Introduction

Delay in calling emergency medical services (EMS) following stroke onset is an important factor that prevents patients accessing hyperacute stroke care and thrombolytic therapy, an intervention which can have a beneficial impact on patient outcome [1]. Delay times range from 3 to 6 hours [2] reducing the chances of optimal recovery for the majority of stroke patients. EMS are typically contacted by witnesses of stroke, not stroke patients [3]. EMS phone calls for stroke are initiated by 2–7% of patients, with remaining contacts made by witnesses including family members (∼41–75%), friends (∼4–20%), or medical/care personnel (∼13–42%) [3]–[7]. Nevertheless, most research to date has focused on the patient rather than witness factors that determine response behaviour following stroke. However, different factors might influence witnesses’ compared with patients’ responses. It has been suggested that family members and friends should be targeted for interventions as they are likely to act on behalf of the patient when stroke occurs [2]. Past public education campaigns [8], [9] included stroke witnesses as a target population, but overall most interventions have been found to be only minimally effective [2], [10], [11]. Such interventions would benefit from a sound understanding of factors that influence witnesses’ response behaviour to stroke.

Despite the advantages of theory-based research and interventions, [12] the field of help-seeking behaviour often lacks rigorous methods [13] including an explicit theoretical basis [14], [15] and systematic intervention development [10]. Descriptive theories of the patient process in seeking medical care outline a number of distinct stages, [16]–[18] including one stage of situational appraisal where a health threat is detected (appraisal stage) and another stage where response behaviours are selected (behavioural stage). Our study explored factors and processes that might influence witness responses to stroke by examining witnesses’ interviews of their stroke experience using two theoretical frameworks that provide explanatory accounts of appraisal and behavioural factors.

First, the Common Sense Self-Regulation Model [19] was selected as it provides a detailed explanatory account of the processes involved in making sense of a health threat. According to the model, individuals faced with a health threat form a mental representation which reflects the individuals’ understanding of the illness. These illness representations include aspects of the illness such as its label (identity), its time-course (timeline), its effects on the person (consequences), its controllability (personal control) and its cause. The Common Sense Self-Regulation Model has previously informed research on health care seeking behaviour [20]. Second, the Theory Domains Framework [21] was selected as it provides a broad and inclusive evidence-based account of all relevant factors outlined in current theories of behaviour. This framework comprises 12 domains of theory-based explanations for behaviour resulting from a comprehensive review of behavioural theories and an expert consensus approach grouping theoretical constructs by commonality.

Methods

Results

Twenty-six stroke witnesses (20 female) were recruited ≤14 days post stroke with the help of nurses at three participating stroke units in North-East England. Witnesses’ relationships to the patients were: wife/husband (n = 14), son/daughter (n = 9), granddaughter (n = 1), formal care-giver (n = 1) and acquaintance (n = 1). All strokes occurred in the patient’s home, except one that occurred in a supermarket. Most contacts with EMS (n = 13) and some to primary care physician offices (n = 2) were made within 1 hour of symptom onset. For those who responded after 1 hour, contact included EMS calls (n = 7), primary care physician (n = 3) and ED visits (n = 1). All variables reported in the next section were identified as influencing witness’ response behaviours.

Discussion

This study outlines multiple interrelated factors likely to influence witnesses’ responses to stroke. Different processes and determinants may result in similar response behaviours and some influence of stroke appraisal and behavioural factors may operate out of immediate awareness.

Heuristics seemed to guide information processing, often regardless of symptom recognition, sometimes contributing to an immediate response. As acute stroke is often a complex presentation, heuristics might play a crucial role in adaptively guiding witness’ appraisal and response processes by ignoring redundant information (e.g. loss of speech requires immediate medical attention regardless of other symptoms). Complexity-reducing mental shortcuts to behaviour might be used in rapid-onset high-threat conditions such as stroke, as deliberation of all relevant stimuli could lead to more harmful outcomes. Heuristics have been shown to result in superior behavioural responses in a variety of situations as compared to more complex processes [23] and may partly explain findings in the delay literature, such as that less severe strokes are associated with longer delay [24] and differential delay patterns for different stroke strategies [25].

Equal proportions of witnesses identified, suspected or failed to recognise stroke (one-third each respectively). Those failing to recognise stroke often formed alternative conclusions about symptom identities/causes. Once a health threat was detected, increasingly elaborate mental representations of the patients’ health condition were formed. Illness representations seemed to influence responses, with those typically not associated with an emergency (e.g. food poisoning) potentially causing delays compared to illness representations typically associated with emergency (e.g. heart attack). Witnessing stroke is likely to lead to a rapid instinctive appraisal and responses followed by more elaborate processes taking additional factors into consideration.

