- Email: [email protected]
Accuracy of stroke diagnosis by registered nurses using the ROSIER tool compared to doctors using neurological assessment on a stroke unit: A prospective audit
International Journal of Nursing Studies 48 (2011) 979–985
Contents lists available at ScienceDirect
International Journal of Nursing Studies journal homepage: www.elsevier.com/ijns
Accuracy of stroke diagnosis by registered nurses using the ROSIER tool compared to doctors using neurological assessment on a stroke unit: A prospective audit Bronagh Byrne a, Peter O’Halloran b,*, Christopher Cardwell c a
Stroke Nurse Specialist, Stroke Unit, Level 6, Daisy Hill Hospital, 5 Hospital Road, Newry, Co.Down BT35 8DR, UK Nursing and Midwifery Research Unit, School of Nursing and Midwifery, Queen’s University Belfast, 10 Malone Road, Belfast BT9 5BN, UK Centre for Public Health, School of Medicine, Dentistry and Biomedical Sciences, Institute of Clinical Sciences, Block B, Queens University Belfast, Royal Victoria Hospital, Grosvenor Road, Belfast BT12 6BA, UK b c
A R T I C L E I N F O
A B S T R A C T
Article history: Received 18 August 2010 Received in revised form 10 December 2010 Accepted 28 January 2011
Background: Recombinant tissue plasminogen activator (rT-PA) is an effective treatment for acute ischaemic stroke when given within 3 h of symptom onset but can be delayed as patients wait for a diagnosis. The ROSIER assessment tool (Fig. 1) has been found to be effective in diagnosing stroke but to date has only been tested when used by doctors. Objective: To compare registered nurses’ ability to diagnose stroke using the ROSIER assessment tool with doctors’ ability to diagnose stroke using traditional neurological assessment. Design, setting, and participants: A prospective audit of all suspected stroke patients (n = 106) admitted to the stroke unit of a district general hospital over an eight month period, assessed by registered nurses trained to use the ROSIER assessment tool to identify stroke. Main outcome measures: Time from admission to the stroke unit until initial assessment by doctors and registered nurses. Comparison of initial diagnosis by doctors and registered nurses with final diagnosis by a consultant for stroke. Results: Of 106 suspected stroke patients, 78 (73.5%) had a final diagnosis of stroke or transient ischaemic attack (TIA) and 28 (26.4%) had an alternative diagnosis. Six patients with TIA were subsequently excluded as they were asymptomatic at the time of assessment, leaving 100 participants in the validation phase of the study. Using the ROSIER tool registered nurses achieved a diagnostic sensitivity for stroke of 98% (95% confidence interval 88–99), positive predictive value (PPV) 83% (95% confidence interval 73–90). Doctors using standard neurological assessment had a similar diagnostic sensitivity of 94% (95% confidence interval 86–98), PPV 80% (95% confidence interval 70–88). The mean time from initial assessment by registered nurses using the ROSIER tool, until assessment by doctor on the stroke unit was 75 min (SD = 65.8 min). Conclusions: Registered nurses working on a stroke unit using the ROSIER assessment tool are able to diagnose stroke with a degree of accuracy comparable to doctors using clinical neurological assessment. Prompt assessment of suspected stroke patients by registered nurses using the ROSIER tool could reduce delays in eligible stroke patients being assessed for rT-PA treatment. ß 2011 Elsevier Ltd. All rights reserved.
Keywords: Diagnosis Nursing assessment Sensitivity and specificity Stroke Thrombolytic therapy
What is already known on the topic * Corresponding author. E-mail addresses: [email protected] (B. Byrne), [email protected] (P. O’Halloran). 0020-7489/$ – see front matter ß 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.ijnurstu.2011.01.015
rT-PA is an effective thrombolytic treatment for acute ischaemic stroke when given within 3 h of symptom onset but can be delayed as patients wait for a diagnosis.
