Visual impairment following stroke is common and estimated to affect two-thirds of all stroke survivors (Rowe et al, 2019). However, there is no standardised screening for visual impairment for ambulance services.
A particular issue is posterior circulation stroke, in which visual impairments such as visual field loss, visual inattention and eye movement disorders are common and are often FAST-negative (FAST stands for signs of a stroke, including facial drooping, arm weakness and speech difficulties, as well as time to call emergency services) (Rowe and VIS Writing Group, 2017). No ambulance service screening is provided for all these visual problems so there is the potential for misdiagnosis or missed diagnosis.
Where strokes affect the occipital lobe only, around 90% of patients will have only visual complaints (Isaeff, 1974). It is therefore important to assess specifically for this. Furthermore, although stroke is more likely to be considered as a diagnosis for older people, one in four stroke survivors are of working age (Stroke Association, 2018); they are often easily misdiagnosed where their primary complaint is visual, typically as migraine.
The consequences of mis- or missed diagnosis are that patients are not directed to the appropriate level of stroke care nor receive treatment within the thrombolysis time window, which means the thrombolysis treatment option is not available to them (Brandt et al, 2000). As a result, visual impairment can be permanent, with life-changing disability; its effects on daily life include loss of confidence, impaired mobility, inability to judge distances and an increased risk of falls (Hepworth and Rowe, 2016). There is a known link between poor vision, quality of life and depression in older people (Jones and Shinton, 2006).
For these reasons it is important that patients with visual impairment are identified by the ambulance service to improve diagnostic accuracy and ensure appropriate onward referral.
Visual impairments related to stroke are wide-ranging (Rowe and VIS Writing Group, 2007; Rowe et al, 2019). Therefore, the authors aimed to develop a quick vision screening tool using simple, validated assessments of visual function that identify the most common visual impairments associated with stroke, and coupled this with a supporting education package.
Methods
Ethical approval
This prospective study had institutional ethical approval (reference number 1782) and complied with the Declaration of Helsinki.
Education package
The development process was reported in line with COREQ guidelines (Tong et al, 2007), which are the accepted standard for evaluation of the methodological quality of qualitative research.
Steering committee
During study development, a steering committee was established to oversee its conduct. The committee comprised two research and clinically active orthoptists, one neuro-ophthalmologist, one paramedic and one stroke survivor.
Focus group meetings
Two facilitated focus group meetings were held with experts in stroke and vision research and paramedic practice as well as stroke survivors; they were identified from responses from advertisements about the study through national professional research networks and specialist interest groups.
Information was collected through semi-structured group interviews. In recognition that some people are more vocal than others, a nominal group technique was used. This is a structured method for group brainstorming that encourages equal contributions from everyone. The authors aimed to have 6–12 participants per focus group meeting.
In the first focus group meeting, a nominal group technique with five stages was used:
Once the content of the education package had been agreed, a smaller writing group was set up from focus group participants and steering committee members. They would develop the content into an education package but keep all focus group participants involved in proofing the emerging guide through email contact. The education package emerged through an iterative process.
A second focus group was held to discuss the final draft of the education package and screening tool—the V-FAST tool—using the same nominal group technique to ensure facilitated equal contribution to discussion and decision-making for the final versions.
The final versions of the education package and screening tool were then checked and approved by the senior education team at the North West Ambulance Service (NWAS) and placed on the NWAS online education platform.
Pilot screening tool evaluation
The evaluation process is reported here in accordance with the STARD guidelines (Cohen et al, 2016), which are the accepted standard for evaluation of methodological quality of cross-sectional cohort studies.
Design
A prospective, cross-sectional comparative study was undertaken in two geographically separate ambulance services in the NWAS. Members of this ambulance service had completed the online education package for the V-FAST screening tool. The target population was patients, recruited as a convenience sample, in the hyperacute phase at pre-admission to hospital with a suspected clinical diagnosis of stroke.
Screening protocol
All patients being attended to by the ambulance service following a call-out with suspected stroke were vision-screened by the ambulance service using the V-FAST vision screening tool.
