients who were younger: the average age of the non-stroke patients was 65 years, compared with the 77 years for stroke patients. However, around one-quarter of all stroke patients are aged < 65 years, so this is not a definitive marker of stroke or not.

The data from this phase have shown that the word ‘stroke’ is rarely used by callers when describing a problem where the final diagnosis is not stroke. In phase 3 it was identified that use of the word ‘stroke’ was associated with a final diagnosis of stroke in around 40% of cases. Taken together, these findings suggest that if a caller mentions the word ‘stroke’ during the first story, there is some probability that they are describing a stroke, but if they do not use the word ‘stroke’, there is a very high probability that they are describing something other than stroke.

The individual items of the FAST show similar patterns in relation to how frequently they are used in stroke compared with non-stroke calls. Both speech problems and limb weakness were reported in < 5% of the calls where the final diagnosis was not stroke, but were reported in around 15% of calls where the final diagnosis was stroke. Although facial weakness was mentioned in < 4% of calls where the final diagnosis was stroke, it was mentioned in well under 1% of the calls where the final diagnosis was not stroke. This suggests that absence of facial weakness may be useful to indicate a non-stroke diagnosis. In addition, although more than one item of the FAST was mentioned in only a small number of cases of stroke, multiple items were never mentioned when the diagnosis was not stroke. Similar to the absence of facial weakness, a lack of multiple FAST items may be useful to indicate a non-stroke diagnosis.

Describing a collapse or fall occurred in 26% of the non-stroke calls; this was less than in the stroke calls (38.5%). These are much higher values than reported elsewhere for terms such as ‘collapse’ or ‘fall’, which have previously been reported as presenting stroke symptoms by 17% and 21% of callers respectively. 20,22 Despite the common occurrence of ‘fall’ as the key presenting symptom in calls where stroke is subsequently diagnosed,46 it has little discriminatory value if as many as one-quarter of the non-stroke calls also mention collapse or fall. It is, however, important that EMDs are able to distinguish between ‘mechanical’ falls where there is a clear external cause, and fall/collapse where an underlying medical condition has led to the presentation as ‘fall’. The AMPDS algorithm for ‘fall’ includes the question ‘What caused the fall?’, which could be held to imply that a mechanical, rather than medical, cause is the precipitating factor. In the non-stroke calls, despite a high proportion of callers reporting a fall, only 2% had a final diagnosis of fall in hospital.

Limitations

The samples of calls for phases 3 and 4 were based within the catchment area of the same regional ambulance trust, but utilised different acute hospital trusts, and also covered different time periods. This means that the populations studied in these phases are not directly comparable, but they do allow separate estimates of sensitivity (phase 3) and specificity (phase 4). The nature of the sampling in phase 3 meant that the stroke population was over sampled, which meant that it was harder to explore the relationship between the words used and the final diagnosis when the final diagnosis was non-stroke. In order to address this, in phase 4 we identified all calls to the EMSs over a 1-week period, thereby allowing us to identify all the calls related to a non-stroke final diagnosis and subsequently explore the words used by the callers in such calls. Within the calls identified, only six had a final diagnosis of stroke, demonstrating that stroke accounts for < 1% of calls to the EMSs. This sample of all calls could have allowed us to more fully explore sensitivity, specificity, PPV and negative predictive value; however, the small sample of true stroke patients would have led to very imprecise estimates of these values. Rather than over analyse or over interpret these data it was felt that a more cautious approach should be taken and so we only analysed calls associated with a non-stroke final diagnosis.

There were very limited data available about the callers as we were only able to record details discussed during the call itself. This generally included the caller’s relationship to the patient. Callers’ genders were attributed according to relationship with the patient, or voice, if it was not apparent from the content of the call. It was not possible to identify the callers’ ages. Patients’ final medical diagnoses were taken from the case notes, but there was no mechanism to verify these independently.

A large proportion of the calls to the EMSs were excluded from the analysis, but it is unlikely that the problems described for these excluded cases are likely to overlap to a great extent with the problems we identified (e.g. calls relating to women in labour or assault injury). In addition, the context of many of the excluded calls is likely to indicate that the problem is due to extrinsic rather than intrinsic factors, such as road traffic accident.

Conclusion

The non-stroke calls were more likely to be made by someone other than the patient themselves and the patients tended to be younger. Less than 5% of callers in the non-stroke calls mentioned either ‘stroke’ or any of the FAST items and none of the calls included more than one FAST item. This compares with the mention of ‘stroke’ in around 40% and FAST items in around 15% of the stroke calls. Mention of ‘stroke’ or ‘FAST’ items in the first story was much more common in stroke than in non-stroke calls and not mentioning these terms occurred frequently in the non-stroke calls. Fall was a common presenting complaint in both the stroke and non-stroke calls, meaning that whether or not it is mentioned is not indicative of a stroke or non-stroke diagnosis.

Summary

In previous phases of this programme of work we identified that the majority of calls made to the EMSs on behalf of someone with suspected stroke were made by a female relative: stroke patients rarely made the EMS call themselves. Callers tended to talk in terms of the consequences of loss of function (e.g. unable to grip, cannot stand) rather than describing symptoms themselves, such as weakness. Nevertheless, findings from the earlier phases indicate that if the patient has, or the caller mentions, one or more of facial weakness, limb weakness, or speech problems (or even uses the word ‘stroke’), there is an increased probability that the person has had a stroke. This phase has added to these data by showing that callers very rarely mention the word ‘stroke’ or any of the FAST items for non-stroke calls. This programme of work is therefore building a picture of the relationship between problems described, or not, and whether the call relates to a stroke or non-stroke diagnosis. There is no one definitive indicator word for a diagnosis of stroke, but a call made by a third party, describing an older patient and mentioning the word ‘stroke’ or one or more of limb or facial weakness, or speech problems has a higher probability of being associated with a final diagnosis of stroke. In contrast, a call made by the patient, who is of a younger age and does not mention the word ‘stroke’, or describe limb or facial weakness, or speech problems has a high probability of having a diagnosis of not stroke. The findings from phases 1–4 of this programme of research will be used in phases 5 and 6 to inform the development of algorithms, protocols and a training package aimed at improving recognition of stroke. Although the aim of the next phase (phase 5) was to develop algorithms and protocols for EMSs and NHS Direct staff to assist in the identification of those with suspected stroke, this was not possible, reasons for this will be now be explained further.

Background

In the UK, people seeking urgent medical assistance call a universal number (999) and are connected to the EMS dispatch centre in closest proximity. All calls to the EMSs in the UK are digitally recorded for training and governance purposes. Once the call is logged and patient location established, an EMD will dispatch the closest ambulance. Calls to EMSs are triaged using AMPDS,57 an algorithm-based system which is also used widely in Europe and North America. In 2007, the AMPDS system was used by more than 3000 public safety agencies in over 23 countries. 58 The design and publishing licence for AMPDS is owned by Priority Dispatch Corp. EMD classification within AMPDS and prioritisation directly impact the speed of ambulance response and the level of medical care (e.g. paramedic) sent.