980
B. Byrne et al. / International Journal of Nursing Studies 48 (2011) 979–985
The ROSIER assessment tool has been shown to be effective in diagnosing stroke but has only been tested when used by doctors. What this paper adds Using the ROSIER tool registered nurses are able to diagnose stroke with a degree of accuracy comparable to doctors using clinical neurological assessment. Training registered nurses to use the ROSIER tool has the potential to reduce delay in diagnosis of stroke and enable more patients to benefit from rT-PA treatment. 1. Background Every year about 16 million people worldwide suffer a first ever stroke, causing a total of 5.7 million deaths (Strong et al., 2007). In 2005 it was estimated that stroke cost the National Health Service in the United Kingdom approximately £2.8 billion a year in direct care costs, and the wider economy some £1.8 billion more in lost productivity and disability (National Audit Office, 2005). Additionally the annual informal care costs (costs of home nursing and care borne by patients’ families) are around £2.4 billion (National Audit Office, 2005). A systematic review of worldwide stroke incidence and early case fatality over the past four decades found that, although stroke incidence has decreased in high income countries it has increased by more than 100% in low to middle income countries (Feigin et al., 2009). 1.1. Recombinant tissue plasminogen activator (rT-PA) treatment in people with acute ischaemic stroke Traditionally treatment following stroke has concentrated on secondary stroke prevention, physical rehabilitation and management of potential complications (Carrozzella and Jauch, 2002). Given that ischaemic stroke accounts for approximately 75% of all cases (Feigin et al., 2009), the introduction of rT-PA for the treatment of acute ischaemic stroke has the potential to revolutionize acute stroke management and become the expected standard of care. Ischaemic stroke is caused by vessel occlusion, cardioembolic, artery to artery embolism, or in situ small vessel disease (Donnan et al., 2008). Occlusion of a vessel disrupts blood flow to an area of the brain unleashing an ischaemic cascade, leading to death of brain tissue. Use of the fibrinolytic intravenous drug rT-PA in eligible patients with acute ischaemic stroke can result in recanalisation of the affected blood vessel, restoring blood flow to the brain and improving outcomes. In the UK rT-PA is recommended for the treatment of acute ischaemic stroke by the National Institute for Health and Clinical Excellence (NICE), with the proviso that it is used in centres with the necessary facilities and appropriately trained staff (NICE, 2007). In particular, the guidance specifies that rT-PA should be administered within 3 h of onset of symptoms and only after brain haemorrhage has been definitively excluded using brain imaging. Disappointingly, low uptake of this acute treatment is apparent in the UK and elsewhere, largely because
of the difficulties imposed by the 3 h time window (Dawson and Walters, 2005; Donnan and Davis, 2008). However, following the results of the ECASS III trial (Hacke et al., 2008) this 3 h time window has recently been extended to 4.5 h in some regions within the UK (SIGN, 2008). Nevertheless, timely intervention is dependent on an early and accurate diagnosis of stroke, which is best done using a validated assessment tool (NICE, 2008). To date, research into the use of these tools has concentrated on their use by paramedics and doctors. Electronic literature searches of the following databases: Cumulative Medicine Index to Nursing and Allied Health Literature (CINAHL), Pubmed, MEDLINE, Cochrane Database, Journals at Ovid and British Nursing Index (BNI) revealed no articles examining registered nurses’ ability to diagnose stroke using the ROSIER tool. However, when a patient with a suspected stroke is admitted to hospital the first member of staff to make an assessment is usually a registered nurse (Harbinson et al., 2003; Mohd Nor et al., 2005; Hinkle et al., 2007). If registered nurses can effectively use a standard assessment tool to diagnose stroke upon admission to hospital this has the potential to accelerate the assessment process for possible administration of rT-PA. Our aim was to compare the speed and accuracy of diagnosis of stroke by registered nurses in a hospital stroke unit using a standard assessment tool with that of the admitting doctors obtained by performing standard neurological assessment. We chose to use the Recognition of Stroke in the Emergency Room (ROSIER) tool (Mohd Nor et al., 2005) because it has demonstrated higher PPV and higher negative predictive value (NPV) than any of the other stroke recognition tools and is recommended by NICE for use in Accident and Emergency departments (NICE, 2008). 2. Methods 2.1. Objective The objective of the study was to compare registered nurses’ ability to diagnose stroke using the ROSIER assessment tool with doctors’ ability to diagnose stroke using traditional neurological assessment. 2.2. Design and setting This was a prospective audit using routinely collected data following the introduction of the ROSIER assessment tool to the practice of registered nurses in the stroke unit of an acute hospital in Northern Ireland. Registered nurses are those who have passed a course at a higher education institution in pre-registration nursing, leading to registration with the Nursing and Midwifery Council in the UK. Twenty registered nurses, with nursing experience ranging between six months and 30 years, were trained in the use of the ROSIER assessment tool (Mohd Nor et al., 2005) by BB. Training was carried out with small groups of three to four registered nurses and involved a PowerPoint presentation and practical demonstration of the ROSIER tool using actual stroke patients with their consent. The ROSIER can be taught in 15 min and performed in 5–10 min (Mohd Nor et al., 2005). The admitting doctors used standard neurological
B. Byrne et al. / International Journal of Nursing Studies 48 (2011) 979–985
assessment to assess the patients. Depending on the experience of the doctor this can take between 10 and 20 min. One of the admitting doctors was a ‘staff’ grade (that is a non-consultant middle-grade doctor, in this case with more than 15 years experience) and the remainder either in the first or second years of a ‘Foundation Programme’ (the overall training structure for the first two years of clinical practice post-medical school). The final, ‘gold-standard’ diagnosis was made by either of two medical consultants for stroke. The audit took place before the introduction of rT-PA as a standard treatment option in the hospital. 2.3. Participants and eligibility criteria Patients with a suspected stroke or transient ischaemic attack (TIA) were admitted to the stroke unit following medical assessment in the Accident and Emergency
department or following direct telephone referral by primary care doctors (but prior to brain imaging). The following were excluded from the audit: patients who had a stroke or TIA whilst an in-patient in hospital and patients admitted to the stroke unit from another ward or hospital with an established diagnosis of stroke or TIA. 2.4. Outcomes and data collection Patients admitted to the stroke unit were assessed by the admitting registered nurse using the ROSIER tool. This tool requires the assessor to score for the presence or absence of a number of common clinical features, including loss of consciousness or syncope; seizure activity; asymmetric facial or limb weakness; speech disturbance; and visual field defect (Fig. 1). A validation study for this tool demonstrated that, for a stroke cut-off rating of +1 or above, the ROSIER
ROSIER scale proforma
Assessment
Date:
Symptom onset
Date Time
Time:
1
GGS E=□
M= □ V= □
BP
*B BM
* If BM < 3.5 mmol/l treat urgently and reassess once blood glucose normal Has there been loss of consciousness or syncope?