Reference standard assessment
Each patient was assessed as per routine NHS clinical care by an admitting stroke clinician in the emergency department using the NIHSS. Specific aspects extracted from the clinical notes for this study included:
Outcome measures
The primary outcome measure was presence or absence of visual impairment (defined as visual field loss, eye movement abnormality or visual attention abnormality) and recorded as a binary measure: present/absent. Results were taken in numerical format from the referral forms completed by both the ambulance service and NIHSS. These were inputted to a trial database.
Statistical methodology and sample size
For pilot assessment of the screening tool, the authors aimed to recruit a minimum of 30 patients. Results were recorded in binary form (present/absent) for visual impairment and for types of visual impairment.
The authors estimated the level of specificity (proportion of patients without visual impairment who are correctly identified), sensitivity (proportion of patients with visual impairment who are correctly identified), positive predictive value (probability that visual impairment is present when indicated by V-FAST), negative predictive value (probability that visual impairment is absent when indicated by V-FAST) and accuracy (overall probability that the patient is correctly classified with or without visual impairment). Corresponding 95% confidence intervals were provided for all calculations.
Results
Education package
The first focus group had eight participants: the facilitator (FR), one research orthoptist, four paramedics and two stroke survivors. Three other people did not attend because of work or family commitments. The second focus group had seven participants: the facilitator (FR), one research orthoptist, one clinical orthoptist, three paramedics and one stroke survivor. Two other people did not attend because of work commitments or illness.
Each focus group was conducted over a 3-hour period and both were held in Liverpool. Notes were taken by the facilitator during the discussions and responses to the questions were documented on flip charts. The facilitator also took notes.
Each question and summaries of discussion are outlined in Table 1.
| Question 1. What is your current knowledge of potential visual problems that may occur due to stroke? |
|---|
| Responses: Very limited knowledge. Little in the way of assessment and a lack of understanding of visual problems. Reliant on the patient reporting visual symptoms. Split knowledge of paramedics depending on their training and role (e.g. university or other training, and paramedic, technician or other role). No extended history-taking skills. No research evidence. Pupillary response checked. Questioning at point of 999 call |
| Question 2. How would you currently assess patients during call-out for possible visual problems? What do you consider important if you are having your vision assessed in an emergency? |
| Responses. Must show benefit to patients. Need knowledge of anatomy and physiology related to vision. Show the benefits of proposed changes to paramedic assessment. Educate on new tool in line with basic FAST training. Impact on patients and families in a centralised model of care. Include patient stories. Information on why change should be made: why does FAST not catch all strokes, with link to anatomy and physiology? Provide impact on percentage of population. Increase in paramedic knowledge versus public knowledge. Ensure assessment is more accurate than other available options, e.g. West Midlands AVVV (sudden onset ataxia; sudden acute onset visual field defect; vertigo; and vomiting) (West Midlands Ambulance Service, 2016); field assessment but no specific method for FAST-negative patients; South West England use of Miami Emergency Neurologic Deficit (MEND) assessment |
| Question 3. What background vision information do you think should be provided in an education manual? |
| Responses: Include the main facts. Percentage of stroke survivors with visual impairment. Percentage of those missed as having visual impairment by paramedic and/or National Institutes of Health Stroke Scale on admission, plus consequences of being missed. Type of visual impairment and why they occur (anatomy and physiology). Patient stories. Evidence—if paramedics detect there is a visual problem, often the doctors do as well. Include information about what else we know occurs; speed of detection leads to more uptake on thrombolysis/thrombectomy. Provide information about transient ischaemic attack and vision, e.g. amaurosis fugax |
| Question 4. What level of detail is required for description of visual assessment techniques? |
| Responses: Full detail of everything required—spoon feed. Ensure information is not open to interpretation. Include one-page prompt guides. Alternatives to assessment when testing in the patient's home versus when the patient is on a stretcher. Include videos/pictures/YouTube access. Important to include percentage of missed cases in FAST-positive versus negative cases. Potential to link with the Stroke Association regarding basic assessment guide for vision. Potential to provide all training material to GPs at later stage |
| Question 5. What type of tips would be useful in training material? |
| Responses: Avoid use of acronyms. Testing tips already used in orthoptic practice. Picture guides |
| Question 6. What are the key features to include in one-page assessment guides? |
| Responses: As for question 5—tips. Use stepwise information and flowcharts. Mix of text boxes and pictures |
| Question 7, What format of training could be effective, e.g. formal lecture, interactive, problem-based learning etc? |
| Responses: Train the trainer events. Use of pre-reading. Formal lectures using Powerpoint to teach each part of the manual in turn. Include practical. Option to start the training by asking paramedics to read the one-page assessment guides and show how they would assess based on their interpretation of them. This would highlight issues with interpretation and ambiguity, which would allow guides to be altered and refined further for accuracy. Then provide full teaching and repeat the practical testing at the end. Aim to train consultants, advanced practitioners and senior paramedics |
V-FAST screening tool and education package
The screening tool was created as a simple checklist (Appendix 1) alongside a reporting form (Appendix 2). It has five sections:
Core elements of the education package are set out in Table 2 (full details in Appendix 3).