The AMPDS protocol is designed to standardise and codify the operation of EMDs while optimising safe and effective patient care through dispatch life support and EMS system response. AMPDS provides EMDs with software that supports decision making during a 999 call. The EMD firstly establishes the patient’s address and contact details and then asks specific questions about the patient’s condition. EMDs must follow the questions that are scripted within AMPDS. AMPDS also provides EMDs with information about specific conditions/problems, as well as online pre-arrival and post-dispatch instructions, appropriate response and referral recommendations. The aim of this phase of the programme was to develop algorithms and protocols for ambulance and NHS Direct staff, informed by earlier phases to assist in the identification of those with suspected stroke.

Methods

Results

Discussion

It was not possible to change the AMPDS algorithm for stroke within the ambulance service or NHS Direct, based on the findings from ESCORTT. Further evidence would be required as AMPDS is based on published standards from a wide range of international institutions including the NAEMSP, the ASTM, the ACEP, the USDOT, the NIH and the AMA. 58

Similarly, NHS Direct algorithms are aligned to DoH, NHS and National Institute for Health and Care Excellence (NICE) policies and are reviewed annually. Too few calls relating to patients who had contacted NHS Direct for suspected stroke could be identified to suggest any changes to the NHS Direct algorithm for stroke.

A new system, NHS Pathways, has recently been developed with the intention of replacing AMPDS. This development took place during the same period as the ESCORTT programme of research. NHS Pathways is a suite of clinical content assessment for triaging telephone calls from the public, based on the symptoms they report when they call. NHS Pathways sets out to deliver a single clinical assessment tool that provides effective triage over the telephone in any setting which takes calls from the public. This includes calls to 999, NHS Direct, GP out-of-hours, NHS 111 and any other Single Point of Access number in place. 60 NHS Pathways was developed and is maintained by a group of NHS clinicians with extensive experience of urgent and emergency care provision, and clinical decision support tools. The system is currently being used by five ambulance trusts in the UK and many ambulance trusts have expressed an interest in using NHS Pathways as their primary clinical assessment tool for incoming 999 calls. The switch over to NHS Pathways from AMPDS will require extensive training of EMDs and changes to information technology (IT) systems, and so the roll-out of NHS Pathways has taken longer than expected. The Ambulance Trust within which the ESCORTT programme of research took place had a service delivery plan in place to move from AMPDS to the NHS Pathways system in 2011. However, at the time of writing this process is yet to take place.

Limitations

Although a specific change methodology approach was not utilised at the time of the study, the steps taken to identify and explore a process of revising current algorithms and scripted questions with the EMSs and NHS Direct have been mapped to Kotter’s eight-step change model. 59 It is not envisaged that using an alternative change model would have provided any additional benefits or would have better facilitated the change process.

Conclusion

It was not possible to change the AMPDS algorithm for stroke based purely on the findings from ESCORTT, and within Ambulance Trusts across the UK AMPDS is to be replaced by NHS Pathways. Future research could evaluate the impact of NHS Pathways in terms of dispatch/prioritisation for suspected stroke, as little is known about the impact of NHS Pathways.

Summary

The aim of this phase of the programme was to develop algorithms and protocols for ambulance and NHS Direct staff, informed by earlier phases, to assist in the identification of those with suspected stroke. It was not possible to change the AMPDS algorithm for stroke within the ambulance service or NHS Direct, based on the findings from ESCORTT. AMPDS is gradually being replaced by NHS Pathways. The findings of the ESCORTT programme may contribute to the development of effective stroke algorithms in this new system.

Background

Stroke outcomes can be improved by timely care. It is therefore vital that EMS personnel are able to recognise the symptoms of suspected stroke and initiate a rapid response. People with suspected stroke should be taken immediately to a hospital ED which offers hyperacute stroke care. Early presentation at hospital provides greater opportunity for time-dependent treatments, such as thrombolysis, and for early admission to specialist stroke care. 3

A training package will need to be developed for use by EMDs informed by the results of phases 1–4. The aim of the training package will be to improve education and increase awareness of staff in recognising the symptoms of stroke and the importance of treating stroke as a medical emergency. The training package will consider the ability of EMDs to identify key words and phrases that might indicate acute stroke. Development of the package will include input from the EMSs, other health-care professionals, academics, and users and carers. The course will be accessible via the internet, completed and evaluated online.

Online delivery allows flexibility of access by allowing participants to collapse time and space. 61 However, online materials must be appropriately designed to promote learning and engage the learner. 62 This training package was intended to complement an existing training package for EMSs – Rapid Emergency Stroke Pathways: OrgaNised Systems and Education (RESPONSE). The RESPONSE course is an interactive online acute stroke course that has been completed by over 2000 health professionals worldwide and is endorsed by the UK Stroke Forum Education and Training.

Methods

Results

The expert committee comprised four academics, six health professionals and two stroke survivors. The committee reviewed the content of the existing RESPONSE acute stroke online course and identified sections that could also be incorporated into the training package. These sections included:

• risk factors for stroke

• ‘what is a stroke?’

• different types of stroke

• suspected stroke and dialling 999

• the signs and symptoms of stroke

• TIA

• other signs and symptoms

• stroke mimics

• thrombolysis.

The expert committee also felt that it was important to include general information about ESCORTT and the purpose of the research programme.

The findings from phases 1–4 were used to inform the sections of the training package. Table 23 gives details of how and where the findings of each phase of the ESCORTT programme were incorporated into the training package. The full training package content can be found in Appendix 2 .

Once the initial training package content had been developed this was circulated to the ESCORTT steering group for members to review. The training package was also discussed at project steering groups.

Members of the expert committee with specific expertise in developing online educational materials emphasised the importance of developing interactive activities that would aid learning. The course therefore also included interactive animations that allowed the EMDs to alter physiological parameters in order to visualise their effects on the post-stroke brain ( Figure 3 ), and a multiple choice quiz ( Figure 4 ).

FIGURE 3.

An interactive animation showing perfusion and oxygen saturation.

FIGURE 4.

Multiple-choice quiz.

Interactive activities also included an online card-sort game in which EMDs had to put cards onto two piles according to the words they thought that the public used to describe the term ‘consciousness’ ( Figure 5 ).

FIGURE 5.

Interactive card-sort game for words that the public use to describe the term ‘consciousness’.

The course also included audio simulations of 999 calls in which miscommunication or poor stroke symptom recognition had resulted in incorrect categorisation, to add context and impact. These calls were based on transcripts of actual calls from phase 3. They were anonymised, with all identifying material removed, before being audio-recorded by members of the research team.

Discussion

This is the first stroke-specific training package to be developed for EMDs with the aim of improving their recognition of stroke. The training package includes stroke-specific information relating to risk factors; what a stroke is; different types of stroke; suspected stroke and dialling 999; TIA; stroke mimics and thrombolysis.

Underpinned by the findings from phases 1–4 of the ESCORTT programme of research, the training package also provides detailed information about what influences people in their initial decision to contact the EMSs at the onset of stroke symptoms; how the public describe the symptoms of stroke during a 999 call; the proportion of patients (or someone on their behalf) who obtain first medical contact from the EMSs for suspected stroke; who is most likely to dial 999 for suspected stroke; important communication issues between EMDs and callers; how often suspected stroke is confirmed as a stroke by ambulance crews and in hospital; details of the words that are used to describe non-stroke-related conditions; and the public’s understanding of different levels of ‘consciousness’ and how these are described during 999 calls.

A range of interactive activities have also been developed to assist learning including interactive animations; a multiple-choice quiz; card-sort activity and re-enactments of anonymised 999 calls.

