Stroke is defined as a neurological deficit attributed to an acute focal injury of the central nervous system by a vascular cause. As it is a common, time-sensitive medical emergency with a high mortality rate (Sacco et al, 2013; Sibson, 2017), prehospital care focuses on early recognition and rapid transportation to an appropriate hospital for treatment (Sibson, 2017). A significant proportion of patients with stroke symptoms are later diagnosed with a non-stroke mimic condition (Winkler et al, 2009), while some spontaneously improve after initial assessment and the condition is labelled a transient ischaemic attack (TIA) (McClelland et al, 2020). Nonetheless, the emergency response should continue until stroke has been excluded.
Approximately 340 000 older people in England are living in care homes across the UK (Wolters et al, 2019). These are split into two main types: nursing homes providing 24-hour personalised nursing care; and residential homes providing assistance with daily activities only (Wolters et al, 2019).
It is well understood that care home patients will be older and have more complex health needs than the general population (Shah et al, 2010; Gordon et al, 2014; Wolters et al, 2019), and they have a higher prevalence of conditions such as stroke, dementia and mental health problems (Shah et al, 2010; Wolters et al, 2019). The presence of such comorbidities and the care environment itself might make prehospital patient assessment and management more challenging (Voss et al, 2018); for example, dementia can affect the recognition and communication of symptoms (Choonara and Williams, 2021).
No previous research containing descriptions of characteristics and emergency assessment processes of patients in care homes with acute stroke has been published. Wider evidence shows that care home patients in general access emergency services more frequently than those living in their own homes (Dwyer et al, 2018), and that healthcare provision is often not tailored towards their needs (Gordon et al, 2014).
It is therefore possible that differences between private and care home populations and settings might influence stroke identification and prehospital care. This has implications for staff training and clinical guidelines, which currently do not differentiate between settings (Association of Ambulance Chief Executives (AACE), 2023).
Aim
This study aimed to describe the ambulance care of adults with suspected stroke according to residential status of any type of care home versus a private address (i.e. patients living in their own homes).
Methods
Objectives
The aims of this study were to:
Study design
Secondary analysis of an existing ambulance clinical record dataset was undertaken. The dataset had previously been used to describe prehospital stroke care in a single UK ambulance service.
Study population
The dataset described care parameters for patients transported to hospital by the North East Ambulance Service NHS Foundation Trust, where a clinical impression of suspected stroke (including TIA or any clinical scenario suggesting possible stroke) had been documented on the electronic patient record system between December 2021 and April 2022 (McClelland et al, 2020).
The region covered by the trust contains six acute stroke units. Clinicians with varying grades and qualifications are employed so, in this report, ambulance clinician is used as a generic term for any responding practitioner. However, stroke patients are usually attended by a minimum of one person qualified to at least emergency medical technician (EMT) level.
Residential status is routinely recorded by ambulance clinicians attending any incident using predefined categories: private address; care home; public place; educational facility; other; and unknown. The care homes category includes both residential and nursing homes, which cannot be separated.
Cases were excluded from the analysis if the residential status recorded was not a care home or private address (i.e. public incident, educational facility, workplace, other or unknown) and where there were obvious data reporting errors that made no sense.
Data analysis
Prehospital time intervals were calculated from routinely captured timestamps. The presence of polypharmacy and multimorbidity were derived from lists of medication and previous medical history recorded in the ambulance dataset using standard definitions; they were defined as more than five prescribed medications (Masnoon et al, 2017) and two or more medical conditions, respectively (Johnston et al, 2019).
Patient demographics, ambulance response times, ambulance clinician assessments and ambulance clinician treatments provided were summarised and categorised according to residential status (i.e. private address or care home) using descriptive statistics.
Categorical data were compared using chi-square testing and continuous data were compared using Mann-Whitney U testing. For the statistical analysis, SPSS software (v. 28.0) was used.
Ethical issues
As a retrospective description of anonymised routine care provided by a single organisation was used, ethical approval was not required.