| Background | Visual impairment because of stroke and acquired brain injury |
| Introduction | Screening instructions and assessment |
| Section 1 | History |
| Section 2 | Eye alignment and movement |
| Section 3 | Reading |
| Section 4 | Visual fields |
| Section 5 | Visual inattention/extinction |
| Supplementary | Visual/stroke mimics and transient ischaemic attack |
| Vision anatomy and physiology |
The ambulance service was given detailed instructions regarding the correct use of the assessments required for screening. The tool was designed for self-training. It had detailed instructions on how to use the tool, how to carry out screening and cut-off results that meet normal or abnormal criteria, as well as guidance and tips.
Screening was timed to take around 2 minutes to administer. A video, as part of the education package, showed step-by-step instructions.
The free-to-access V-FAST tool is available at www.vision-research.co.uk.
Pilot screening tool evaluation
The pilot study recruited 43 patients requiring assessment during 999 call-outs for suspected stroke (Figure 1); there were 18 (42%) women and 25 (58%) men. All were admitted to either of two regional hospitals, both of which have adjoining hyperacute and acute stroke units. There was no distinction between referral for FAST-positive or negative patients regarding the hospital they were admitted to. Overall, visual problems were noted on V-FAST assessment in 26 patients (60.5%). In 21 patients (48.8%), only a partial visual assessment could be made because of the patient's ability at the time.
The stroke assessment was FAST positive in 29 patients (67.4%) and FAST negative in five patients (11.6%). This was not recorded for the rest of the patients. Visual problems were present in 22 of 29 (75.9%) FAST-positive patients and in four of five (80%) FAST-negative patients (Table 3A).
| A. Visual impairment in FAST positive/negative tests (n=43) | ||||
|---|---|---|---|---|
| FAST | ||||
| Positive | Negative | Not recorded | ||
| Visual impairment | No | 7 | 1 | 5 |
| Yes | 22 | 4 | 4 | |
Section 1. Visual symptoms and observations
Thirteen patients (30.2%) reported new visual problems and eight (18.6%) reported their vision was different. No visual symptoms were reported in the remainder. Five had specific new visual symptoms recorded, including blurred vision, sudden drop of the right eyelid, pre-existing blindness but with dense visual neglect, right eye turning out, and hazy vision.
Abnormal lid position (i.e. ptosis or asymmetry of right and left lids) was noted in eight patients (18.6%), unequal (right versus left) pupil reactions in four (9.3%), squint/eye turn in four (9.3%), closing one eye in five cases (11.6%) (which could indicate diplopia) and moving head to see in seven cases (16.3%) which could be an indicator for visual field loss.
Section 2. Reading
Impaired reading ability was noted in nine patients (20.9%), with the assessment done with reading glasses if required. It was not determined whether the reading difficulty was eye- or cognition-related.
Section 3. Eye position and movement
Abnormal eye position was noted in five patients (11.6%); two had a downward positioned eye and one patient each had an inward, outward and elevated positioned eye. Abnormal eye movements were noted in 12 patients. There were impaired upgaze in one, impaired downgaze in two, impaired right gaze in one, impaired left gaze in three and nystagmus in five.
Section 4. Visual fields
Abnormal visual fields to confrontation were recorded in eight patients (18.6%). Specific defects included a defect in the right central area, bilateral inferior quadrant defects, general restriction of the visual fields, partial left-sided homonymous hemianopia and two cases of partial right-sided homonymous hemianopia.