Based on the previous success of the RESPONSE acute stroke online course,63 the training package was developed in a similar format for delivery online. Online delivery allows flexibility of access61 and the development of a range of interactive activities (e.g. listening to re-enactments of 999 calls and deciding which dispatch code to assign to that call). Interactive activities are designed to promote learning through enabling the learner to engage with the training material. However, it was acknowledged that online learning may be new to some and might not suit the learning style of others. The EMD evaluation of the training package, including its delivery mode and content, will be explored in the next phase (phase 7).

Conclusion

This is the first stroke-specific training package to be developed for EMDs, and is underpinned by the findings from each phase of the ESCORTT programme of research. A range of interactive activities have also been developed to assist learning including interactive animations; a multiple-choice quiz; card-sort activity; and re-enactments of anonymised 999 calls. The development of a stroke-specific training package will provide an exceptional learning opportunity for EMDs in improving their awareness of the key words and phrases that can indicate acute stroke during calls to the EMSs. Online delivery promotes flexibility of access by allowing participants to complete the training at the times most convenient for them.

Summary

The next phase (phase 7) of this programme of research involved the implementation of a training package in one EMD control centre. The next phase will outline an evaluation of EMDs’ satisfaction with the training package, including the method of assessment, usability, level of content, self-perception of increased knowledge and understanding of emergency stroke issues, and any changes that they felt should be made to the training package.

Background

It is important that stroke is recognised at the earliest opportunity to ensure that an ambulance is dispatched with an appropriate level of priority, therefore facilitating early presentation and rapid specialist treatment. Informed by the results from phases 1–4 a training package was developed for use by EMDs. The findings from phases 1–4 were used to inform the sections of the training package that specifically relate to:

• How often is suspected stroke confirmed as a stroke by ambulance crews and in hospital? (Phase 1.)

• What influences the public’s initial decision to contact EMSs at the onset of stroke symptoms? (Phase 2.)

• Communication between the EMD and caller. (Phase 2.)

• Suspected stroke – describing the symptoms. (Phase 3.)

• The proportion of callers who obtain first medical contact from the EMSs for suspected stroke. (Phase 3.)

• Who is most likely to dial 999 for suspected stroke? (Phase 3.)

• How stroke symptoms may be described by the public? (Phase 3.)

• Callers’ understanding the term ‘conscious’. (Phase 3a.)

• Do non-stroke callers mention ‘stroke’ and/or the FAST symptoms? (Phase 4.)

Phase 7 explores the implementation of the training package in one EMD control centre. It outlines an evaluation of EMDs’ satisfaction with the training package overall, the method of assessment, usability and level of content. EMDs were also asked about any increases in knowledge of the anatomy and physiology in relation to stroke, knowledge of the signs and symptoms of stroke, any increases in understanding that stroke should be treated as a medical emergency, and about any changes that they felt should be made to the training package. The impact of the training package on the recognition of stroke by EMDs will be evaluated in phase 8.

Methods

Results

The training package was undertaken by two educational and training managers who in turn rolled it out to 67 call handlers, 76% of whom were female.

Discussion

The training package was successfully undertaken and completed by EMD call handlers in one EMD control centre. The majority of respondents found the training package to be informative, generally raising knowledge of stroke, its symptoms and the importance of stroke as a medical emergency. To our knowledge, no previous studies have evaluated stroke-specific EMD training with the aim of improving EMD recognition of stroke. Although the majority of participants reported an increase in knowledge of stroke symptoms, the management of acute stroke and stroke as a medical emergency, we did not formally test knowledge before the training package was completed and so this is a subjective outcome.

In health care, as in many other educational fields, there have been recent advances in the use of technology as a tool for facilitating student learning, particularly for those accessing courses from the practice setting. 65 There is also a demand for continuing professional development and updating of knowledge and skills. 65 As in previous studies66 this evaluation has highlighted a demand for stroke-specific training. The training package takes only 2 hours to complete and is free to access, both features known to facilitate the uptake of learning opportunities. 66 The training package can be accessed via any computer with internet access, at home or work. Staff reported that they had enjoyed completing the course, but also stressed the importance of accessibility of suitable training opportunities in relation to their continuing professional development.

Limitations

The training package was implemented and evaluated in only one EMD control centre and so the generalisability of the findings of this phase is limited. Although the uptake of the training package has been encouraging, e-learning is an approach that does not suit everybody’s learning style or technological abilities. Other barriers to the use of e-learning may also include a lack of personal discipline65 and out-of-date computer equipment which may be inadequate for the use of some elements of an e-learning training package, such as the use of interactive animations.

Emergency medical dispatchers were asked to rate their own improvements in stroke knowledge and so this is a subjective outcome. Stroke knowledge was assessed only following implementation of the training package and so no baseline comparisons can be made.

Conclusion

We believe that this is the first study to (develop and) evaluate stroke training for EMDs with the aim of improving recognition of stroke. We have also shown that online learning for EMDs is well received and can increase self-reported stroke knowledge and that EMDs welcome the opportunity for continuing professional development. Further information about the course can be found at http://ukfst.org/courses/50/.

Summary

Following the successful development and implementation of a stroke-specific training package for EMDs, the impact of this newly-developed, stroke-specific, online training package on the recognition of stroke by EMDs will be evaluated in phase 8. In phase 7, two trainers and 67 EMDs completed the training package. Phase 8 will evaluate the accuracy of dispatch diagnosis of EMDs using an interrupted time series design.

Background

Stroke is a leading cause of mortality and disability worldwide1 and is increasingly recognised as a time-dependent medical emergency in which early presentation to specialist care reduces death and dependency. 3

Up to 70% of all stroke patients obtain first medical contact from the EMSs. 3,7,24 Calls to the EMSs are triaged using AMPDS,8 a system also used widely in Europe and North America. EMDs use this system to categorise ambulance response and decide on the level of medical care sent. If the EMD suspects a time critical condition such as stroke, an ambulance can be dispatched as a high priority (category A, currently up to a 19-minute response). In the UK the categories for response prioritisation are pre-determined by the DoH.

Identifying ‘true stroke’ from an EMS call is challenging. Although AMPDS is effective at ruling out acute stroke in people with other conditions, it is poor at correctly identifying acute stroke, with > 50% of strokes being misclassified. 9–11 It is important that stroke is recognised at the earliest opportunity to ensure that an ambulance is dispatched with an appropriate level of priority, therefore facilitating early presentation and rapid specialist treatment.

One way of facilitating rapid EMS transport to hospital, thereby improving the chance for early presentation is through enhancing communication between the general public and the EMSs. As part of a programme of research, we explored the interaction between the public and the EMSs during emergency calls for stroke, in order to inform the content of stroke-specific, online training for EMDs; this included exploring callers’ experiences12 and identifying the key words used by the public, to describe and that are indicative of stroke. 13 The aim of this study was to evaluate the impact of this newly-developed, stroke-specific, online training package on the recognition of stroke by EMDs, and on the impact of stroke recognition on the time between the call to EMSs and stroke patients reaching hospital.

Methods

Data analysis

Patient demographics, stroke characteristics and diagnosis data are presented overall and for each period. Time series plots are presented to illustrate trends for each outcome.