This study might help to explain other common findings in the stroke delay literature, such as the gap between stroke knowledge and response behaviour [24], [26]. Knowledge appeared to be beneficial, but neither necessary nor sufficient factor to influence appropriate responses for the following reasons:

1. Knowledge of symptoms might not lead to stroke recognition (e.g. a witness has factual knowledge of all symptoms but fails to match knowledge with an occurring stroke in the moment);
2. Stroke (or other serious illness) recognition might occur even if witnesses have poor stroke knowledge (e.g. a witness only knows a limited number of symptoms but encounters one of these, recognises stroke and responds adequately);
3. Following stroke recognition, the appropriate response might be unknown (e.g. a witness recognizes stroke, but thinks it is right to contact the primary care physician instead of EMS);
4. Knowledge effects might be masked by other factors with potentially stronger influence on behaviour (e.g. a witness recognizes stroke and the need to contact EMS, but delays to avoid inconveniencing busy services).

Further research needs to explore the processes through which knowledge influences responses, using explanatory frameworks such as the Common Sense Self Regulation Model [19]. Descriptive accounts of the varying relationship between knowledge and response can only be advanced by exploring how such effects are being generated.

Social influence seemed to play a key role. Witnesses typically engaged in conversations with patients, other witnesses and health care professionals, all of which appeared to influence behaviour with the potential to both increase or decrease delay. Some witness/patient interactions included shared decision-making elements with most patients reluctant to engage EMS, suggesting that compared to patients, witnesses’ threshold for initiating an emergency response might be lower.

Our findings suggest several factors that offer promising opportunities for the development of delay-reducing interventions, some of which are discussed below. Key factors to target are heuristics that can lead to swift behavioural performance. Behavioural interventions might use unspecific and automatic perceptions of ‘something is wrong’ as triggers for an immediate and more focused appraisal for stroke using the FAST check list [8]. This could potentially accelerate the initiation of the appropriate response by the witness.

Furthermore, as stroke identification is complex, suspicion of stroke could be linked with immediate contact of EMS. Evidence suggests witnesses are mostly correct in their diagnosis when suspecting stroke [3]. Action plans [27] to pre-specify calling EMS upon suspecting stroke could decrease delay. Additionally, coping planning [27] for probable behavioural tendencies (e.g. contacting peers or non-EMS professionals) could be counteracted by planning to immediately contact EMS on suspicion of stroke, thereby ignoring competing behavioural options.

The social elements of witnessing stroke could be used by at-risk patients to collaboratively plan [28] response patterns in the event of stroke. Pre-specifying perceived stroke symptoms as triggers for social interaction to immediately call EMS could interfere with time consuming deliberation processes. Further intervention targets might constitute strengthening of the link between contacting EMS and positive outcomes in the short (e.g. quick help) and long-term (e.g. better recovery chances) through accessing effective treatments.

The strengths of this paper are the focus on the key population of stroke witnesses and the use of psychological theory to explore factors relevant for determining response behaviours. The limitations of this research are that the relative importance of identified factors and their likely interactions cannot be determined using qualitative methods. Moreover, the use of different theoretical frameworks, might have uncovered additional important factors not captured within the models employed in the current analysis. Lastly, retrospective accounts of witness’ experience are prone to recall bias and the sample obtained for this study might not be representative of the population of stroke witnesses.

Future research should include both explanatory and descriptive frameworks to understand the process underlying delay. Carefully unpacking witnesses’ behavioural response sequences when encountering stroke and determining the factors important for different response behaviours would further increase our understanding. In addition, interactions between witnesses and patients need to be assessed. The differences in processes and determining factors between witnesses and patients need exploration to understand common and distinct targets for appropriate behavioural interventions. Lastly, given the overlap of stroke symptoms with other health conditions and the significant risk of co-morbidity in patients at-risk of stroke, research needs to determine whether EMS responses should be advocated at the symptom level (i.e. symptom X  =  seek care Z) or at the disease level (i.e. symptom X indicates disease Y  =  seek care Z) [14]. Although seemingly similar, the two levels of focus would have implications for stroke interventions and impact on how these could be dovetailed with those for other conditions.

Customary calls for more education and awareness of stroke should be supplemented with elaborations on the specific factors that interventions should actively target. Focusing on the key population of witnesses, the current study advances our cumulative understanding of delay between stroke symptom onset and calling EMS and suggests specific targets to maximise stroke response efficiency.