Section 5. Visual extinction
Abnormal extinction responses were noted in three patients (7%) to the right side and three patients to the left side (7%).
Emergency department assessment
Stroke diagnosis was confirmed in 25 patients, medical or other causes in 13 patients and transient ischaemic attack in five patients (Table 3B).
An admission assessment was available for 33 patients; the NIHSS was recorded in 26 patients. Visual problems were confirmed in 12 patients (36.4%) during stroke evaluation and for seven patients (26.9%) by NIHSS. The total score on NIHSS was a mean of 7.46 (SD 6.69; range 0–28). Four patients (15.4%) had horizontal gaze issues noted on NIHSS, four had visual field loss (15.4%), and five (19.2%) had extinction/inattention issues noted on the NIHSS.
A comparison of V-FAST assessments by the ambulance service with emergency department visual assessments are outlined in Table 3C. Sensitivity was 85.7% (95% CI [42.1–99.6%]) and specificity was 42.1% (95% CI [20.3–66.5%]). Positive predictive value was 35.3% (95% CI [25.1–47.1%]) and negative predictive value of 88.9% (95% CI [54.7–98.2%]) with overall accuracy of 53.9% (95% CI [33.4–73.4%]).
Discussion
The V-FAST vision screening tool was developed for use by the ambulance service in 999 call-outs for suspected stroke. It is coupled with an education package that includes a video. V-FAST was developed as an adjunct to routine stroke assessment, specifically to improve the detection of vision problems and support added evaluation of potential FAST-negative strokes.
V-FAST is not the first adaptation of FAST to include vision. Table 4 outlines stroke screening checklists that include a vision component; these are not used as standard in the ambulance service.
| Face | Arm | Speech | Coordination/balance | Limb | Consciousness | Vertigo | Vomiting | Ataxia | Eyes | Vision components | |
|---|---|---|---|---|---|---|---|---|---|---|---|
| ABCD-E2 (Gadi and Huwez, 2011; Gulli and Markus, 2012) | x | x | x | x | Blindness, diplopia, pupils | ||||||
| BEFAST (Aroor et al, 2017) | x | x | x | x | x | Blurred vision, double vision, persistent vision trouble | |||||
| FAST AV (Huwez and Casswell, 2013) | x | x | X | x | x | Visual disturbance | |||||
| FAST AB (Huwez and Casswell, 2013) | x | x | x | x | x | Blindness | |||||
| FAST AVVV (West Midlands Ambulance Service, 2016) | x | x | x | x | x | x | x | Visual fields | |||
| Give Me 5 for Stroke (Stroke Collaboration) | x | x | x | x | Vision all or partly lost | ||||||
| MedPACS (Rudd et al, 2016) | |||||||||||
| MEND (Brontons et al, 2018) | x | x | x | x | x | x | Visual fields, horizontal gaze, nystagmus | ||||
| Rosier (Mingfeng et al, 2012) | x | x | x | x | x | Visual fields | |||||
| V-FAST | x | x | x | x | Symptoms, pupils, alignment, movement, visual fields, extinction |
In 2013, FAST AV and FAST AB were developed; AV relates to ataxia and visual disturbance and AB relates to ataxia and blindness. Adding AV or AB to FAST increases sensitivity for early recognition of posterior circulation stroke (Huwez and Casswell, 2013). BE-FAST includes sudden loss of balance or coordination (B) and eyesight changes (E). BE-FAST was reported to reduce the number of missed FAST-negative strokes by about 10% and was particularly relevant to posterior circulation stroke (Aroor et al, 2017). FAST-AVVV, developed in 2016, added acute sudden onset ataxia (A), sudden acute onset visual field defect (V), vertigo (V) and vomiting (V) (West Midlands Ambulance Service, 2016).
Further adaptations include Give Me 5 for Stroke (Stroke Collaborative, 2008), developed in place of FAST. It includes walk, talk, reach, see (‘Is their vision all or partly lost?’) and feel. The ABCD-E2 tool includes ataxia, blindness (unilateral/bilateral), consciousness, dysphagia, eye 1 (diplopia) and eye 2 (pupils) (Gadi and Huwez, 2011; Gulli and Markus, 2012). The Miami Emergency Neurologic Deficit checklist involves 12 assessments including mental status, cranial nerves and limb function (Brotons, 2018). MedPACS (Medical Prehospital Assessment for Code Stroke) includes vision assessment alongside face, arm, leg and speech (Rudd, 2016) as does the ROSIER (Recognition of Stroke in the Emergency Department) assessment tool (Mingfeng et al, 2012).