Results
The original dataset contained 1621 suspected stroke patients attended by ambulance clinicians between December 2021 and April 2022. Exclusion criteria removed 126 cases: 119 public incidents; one unknown location; and six private address cases with obvious nonsensical errors in data recording. Of the remaining 1495 cases, 171 were from care homes and 1324 were from private addresses (Figure 1).
The median age of patients in care homes was 11 years higher than that of those from private addresses. Care home patients had on average two more medications prescribed and two more comorbidities than people from private addresses. Rates of atrial fibrillation and dementia were higher in the care home group. The private address group had higher systolic blood pressure readings whereas the care home group had lower Glasgow Coma Scale scores, lower ACVPU (alert, confusion, verbal, pain, unresponsive) scores and higher NEWS2 scores (Table 1).
| Care home (n=171) | Private address (n=1324) | P value | |
|---|---|---|---|
| Age (years): median (IQR) | 86 (79–91) | 75 (63–83) | <0.001* |
| Sex | |||
| Female (%) | 94 (55.0%) | 672 (50.8%) | |
| Male (%) | 77 (45.0%) | 652 (49.2%) | 0.29 |
| Polypharmacy (%) | 116 (67.8%) | 625 (47.2%) | <0.001* |
| Number of prescribed medications: median (IQR) | 6 (4–9) | 4 (1–7) | <0.001* |
| Multimorbidity (%) | 163 (95.3%) | 1065 (80.4%) | <0.001* |
| Number of comorbidities: median (IQR) | 5 (3–6) | 3 (2–5) | <0.001* |
| Atrial fibrillation listed (%) | 40 (23.4%) | 181 (13.7%) | <0.001* |
| Dementia and/or Alzheimer's listed (%) | 58 (33.9%) | 62 (4.7%) | <0.001* |
| Previous stroke listed (%) | 27 (15.8%) | 192 (14.5%) | 0.65 |
| Diabetes (types 1 and 2 grouped) listed (%) | 28 (16.4%) | 229 (17.3%) | 0.76 |
| Heart rate: median (IQR) |
n=170 |
n=1319 |
0.164 |
| Blood pressure: median systolic (IQR) |
n=167 |
n=1315 |
<0.001* |
| Capillary blood glucose (mmol/l): median (IQR) |
n=164 |
n=1289 |
0.409 |
| Body temperature (°C): median (IQR) |
n=162 |
n=1267 |
0.340 |
| Glasgow coma score: median (IQR) |
n=170 |
n=1316 |
<0.001* |
| ACVPU assessment (alert: 5; confused: 4; verbal: 3; pain=2; unresponsive=1): median (IQR) |
n=156 |
n=1246 |
<0.001* |
| NEWS2 score: median (IQR) |
n=118 |
n=944 |
<0.001* |
ACVPU: alert, confusion, verbal, pain, unresponsive;
GCS: Glasgow Coma Scale; IQR: interquartile range;
NEWS2: National Early Warning Score.
Both groups received similar assessments from ambulance clinicians, including documentation of the face, arm, speech, time (FAST) test and a 12-lead electrocardiogram. When looking at the NHS Ambulance Care Quality stroke assessment bundle (NHS Digital, 2020) of FAST, blood pressure and capillary blood glucose, the private address group was more likely to have a blood pressure assessment completed. However, other physiological observation recordings were similar. In terms of the FAST items, a higher proportion of care home patients were unable to participate in this assessment but, when this was possible, a facial droop was recorded more often. Other components of the FAST assessment were similar across both groups (Table 2).