The effect of the intervention on:

1. accuracy of dispatch diagnosis was investigated using binomial (grouped) logistic regression, with the number of subjects with a dispatch diagnosis of stroke as the numerator and the number of subjects with a final diagnosis of stroke as denominator for each observation

2. call to arrival at scene and call to arrival at hospital were investigated using linear modelling, with observations weighted by that week’s number admitted with a final diagnosis of stroke.

Analysis was performed using complete cases (i.e. those with data available for both dispatch and final diagnosis). It was suspected that, given the non-contiguous nature of the observation periods, serial autocorrelation would be absent or weak. However, potential autocorrelation [due to the time of the weekly diagnosis rates and clustering (overdispersion) of the accuracy within observations] was investigated; standard errors would be adjusted for any observed lack of independence or overdispersion.

Regression models included the intervention factor ‘pre’, ‘during’, ‘post’: this segments the regression model and allows a ‘jump’ in outcome on transition from one period to the next. An overall linear trend over the period of data collection and an interaction between the intervention factor and the linear trend (to allow the intervention to influence any underlying trend and to allow a gradual impact of the introduction of the intervention) were also included. Where there was no evidence of either an overall trend or difference in trend between periods (p > 0.15), the corresponding term was removed. Findings from the more parsimonious model are presented. Sensitivity analysis was performed to assess the potential impact of missing dispatch and/or final diagnosis by imputing possible diagnoses, including extreme imputations (dispatch diagnosis as ‘not stroke’ and final diagnosis as ‘stroke’, and vice-versa). For the time intervals, sensitivity analyses was performed by repeating the modelling on geometric rather than arithmetic means for each observation (to reduce the potential influence of outlying times). Analysis was performed using SPSS (versions 19 and 20) and Intercooled Stata (version 11.0; StataCorp LP, College Station, TX, USA). Unless otherwise stated, inferential analyses used a 5% significance level; 95% CIs are presented.

Results

Over the 26 observation weeks, 464 patients met the study’s inclusion criteria. Sixty-six patients were included due to a final diagnosis of stroke only, 251 patients were included due to having a stroke dispatch code only, and 147 patients met both these inclusion criteria. Their median (IQR) age was 75 (62–83) years; 241 (51.9%) were female. Data were collected on 174 (mean 19.3 per week), 116 (mean 16.6 per week) and 174 (mean 17.4 per week) patients over the pre-, during- and post-implementation periods respectively. Dispatch data were available for 450 patients and, of these, 398 (88.4%) were dispatched as stroke. A final confirmed diagnosis was recorded for 424 patients and, of these, 213 (50.2%) had a final diagnosis of stroke ( Table 25 ).

The characteristics of the stroke patients are shown in Table 26 . Of the 213 patients with a final diagnosis of stroke, dispatch data were available for 199, of whom 147 (73.9%) had been dispatched as stroke.

For the patients with a final diagnosis of stroke, the proportions dispatched as ‘stroke’ or ‘not stroke’ (dispatch accuracy) for each of the three periods, and for each of the 26 observation weeks are given in Table 27 .

Trends in the number dispatched as stroke and the number dispatched as not stroke are illustrated in Figure 6 and the corresponding diagnostic accuracy is shown in Figure 7 . Logistic regression showed no significant change in trend in diagnostic accuracy across the observation weeks (p = 0.18), nor an overall trend in diagnostic accuracy across the observation weeks (p = 0.85). However, a significant change in dispatch diagnosis accuracy between periods was detected (p = 0.003), reflecting an improvement in dispatch accuracy relative to the pre-implementation period for both the during-implementation (OR 4.10, 95% CI 1.58 to 10.66) and post-implementation (OR 2.30, 95% CI 1.07 to 4.92) periods; the difference in dispatch accuracy between during- and post-implementation periods was not significant (p = 0.29). There was no evidence of (first-order) autocorrelation of residuals (r = −0.0123, p = 0.95), so no adjustment of standard errors was necessary. These findings were robust to the various imputations applied: the difference between periods in proportions correctly dispatched as stroke was significant for all imputations, with p-values ranging from 0.001 to 0.017. When data from the during- and post-implementation periods were combined and compared with the pre-implementation data the overall effect of training was significant (p = 0.002; OR 2.90, 95% CI 1.50 to 5.61).

FIGURE 6.

Weekly numbers of patients with final diagnosis of stroke dispatched as stroke and not stroke.

FIGURE 7.

Weekly percentages of patients dispatched as stroke (denominators are patients with a final diagnosis of stroke).

For the 199 patients with a final diagnosis of stroke and dispatch data, time of call was missing for three (1.5%). Of the remainder, 25 had missing arrival at hospital time. Twenty-two of these patients had admission to hospital time recorded and one further patient had triage time recorded, so these were used for the time of arrival at hospital. One hundred and ninety-four patients had mean (standard deviation; SD) call to arrival at scene 13.1 (12.0) minutes (median 9, IQR 7–15, range 3–99 minutes) and mean (SD) call to arrival at hospital 47.0 (16.3) minutes (median 44, IQR 36–53, range 22–124 minutes).

There was a marginally non-significant reduction in mean time from the EMS call to arrival of the ambulance at the scene of 2.8 minutes (95% CI −0.2 to 5.9 minutes; p = 0.068) between pre- and post-implementation periods ( Table 28 ). However, these mean times increased significantly overall during implementation (7.0 minutes, 95% CI 3.8 to 10.3 minutes; p < 0.001). There was no evidence of autocorrelation of residuals (r = −0.16; p = 0.40), so adjustment of standard errors was not deemed necessary. Overall, there was no significant underlying trend in mean time from the EMS call to arrival of the ambulance at the scene (p = 0.18), but potentially appeared to decline over the during-implementation period (p = 0.081). Findings were similar when geometric rather than arithmetic means were modelled.

The mean call to arrival at hospital time was unchanged between pre- and post-implementation periods (difference −0.1 minutes, 95% CI −5.5 to 5.3 minutes; p = 0.23) ( Figure 8 ). However, there was a significant short-term increase during implementation of 7.9 minutes (95% CI 2.2 to 13.6 minutes; p = 0.009). There was no evidence of autocorrelation of residuals (r = −0.26, p = 0.16), so adjustment of standard errors was not deemed necessary. Overall, there was a significant underlying trend in mean time from the EMS call to arrival of the ambulance at the scene (p = 0.028), but no evidence that this underlying trend varied between periods (p = 0.55). Findings were similar when geometric rather than arithmetic means were modelled.

FIGURE 8.

Weekly mean times from call to arrival at scene and times from call to arrival at hospital.

Discussion

Implementation of stroke-specific online training resulted in an increased recognition of stroke by EMDs and a modest reduction in the time from the call to the ambulance’s arrival at the scene.