Although these checklists include an assessment of vision, most were limited in the type of visual deficit being screened. V-FAST was developed on the basis of the most common types of visual impairment detected in two large prospective stroke/vision studies (Rowe and VIS Writing Group, 2017; Rowe et al, 2019) with consensus reached from focus group meetings on the key elements to include in a short (approximate 2-minute) screening assessment. The components of V-FAST include documentation of visual symptoms and observations, checks of eye position and eye movements (horizontal, vertical and nystagmus), visual fields and visual extinction. Although balance and coordination do not have a specific section, the symptoms section asks about sudden onset dizziness and balance issues, which are common questions alongside vision assessments.
V-FAST detected visual impairment in 75.9% of FAST-positive and 80% of FAST-negative stroke diagnoses. Sensitivity and specificity were 85.7% and 42.1% respectively when comparing V-FAST responses to NIHSS visual impairment detection rates. The low specificity is because there were a high number of false positives, which could be because V-FAST can detect a wider range of visual impairment than the NIHSS. The NIHSS is limited regarding visual function and does not consider visual acuity, reading ability, eye alignment and ocular movements. The types of vision problems identified included blurred vision, abnormal lid position, strabismus, eye movement defects, visual field loss and visual neglect. While many of these can be caused by conditions other than stroke, they are also common consequences of stroke (Rowe and VIS Writing Group, 2017; Rowe et al, 2019). The importance of diagnosing these visual problems is that they may be the only presenting sign/symptom of stroke so may affect the initial medical assessment and treatment of the patient with a potential stroke.
FAST is a well-established global tool for early recognition of stroke. It is particularly useful in the detection of anterior circulation strokes. V-FAST is a mnemonic that has been used in awareness messages regarding stroke and vision for many years (Rowe, 2020) and adapted into a formal screening tool in this parametric visual evaluation study. It encapsulates the essence of the FAST message but adds the vision component, which is important given the high incidence (circa 60%) of new onset visual problems in acute stroke (Rowe et al, 2019).
It is essential to continuously strive for improvement in detection accuracy to maximise access to treatment pathways in a timely manner, such as expedited access to hyperacute stroke units, thrombolysis and thrombectomy, but also for secondary prevention management. This is of particular relevance to posterior circulation strokes (Casswell and Huwez, 2011). The potential for misdiagnosis is high with stroke mimics such as migraine, which also presents with sudden onset visual disturbance. Therefore, the education package developed alongside V-FAST is useful in providing background information about stroke mimics and an understanding of the visual system. Arguably, there may be a knock-on positive benefit to using the assessment in any neurological-related assessment other than stroke in addition to its use in eye-related incidents, although this warrants further research.
Limitations
V-FAST was piloted in 43 call-out assessments, and this small sample size is a limitation. This study used a convenience sample so is not representative of a general stroke cohort. Ambulance staff did not use the tool on all patients with suspected stroke and may have been more likely to use it when they suspected a visual problem; this is a recruitment bias.
This was a pragmatic study and, as part of the tool, patients were asked if they were experiencing any new problems with their eyes or vision. However, no orthoptic or ophthalmology reviews were collected; therefore, it is not possible to confirm whether the visual impairments reported were new or old. Types of visual problems documented in this small cohort were, however, typical of those reported in larger studies of stroke-related visual impairment.
In addition, the pilot study was limited to a geographic area. Further studies for formal validation with a larger sample size and larger geographical area are recommended.
V-FAST was compared to stroke and NIHSS assessments made in the emergency department. Further studies could consider comparisons to specialist eye assessments and compare imaging results for anterior versus posterior strokes in relation to FAST-positive or negative strokes with and without visual impairment.
Conclusions
V-FAST is a new, rapid, 2-minute vision screening assessment for use specifically in patients with a suspected stroke. It is supported by an education package with detailed instructions and a video guide. Initial pilot results are promising. Further studies are warranted to determine if sensitivity, specificity and overall accuracy improve in larger studies.