| Care home (n=171) | Private address (n=1324) | P value | |
|---|---|---|---|
| Blood pressure assessment completed (%) | 167 (97.7%) | 1315 (99.3%) | 0.028* |
| Capillary blood glucose assessment completed (%) | 164 (95.9%) | 1291 (97.5%) | 0.22 |
| 12-lead electrocardiogram assessment completed (%) | 63 (36.8%) | 570 (43.1%) | 0.12 |
| Clinician attempted FAST assessment (%) | 170 (99.4%) | 1305 (98.6%) | 0.36 |
| Clinician attempted FAST, but patient unable to participate in the test (%) |
n=170 |
n=1305 |
<0.001* |
| FAST positive result (%) |
n=153 |
n=1266 |
0.11 |
| Facial weakness symptom recorded present (%) |
n=153 |
n=1266 |
<0.001* |
| Arm weakness symptom recorded present (%) |
n=153 |
n=1266 |
0.77 |
| Speech abnormality symptom recorded present (%) |
n=153 |
n=1266 |
0.94 |
| Time of onset recorded by ambulance clinician (%)§ |
n=170 |
n=1305 |
0.76 |
| Leg weakness symptom recorded present (%) |
n=153 |
n=1266 |
0.28 |
| Number of stroke symptoms recorded (face, arms, speech and/or leg) (1–4) (%) | n=134 | n=1043 | |
| 1 | 1=48 (35.8%) | 1=375 (36.0%) | 0.98 |
| 2 | 2=38 (28.4%) | 2=274 (26.3%) | 0.60 |
| 3 | 3=22 (16.4%) | 3=188 (18.0%) | 0.64 |
| 4 | 4=26 (19.4%) | 4=206 (19.8%) | 0.92 |
FAST: face, arm, speech, time test; GCS: Glasgow Coma Scale
The ambulance response time and transfer time to hospital were similar in both groups. However, the on-scene time (OST) was significantly longer by a median of 6 minutes in the care home group. Private address patients were more likely to have intravenous cannulation attempted by paramedics, but other interventions by ambulance clinicians, such as using a hospital pre-alert call and administering drugs, were similar in both groups (Table 3).
| Care home (n=171) | Private address (n=1324) | P value | |
|---|---|---|---|
| 999 call to on-scene time (minutes): median (IQR) | 23 (13–35.5) | 25 (15–41) | 0.19 |
| On-scene time (minutes): median (IQR) | 41 (32–49.5) | 35 (27–45) | <0.001* |
| Leave scene to hospital arrival time (minutes): median (IQR) | 15 (9–21) | 16 (11–23) | 0.19 |
| Overall time of prehospital phase (999 call to hospital arrival): median (IQR) | 83 (75–106) | 80 (63–107) | 0.60 |
| Patient destination: | 0.11 | ||
| Stroke unit (%) | 33 (19.3%) | 330 (24.9%) | |
| Non-stroke unit (%) | 138 (80.7%) | 994 (75.1%) | |
| Hospital pre-alert call documented (%) | 67 (39.2%) | 584 (44.1%) | 0.22 |
| Intravenous cannulation attempted (%) | 41 (24.0%) | 423 (31.9%) | 0.03* |
| Intravenous cannulation success (%) |
n=41 |
n=423 |
0.06 |
| Number of cases where intravenous drugs were given by paramedics (if cannulated successfully) (%) |
n=30 |
n=357 |
0.72 |
IQR: interquartile range
Discussion
This retrospective examination of ambulance records demonstrated that there are potentially important differences between suspected stroke patient characteristics and the care received according to the setting where they live.
Care home patients with suspected stroke were nine years older on average than their private address counterparts, with a median age of 86 years. Consistent with these results, a large European study has previously reported a mean age of 83 years among 4156 care home residents presenting to emergency medical services (Onder et al, 2012). The authors are not aware of any previous detailed descriptions about suspected stroke admissions. It has, however, been reported in cross-sectional community studies that care home patients are typically older and have more complex care needs (Shah et al, 2010; Wolters et al, 2019), as reflected by the higher age, rates of polypharmacy and rates of multimorbidity shown in the results of the present study.