Prior to training, for every 10 stroke calls made, approximately six were correctly identified by the EMDs. This increased to 8 out of 10 after training was implemented and was just short of 9 out of 10 in the during-implementation period. The recognition of stroke even at baseline was much higher in comparison with previous reports in the literature, where studies have shown that EMD sensitivity for identifying stroke is < 50%,7–9 and the findings in phase 1. One explanation for this may be the FAST campaign, which was coming to an end when phase 8 started; the FAST may have raised awareness of stroke in both the public and EMDs. Despite the relatively high proportion correctly recognised as stroke by EMDs at baseline, the difference between periods in proportions of strokes, correctly dispatched as such, was both clinically and statistically significant. This effect was robust to assumptions made about missing dispatch and final diagnosis, and there was no evidence of non-independence. The effect of the training was observed quickly, with a steep rise in the proportion of strokes being dispatched as such during implementation. These proportions were similar or better during implementation than post implementation. The difference between pre- and during-implementation periods was seen as a step change close to the time that the training package was introduced, rather than a trend, evidence of which would have been visible through the use of an interrupted time series design. This step change strengthens the evidence for the effectiveness of the intervention because it suggests that the cause of the change happened close to the time of the introduction of the training. Further evidence to support the introduction of the training as the cause of the improvement in correctly dispatching strokes as such comes from the expected rate of effect. The training reached maximal effect at the time of the third observation week in the during-implementation period, which would be consistent with increasing numbers of EMDs having received the training and being able to implement it in practice: the critical mass of EMDs needed to bring about a demonstrable effect.

One method of improving recognition of stroke would be to alter the algorithm used by EMDs. At the outset of our programme of work it had been identified that it would not be possible to change the AMPDS system and so we opted to develop training that would complement existing knowledge and skills of the EMDs. In addition, the versions of AMPDS changed between phases 1 and 8 (from v11.3 to v12.2). The latter version includes the FAST, which could also have contributed to the higher level of stroke recognition during the pre-implementation period. However, the inclusion of the FAST within the AMPDS system does not negate the value of the training. Our training is aimed at aiding EMDs recognise stroke, primarily during what we termed in phase 3, the first story. The idea being that the training will guide and support the EMDs in their recognition of the call as being one relating to stroke, resulting in the EMD choosing the stroke algorithm (and using a dispatch code, priority level, and emergency vehicle commensurate with stroke). As the FAST is only included in the stroke algorithm, the value of the training is that it will guide EMDs to that algorithm and thereby allow use of the FAST.

As a result of this approach we believe that this is the first study to (develop and) evaluate stroke training for EMDs and subsequently demonstrate an improvement in the recognition of stroke by EMDs. To our knowledge, no previous studies have evaluated EMD training in relation to the recognition of stroke; therefore, this is the first study to improve the recognition of stroke by EMDs. Previous studies have evaluated educational programmes aimed at improving paramedic, hospital and community awareness. 53 This training increased paramedic diagnosis of stroke from 61% to 79%53 and demonstrated that stroke-specific training for EMS personnel can be effective in improving stroke recognition. This suggests that there is a potential for increasing pre-hospital recognition of stroke further by including training for ambulance staff. We have shown that online learning for EMS staff can increase stroke knowledge and provides the opportunity for continuing professional development,63 although we did not study the impact on diagnostic accuracy or timeliness of service.

There was a significant difference in the number of stroke patients correctly dispatched as such between periods, but there was no significant overall trend in observation week. This may be due partly to the wide variation in the number of stroke patients who accessed the EMSs over the study period, which ranged from 3 to 17 patients per week. In addition, there were differences in the average number of patients identified in the observation weeks between periods. In the pre-implementation period there was an average of nine patients per observation week, but in both of the subsequent periods there was an average of six patients per week. However, prior to the beginning of the study the national FAST mass media campaign40 had just ended. This may explain the larger numbers of patients accessing hospital via the EMSs in the pre-implementation period. As the impact of the FAST campaign dwindled, more patients in the during- and post-implementation periods may have accessed emergency help by other means such as direct attendance at the ED, and therefore would not have been included in our study.

As well as increasing recognition of stroke by EMDs, the training package reduced pre-hospital delay in terms of call to arrival at scene time (although this failed to reach traditional levels of significance), but did not affect call to arrival at hospital time. One previous study reported that following the implementation of a training package for paramedics the time from dispatch to arrival at hospital increased from 42.2 minutes to 45.8 minutes. 53 Our study shows that EMD training not only improves recognition of stroke but may have the potential to contribute towards reducing pre-hospital delays, at least in terms of call to arrival at scene time. There are potentially many factors that can influence the time between a call being made and a patient arriving at hospital, from the dispatch code and priority through to road conditions. An improvement in the correct identification of stroke patients has the greatest probability of influencing the time between call and arrival at scene through ensuring the correct dispatch code, vehicle, and priority are used. Our findings are consistent with such an effect. The reason that the effect on time from call to arrival at scene was not reflected in a reduced time from call to arrival at hospital is unclear. Also unclear is the reason for the pattern of times during implementation, in particular the number of sharp increases and decreases. Potential explanations outside of the influence of the EMSs include presence of road works, shorter daylight hours and the recording of data in December and early January. It is difficult to provide explanations for these patterns, but sampling over longer time frames, to include the same annual periods, and/or increasing the frequency of observations may have helped with interpreting the data.

This modest reduction in time may assist in expediting assessment and delivery of thrombolytic therapy for eligible patients. It is possible that this opportunity to deliver thrombolysis may be missed when non-stroke categories are used by EMDs. However, the delay in contacting the EMSs continues to be one of the main causes of patient exclusion, in receiving thrombolysis, with around 5% of stroke patients currently receiving thrombolysis each year in England. 6 Reasons for pre-hospital delay often include poor awareness of stroke symptoms, reluctance to seek medical help and stroke not being viewed as a medical emergency. 67,68

Limitations

The nature of the intervention meant it could not be evaluated as a single centre randomised controlled trial and so we used an interrupted time series design. Although this design is not as robust as randomised controlled trial methodology, it is more robust than a simple before and after study. Alternatively, the intervention could have been evaluated using a cluster randomised controlled trial, or even a stepped-wedge cluster randomised controlled trial, but this would have meant a much larger-scale study, which would have been beyond the limit of the resources available.

We tried to ensure that data capture was over a long period in order to obtain a large enough sample to demonstrate the impact of the intervention. By sampling over 18 months we can be reasonably certain the sample was representative of the annual intake of stroke patients at the centre, and that there was unlikely to be an impact of seasonal variation. In contrast, an 18-month period was a relatively short time series and so has limited the potential for investigation and explanation of underlying trends in the data, for example, in the time from call to arrival at scene and hospital. However, this does not invalidate the findings of improved recognition of stroke by the EMDs and the modest reduction in times from pre- to post-implementation.

The intervention was implemented in one ambulance service, which may limit generalisability. Also, the EMD recognition of stroke at baseline was high compared with other studies, so it could be argued that the EMDs’ prior awareness of stroke facilitated further learning. The sample was identified through a retrospective audit of hospital and EMS records, meaning that the data collected for the study was dependent on previously documented information, and so it was not possible to verify the information through independent assessment. Given the nature of the data that were used, this is unlikely to have affected the results of the study: dispatch data are an objective measure of the code used by EMDs; the stroke register is comprehensive, has been in place over 17 years and is managed by an experienced team.

Conclusion

This is the first study to develop, implement and evaluate the impact of a training package for EMDs with the aim of improving the recognition of stroke. The findings suggest that in addition to improving the recognition of stroke by EMDs, the training has the potential to contribute to a reduction in pre-hospital delays.

Summary

Phase 8 is the culmination of a programme of research, which explored stroke symptoms and how they are used in the communication between the public and EMDs in order to develop online training aimed at raising awareness of stroke for EMDs. This is the first study to develop and evaluate the impact of such an intervention. We have shown that our stroke-specific online training package can improve the recognition of stroke by EMDs.

Contributions of authors

Professor Caroline L Watkins (Professor of Stroke and Older People’s Care, University of Central Lancashire): substantial contributions to the design and methodology, advised on the data collection and data analysis for all phases. Contributed to the writing, revision and approval of the final report.

Dr Stephanie P Jones (Research Fellow, University of Central Lancashire): substantial contributions to the design and methodology, data collection and data analysis for all phases. Contributed to the writing, revision and approval of the final report.

Dr Michael J Leathley (Principal Lecturer, University of Central Lancashire): substantial contributions to the design and methodology, advised on the data collection and data analysis for all phases. Contributed to the writing, revision and approval of the final report.

Professor Gary A Ford (Jacobson Chair of Clinical Pharmacology, Newcastle University): substantial contributions to the design and methodology, advised on the data collection and data analysis for all phases. Contributed to the writing, revision and approval of the final report.

Professor Tom Quinn (Professor of Clinical Practice, University of Surrey): substantial contributions to the design and methodology, advised on the data collection and data analysis for all phases. Contributed to the writing, revision and approval of the final report.

Miss Joanna J McAdam (Research Associate, University of Central Lancashire): substantial contributions to the design and methodology, data collection and data analysis for all phases. Contributed to the writing, revision and approval of the final report.

Dr Josephine ME Gibson (Senior Lecturer, University of Central Lancashire): substantial contributions to the design and methodology, data collection and data analysis for all phases. Contributed to the writing, revision and approval of the final report.

Professor Kevin C Mackway-Jones (Medical Director, North West Ambulance Service): contributed to the design and methodology, the revision and approval of the final report.

Mr Stuart Durham (Consultant in Emergency Medicine, Royal Preston Hospital): substantial contributions to the design and methodology. Contributed to the revision and approval of the final report.

Dr David Britt (Honorary Research Fellow and Patient Adviser, University of Liverpool): substantial contributions to the design and methodology. Contributed to the revision and approval of the final report.

Dr Sara Morris (Senior Research Associate, University of Lancaster): substantial contributions to the design and methodology. Contributed to the revision and approval of the final report.

Dr Mark O’Donnell (Consultant Stroke Physician, Blackpool Victoria Hospital): contributed to the interpretation of the data, the revision and approval of the final report.

Dr Hedley CA Emsley (Consultant Neurologist, Royal Preston Hospital): contributed to the interpretation of the data, the revision and approval of the final report.

Dr Shuja Punekar (Consultant Physician, Royal Preston Hospital): contributed to the interpretation of the data, the revision and approval of the final report.

Dr Anil Sharma (Consultant Physician, Aintree Hospital): contributed to the interpretation of the data, the revision and approval of the final report.

Dr Chris J Sutton (Principal Lecturer, University of Central Lancashire): substantial contributions to the design, methodology and data analysis. Contributed to the writing, revision and approval of the final report.

Disclaimers

This report presents independent research funded by the National Institute for Health Research (NIHR). The views and opinions expressed by authors in this publication are those of the authors and do not necessarily reflect those of the NHS, the NIHR, CCF, NETSCC, PGfAR or the Department of Health.

About the course

The aim of this course is to improve the knowledge and skills of emergency medical call handlers in the identification of patients with acute stroke.

This course will help you to:

• understand why stroke is a medical emergency

• appreciate the importance of early identification

• recognise the symptoms of stroke and how the symptoms are reported by members of the general public.

Background

Stroke is the third most common cause of death in the UK and the leading cause of adult neurological disability.

The National Stroke Strategy (2007) recommends that all patients with suspected acute stroke are immediately transferred by ambulance to hospital.

The National Stroke Strategy also recommends that public and professional awareness of the risk factors and symptoms of stroke should be improved and that stroke should be treated as a medical emergency.

This content of this course has been informed by a large research project called ESCORTT (Emergency Stroke Calls: Obtaining Rapid Telephone Triage). ESCORTT has involved reviewing patient records, interviewing patients and other people who have contacted the emergency services for suspected stroke, as well as listening to actual 999 calls for stroke and other conditions.

Public awareness

The symptoms and severity of stroke can vary from person to person. Because of this, members of the public may have difficulty recognising that they have the symptoms of a stroke (Panicoli et al. 69).

From 2009 onwards, a national campaign led by the Department of Health aimed to raise public awareness. By helping members of the public and health professionals to identify stroke symptoms and to ensure that stroke is treated as a medical emergency by dialling 999.

Routes into hospital

When a person has dialled 999 with a suspected stroke, they will receive a category A (currently up to a 19-minute response) and should be transported as quickly as possible to the nearest Trust with an Acute Stroke Unit. Some patients (e.g. those suitable for thrombolysis) may need to be transported to specialist centres in accordance with local agreements.

Local agreements may also require ambulance personnel to contact the stroke team as the patient is transported to hospital. Depending on local agreements the patient will need to be taken to the Accident and Emergency (A&E) Department or directly to the Acute Stroke Unit. Receiving clinicians must be made aware that stroke is a medical emergency.

Stroke is a medical emergency

It is imperative to seek immediate medical treatment at the first sign of stroke, by calling 999.

The quicker a person is treated, the more brain will be saved and the better the chance of a complete recovery. 1.9 million neurons are lost for each minute that a stroke is left untreated.

Time is brain!

What is a stroke?

A stroke is a disruption of blood flow to the brain or part of the brain due to a blood clot (blockage) or bleed (haemorrhage).

Symptoms of stroke may vary from problems such as difficulties speaking or inability to use the arm or leg to loss of consciousness.

There are different types of stroke

Transient ischaemic attack (TIA)

What if the symptoms of stroke go away quickly? An ischaemic stroke event that resolves quickly is called a transient ischaemic attack or TIA.

Most TIAs only last between 5–30 minutes and present with the symptoms of stroke. A person who has had a TIA has a 7–10% risk of having a stroke in the next 7 days.

Those with TIA are more likely to go on to have a stroke within 24 hours of onset of symptoms, than a person with chest pain going on to have a heart attack.

TIA is a medical emergency, particularly for those who have more than one TIA in a week.

Any patient with a suspected TIA requires an urgent full neurological assessment. All TIA patients should be taken to A&E even if though their symptoms may have resolved. If the patient does not want to go to hospital, they should be informed of the risk of having a stroke and that they need a neurological assessment.

Animation

The following animations demonstrate how the obstruction of different arteries in the brain results in a range of symptoms.

Stroke is a medical emergency. If people suspect that they have had a stroke they should dial 999 immediately. This is because 1.9 million neurons (nerve cells) are lost for every minute in which stroke is left untreated (Saver and Lutsep70).

The following animations demonstrate how the obstruction of different arteries in the brain results in a range of symptoms.

Use the sliders to alter the physiological parameters, to see the effects on the post-stroke brain.

Suspected stroke and dialling 999

On the onset of symptoms two-thirds of stroke patients obtain first medical contact from the emergency medical services (EMSs).

In nearly all cases of suspected stroke, people who call 999 are family members, friends, neighbours or unrelated bystanders.

Only 3% of calls are made by the patients themselves.

The symptoms and severity of stroke can vary from person to person and because of this, members of the public may have difficulty recognising the symptoms of a stroke.

Around one-third of callers recognise the problem as a stroke.