Approximately one in nine (11.4%) suspected stroke patients in the dataset were transported from care homes, which demonstrates that ambulance clinicians encounter this scenario regularly. Publicly available Sentinel Stroke National Audit Programme (SSNAP) (2023) data show that approximately 2.2% (1945/89014) of all NHS confirmed hospital stroke admissions in England and Wales came from care homes between April 2021 and March 2022; however, the national total includes patients who were not admitted by emergency ambulance, such as those self-presenting and inpatients.
If the north east is typical of other regions, the large difference between the frequency of care home patients with suspected stroke in its ambulance data at 11.2% versus the 2.2% confirmed stroke admissions from care homes in SSNAP (2023) implies that a high proportion of stroke mimic cases may be coming from these settings, possibly reflecting the difficulty of performing accurate clinical assessments in this context.
It is well understood in both practice and research that successful prehospital stroke care is defined by early recognition and timely conveyance to a stroke-receiving hospital because of the potential impact on patient outcomes (AACE, 2023); for example, when ischaemic stroke is eligible for thrombolysis, each 15-minute reduction in time to treatment is associated with one additional month of disability-free life (Meretoja et al, 2014). The care process data in the present study showed that the ambulance response time and transfer time to hospital were similar in all settings. However, the median OST was longer in the care home group by 6 minutes. This could reflect that care home patients are more challenging to assess, require a more complex manoeuvre to board the ambulance because of reduced mobility and/or are located in larger buildings. Further research would be required to understand whether this difference is amenable to intervention.
The analysis was reassuring as it demonstrated that patients in care homes with stroke symptoms received a similar clinical assessment to those seen at private addresses. More than 95% received the NHS Ambulance Care Quality stroke assessment bundle (NHS Digital, 2020) of FAST, blood pressure and capillary blood glucose measurements.
A new finding was that the care home group had higher rates of being unable to complete FAST, which appeared to be owing to their reduced level of responsiveness. Additionally, since care home patients typically have more complex health needs, the presence of pre-existing communication and movement difficulties resulting from conditions such as dementia and Parkinson's disease could interfere with assessment (Shah et al, 2010; Wolters et al, 2019).
When a FAST assessment was possible, care home patients were more likely than private address patients to present with facial weakness. This might be related to challenges in assessment or be present already because of common comorbidities such as previous stroke.
Recommendations for research and practice
Although the median longer OST among the care home group was quite small, at a population level the resulting delay in admission to hospital could impact negatively on outcomes.
Future research to explore the factors contributing towards the longer OST in care home patients could be done using a similar research design to that of Li et al (2018), who retrospectively separated and analysed each part of the ambulance care process to understand the components that delayed OST, such as patient extrication. Ambulance clinicians should be made aware that reducing OST is particularly important for care home patients, some of whom are eligible for emergency stroke treatments in hospital despite their higher levels of comorbidities and dependency.
Research should also explore how prehospital stroke assessment is affected by rising age and specific common conditions such as dementia. By linking to hospital data, it would be possible to examine the predictive accuracy of different assessment approaches across various settings and understand how to improve clinical guidelines and training for clinicians.
Limitations
The data represented emergency contacts in a single service over a short time frame, so caution should be used when generalising the results to wider populations.
It was not possible to distinguish between nursing and residential care populations, which could provide further insights into reasons for differences in assessment and treatment compared to private addresses.
The study included only patients with a suspected stroke or TIA as assessed by ambulance clinicians; there was no data to confirm whether a diagnosis of stroke or other condition was made in hospital.
Furthermore, the clinical information was obtained through electronic patient report forms completed by ambulance clinicians during routine care, which may contain errors or omissions, especially given the stress exerted by time-critical situations (Leblanc et al, 2012).
Conclusion
In this first study to specifically investigate prehospital emergency stroke care in different residential settings, it was found that care home patients had more complex background health needs and variations in assessment and care processes.
Clinician awareness of differences between residential settings may be important for improving the delivery of emergency stroke care. Further research is needed to optimise ambulance assessment of care home patients presenting with suspected stroke.