How can I recognise a stroke? – The signs and symptoms

The signs and symptoms of stroke can vary from person to person depending on what part of the brain has been affected. A person may present with one, or a combination of the symptoms listed below.

The symptoms of stroke are usually:

• sudden weakness, numbness, or paralysis of the face, arm, or leg, on one side of the body

• sudden difficulty walking, loss of balance or co-ordination, known as ataxia

• sudden, severe headache with no known cause

• sudden trouble seeing in one or both eyes.

Visual disturbance may include blurred or double vision or hemianopia, whereby blindness in the visual field of both eyes means that the patient can see left with one eye and right with the other. Other visuospatial problems may include inattention where the person does not recognise one side of their body as being part of them or are unable to acknowledge left/right visual-space.

Other symptoms

The following symptoms may occur in combination with those listed above or they may occur alone. These symptoms may also arise from other acute illness:

• confusion, which may be as a result of the event or related to the patient’s prior medical history

• some patients will be agitated and may be aggressive, this could be as a result of hypoxia (lack of oxygen)

• patients are likely to feel frightened, anxious and are unable to make sense of what is happening to them

• patients and their carers will require reassurance

• body temperature often rises in the acute phase of stroke and so patients may have a warm flushed appearance

• vomiting

• incontinence.

Stroke mimics

Hypoglycaemia – blood glucose monitoring should be carried out on any patient that is neurologically compromised.

Bell’s palsy – a condition in which one side of the face becomes paralysed, the symptoms are usually temporary.

Previous stroke – this may have left the person with a residual deficit.

Hemiplegic migraine – symptoms can include sudden hemiplegia (paralysis on one side of the body) fever, the symptoms of meningitis without the actual illness.

Consciousness – ranging from confusion to profound coma.

Intoxication – caused by alcohol or drugs.

Acute confusional state – may be a posterior circulation episode. Commonly, particularly with the older patient, an acute confusional state is due to sepsis.

Emergency Stroke Calls: Obtaining Rapid Telephone Triage (ESCORTT)

What did we do?

Through interviewing people who had dialled 999 and by listening to 999 calls we have identified factors that can improve the accuracy of early diagnosis of stroke. Using this information, this course has been developed for call handlers.

The impact of the training package was assessed to see if stroke patients can be identified and diagnosed more quickly by call handlers who have completed this training. An evaluation of the training package took place in one EMS regional control centre. Sixty-nine call handlers completed the training.

The evaluation showed that before implementation of the training package call handlers correctly identified 63% of stroke patients; this increased to 80% following completion of the training.

Suspected stroke: describing the symptoms

We listened to 592 calls to the EMSs for patients who went on to have a confirmed diagnosis of stroke in hospital. Overall, collapse or fall was the most commonly reported problem (40%), with stroke the second most common (37%).

Other common reported problems included:

• 17% said change in consciousness level

• 15% said limb weakness

• 8% said previous history of stroke

• 4% said facial weakness.

Stroke patients who have facial weakness and speech problems are more likely to be given a correct diagnosis of stroke by call handlers. A correct diagnosis of stroke by the EMSs results in a quicker journey to hospital.

People who dial 999 about non-stroke conditions rarely say stroke, limb weakness, speech problems, or facial weakness. However, the word fall is commonly used in both stroke and non-stroke calls.

Other descriptions included:

• funny turn

• confused

• unable to move

• dizzy

• unresponsive

• cannot speak

• seizure.

What have we found?

If a person has had a stroke, they or someone calling on their behalf are more likely to dial 999 if they perceive their symptoms to be serious.

In nearly all cases of suspected stroke, people who call 999 are family members, friends, neighbours or unrelated bystanders. Only 3% of calls are made by the patients themselves.

After someone has had a stroke, the person ringing 999 may also use the terms or descriptions: fall, collapse and problems speaking. The likelihood of it being stroke is increased if the caller uses two or more of the following descriptors together:

• stroke

• fall

• collapse

• problems speaking.

Callers often feel reassured by the advice given by the EMSs during the 999 call, but some can be uncertain about whether an ambulance had been dispatched or not.

When reviewing EMS patient report forms, a diagnosis of stroke was most likely to be made by call handlers and ambulance personnel when stroke patients had facial weakness and/or speech problems.

Levels of consciousness

The terms consciousness and unconsciousness are used by call handlers but callers have difficulty understanding different levels of consciousness.

For example, callers often describe levels of consciousness in the following ways:

• not awake

• unresponsive

• asleep

• cannot wake them up

• not with it

• keep drifting in and out

• semi-coma

• delirious

• disorientated

• cannot seem to recognise me

• consciousness.

Stroke patients may have altered consciousness. If a patient is completely unconscious there could be many other reasons. The terms consciousness and unconsciousness are used by call handlers to describe different levels of consciousness, however, the caller has a different vocabulary in this respect.

Levels of consciousness activity

Try the exercise below to see what words callers are likely to use to describe different levels of consciousness.

Risk factors

A patient’s past medical history may include some of the risk factors for stroke. There are a number of factors that increase the risk of stroke. Risk factors can be divided into three categories:

Non-modifiable: Those the patient cannot change.

Treatable: Those that can be treated.

Modifiable: Those that can be improved through positive changes in lifestyle.

Non-modifiable risk factors

Age: 1 in 4 men and 1 in 5 women over the age of 45 will have a stroke. The risk of stroke increases with age.

Family history: Stroke is more common in people whose close relatives have suffered a stroke.

Ethnicity: Stroke is more prevalent in some ethnic minority groups.

Treatable risk factors

Hypertension: High blood pressure is the greatest risk factor for stroke. Medication can be prescribed to lower blood pressure.

Previous stroke or Transient Ischaemic Attack (TIA): Greatly increases the risk of a recurrent event. Medication is required, and in some cases surgery.

Diabetes: Blood glucose levels need to be carefully monitored and controlled.

Heart disease: Atrial fibrillation in particular increases the risk of stroke. Medication to thin the blood (for example warfarin) or to reduce cholesterol (for example statins) may be prescribed.

High cholesterol: Cholesterol levels need to be carefully monitored and controlled.

Modifiable risk factors (lifestyle)

Smoking.

Obesity.

Physical inactivity.

Excessive alcohol consumption.

Drug use.

Thrombolysis

For those who do experience a stroke or TIA, rapid access to effective stroke care and treatment can reduce death and dependency by enabling the immediate provision of interventions such as thrombolysis. Access to organised stroke care is also important, as it has been shown to save lives and reduce disability.

Thrombolysis is a ‘clot-busting’ medicine that can dissolve blood clots following acute ischaemic stroke.

Thrombolysis for ischaemic stroke is becoming widely available across the UK. Recent research from the IST3 trial and a systematic review have provided more detailed information about which patients may benefit from thrombolysis and at what time-points.

The following must be answered yes if a patient is to be considered eligible for thrombolysis.

Does the patient have the symptoms and signs of acute stroke?

Is there a clear time of onset within the last 6 hours?

Is the patient over 18 years old?

Not all patients with ischaemic stroke are eligible for thrombolysis and eligibility will be assessed further by the ambulance crew and by hospital staff on arrival at hospital.

Multiple-choice quiz

To access the multiple-choice quiz, click on assessments on the tool bar on the left of your screen. When you have completed the quiz click submit, your submission will be graded immediately.

To pass the quiz you must achieve 60%. If you successfully complete the quiz you will receive a certificate of completion, an electronic certificate will be sent to you.

Your certificate of completion will only be sent out when you have completed the online module evaluation form. This can be accessed by clicking on assessments and then module evaluation.

ESCORTT phase 1: identifying stroke: the association between patient characteristics and stroke symptoms with pre-hospital diagnosis and time to admission

ESCORTT phase 2: callers’ experiences of making emergency calls at the onset of acute stroke: a qualitative study

ESCORTT phase 3: The Identification of Acute Stroke: An Analysis of Emergency Calls

ESCORTT phase 3a: “Is he awake?”: Dialogues between callers and call handlers about consciousness during emergency calls for suspected acute stroke

ESCORTT phase 4: exploring the words used by callers for non-stroke-related calls to the emergency medical services and a comparison with the words used describing stroke-related calls

ESCORTT phase 5: Developing algorithms and protocols for ambulance and NHS Direct staff to assist in the identification of those with suspected stroke

Outputs

Changes to AMPDS and NHS Direct protocols.

Costings

Researchers will be reimbursed for any travel expenses. No other costs are anticipated.

ESCORTT phase 6: Developing an on-line training package for Emergency Medical Dispatchers to assist in the identification of those with suspected stroke

ESCORTT phase 7: Implementation and evaluation of an on-line stroke-specific training package for Emergency Medical Dispatchers

ESCORTT phase 8: Training Emergency Services’ Dispatchers to recognise stroke: The evaluation and impact of a training package

Phase 1

Leathley MJ, Jones SP, Sutton CJ, McLoughlin ASR, Emsley HCA, Punekar S, et al. The relationship between stroke symptoms and diagnosis: a cohort study. Emerg Med J submitted. Outcome awaited.

Jones S, Leathley M, McDonald K, Watkins C, on behalf of the ESCORTT Group. Pre-hospital Identification of Stroke. Third UK Stroke Forum Conference Book of Abstracts. 2008:22–3.

Sutton CJ, Dickinson H, Jones SP, Leathley MJ, McLoughlin A, Watkins CL, on behalf of the ESCORTT Group. Emergency Stroke Calls Obtaining Rapid Telephone Triage (ESCORTT) Phase 1: how and when are patients diagnosed with stroke? Int J Stroke 2010;5(Suppl. 3):15. [Presented as poster at the Fifth UK Stroke Forum Conference, 30 November to 2 December 2010, Glasgow, UK.]

Phase 2

Jones SP, Dickinson HA, Ford GA, Gibson JME, Leathley MJ, McAdam JJ, on behalf of the ESCORTT Group. Callers’ experiences of making emergency calls at the onset of acute stroke: a qualitative study. Emerg Med J 2012;29:502–5.

Jones S, Dickinson H, Gibson JME, Leathley M, McAdam J, McDonald K, on behalf of the ESCORTT Group. Caller experiences of dialling 999 for suspected stroke. Int J Stroke 2009;4(Suppl. 2):13. [Presented as poster at the Fourth UK Stroke Forum Conference, 1–3 December 2009, Glasgow, UK.]

Phase 3

Jones SP, Carter B, Ford GA, Gibson JME, Leathley MJ, McAdam JJ, on behalf of the ESCORTT group. The identification of acute stroke: an analysis of emergency calls. Int J Stroke 2013;8:408–12.

Jones SP, McAdam JJ, Gibson JME, Leathley MJ, Dodd A, Watkins CL, on behalf of the ESCORTT Group. How do callers describe symptoms when contacting emergency services at the onset of acute stroke? A study of the ‘first story’ given by callers. Int J Stroke 2010;5(Suppl. 3):14. [Presented as poster at the Fifth UK Stroke Forum Conference, 30 November to 2 December 2010, Glasgow, UK.]

Phase 3a

Gibson JME, Bullock M, Ford GA, Jones SP, Leathley MJ, McAdam JJ, et al., on behalf of the ESCORTT Group. ‘Is he awake?’: dialogues between callers and call handlers about consciousness during emergency calls for suspected acute stroke. Emerg Med J 2012;30:414–18.

Jones SP, McAdam JJ, McLoughlin ASR, Gibson JM, Leathley MJ, Watkins CL, on behalf of ESCORTT Group. Understanding of the term conscious when used during emergency calls for suspected stroke. Cerebrovasc Dis 2011;31(Suppl. 2):234. [Presented as poster by Jones SP at the 10th European Stroke Conference, 24–27 May 2011, Hamburg, Germany.]

Phase 4

Jones SP, Gibson JME, McAdam JJ, Leathley MJ, Ford GA, Quinn T, et al., on behalf of the ESCORTT Group. ‘Can you send an ambulance please?’: a comparison of callers’ requests for emergency medical dispatch in stroke and non-stroke calls. Emerg Med J 2013. doi: emermed-2013-202752.

Phase 7

Jones SP, Ford GA, Leathley MJ, McAdam JJ, McLoughlin AS, Quinn T, et al., on behalf of the ESCORTT Group. An evaluation of stroke-specific training for emergency medical dispatchers. Cerebrovasc Dis 2012;33(Suppl. 2):616–17. [Presented as poster at the 11th European Stroke Conference, 22–25 May 2012, Lisbon, Portugal.]

Jones SP, Leathley MJ. Emergency Stroke Calls: Obtaining Rapid Telephone Triage: An Evaluation of Stroke-specific Training for Emergency Medical Dispatchers. North West Ambulance Service NHS Trust paramedic training event, 15 August 2012. North West Ambulance Service NHS Trust Head Quarters, Bolton, UK.

Phase 8

Watkins CL, Leathley MJ, Jones SP, Ford GA, Quinn CJ, Sutton CJ, on behalf of the ESCORTT Group. Training emergency services’ dispatchers to recognise stroke: the evaluation and impact of a training package. BMC Health Serv Res 2013;13:318.

Jones SP, Sharma A, Dodd A, Leathley MJ, McAdam JJ, Quinn T, et al., on behalf of the ESCORTT Group. Training call handlers to recognise stroke: the evaluation and impact of a training package. Int J Stroke 2011;6(Suppl. 2):5. [Oral presentation by Jones SP at the Sixth UK Stroke Forum Conference, 30 November 2011, Glasgow, UK.]

All phases

Jones SP, Watkins C. Emergency stroke calls: obtaining rapid telephone triage. Stroke Matters 2009;2:14.

McAdam JJ, Jones SP. Emergency Stroke Calls: Obtaining Rapid Telephone Triage (ESCORTT): An Overview. World Stroke Day Conference – Advancing Stroke Care Through Research, 28 October 2011, Wigan, UK.

Jones SP. Emergency Stroke Calls: Obtaining Rapid Telephone Triage. National Ambulance Research Group Meeting, 31 January 2012, London, UK.

Watkins CL, Jones SP, Leathley MJ. Emergency Stroke Calls: Obtaining Rapid Telephone Triage: An Overview of the Results. North West Ambulance Service NHS Trust, Medical Senior Management Team Meeting, 11 July 2012. North West Ambulance Service NHS Trust, Bolton, UK.

Watkins CL. Emergency Stroke Calls: Obtaining Rapid Telephone Triage (ESCORTT): The Results. World Stroke Day Conference, 29 October 2012, Wigan, UK.