Y (-1) □
N (0) □
Has there been seizure activity?
Y (-1) □
N (0) □
I. Asymmetric facial weakness
Y (+1) □
N (0) □
II. Asymmetric arm weakness
Y (+1) □
N (0) □
III. Asymmetric leg weakness
Y (+1) □
N (0) □
IV. Speech disturbance
Y (+1) □
N (0) □
V. Visual field defect
Y (+1) □
N (0) □
Is there a NEW ACUTE onset (or on awakening from sleep)
*Total Score _____ (-2 to +5) Provisional diagnosis: □ Stroke
□
981
Non-stroke (specify)
___________________
*Stroke is unlikely but not completely excluded if total scores are ≤ 0.
BM=blood glucose; BP=blood pressure (mm Hg); GCS=Glasgow Coma Scale; E=eye; M=motor; V=verbal component. Fig. 1. ROSIER scale proforma.
982
B. Byrne et al. / International Journal of Nursing Studies 48 (2011) 979–985
scale had a sensitivity of 93%, (95% confidence interval 89– 97), specificity 83% (77–89), PPV 90% (95% confidence interval 85–95), and NPV 88% (95% confidence interval 83– 93) (Mohd Nor et al., 2005). In our study, if a score of one to five was obtained, a patient was considered to have had a stroke or TIA. The doctor on duty was then informed of the patient’s arrival and condition but was blind to the ROSIER score at this stage. The admitting doctor then made a provisional diagnosis using clinical neurological assessment, as standardized in a local Stroke Integrated Care Pathway document. The pathway offered the admitting doctor four options as to how confident they were that the patient had a stroke: ‘definite’, ‘probable’, ‘possible’ or ‘other’. The first three options were regarded as a positive diagnosis of stroke or TIA. Data were collected from the notes of all patients admitted to the stroke unit with suspected stroke or TIA between July 2008 and February 2009 by BB using a predesigned audit form. This included basic demographic data; previous stroke history; time of onset of symptoms; route and time of admission; time and outcome of assessment by the admitting doctor and outcome of assessment by nursing staff using the ROSIER tool; and the final medical diagnosis. The final medical diagnosis was made by either of two consultants for stroke based on clinical findings and the result of brain imaging, normally within 24 h of admission.
Table 1 Demographic details and final diagnosis. Variable
Stroke/TIA (n = 78 (%))
No stroke (n = 28 (%))
Age years Mean (SD) Men Route of admission A&E Primary care Previous history of stroke Stroke classification Total anterior circulation stroke Partial anterior circulation stroke Lacunar stroke Posterior circulation stroke Primary intracerebral haemorrhage TIA
72 (14.6) 47 (60.3)
64 (21) 16 (57.1)
42 (53.8) 37 (47.4) 19 (24.4)
15 (53.6) 12 (42.9) 15 (53.6)
15 (19.2) 22 (28.2) 12 (15.4) 14 (17.9) 6 (7.7) 9 (11.5)
Time of assessment by doctor on the stroke unit was recorded for 55 (51%) of the original 106 patients. Registered nurses recorded time of onset of patients’ symptoms and time of completion of ROSIER on 103 (97%) patients. Where documented, the mean time from initial assessment by nurses using the ROSIER tool, until doctors assessment on the stroke unit was 75 min (SD = 65.8 min). 3.2. Final diagnosis
2.5. Sample size and analysis In a sample of 100 patients admitted to the stroke unit with suspected stroke or TIA we would expect approximately 80 to have had a stroke or TIA providing a 95% probability of estimating the true sensitivity of the ROSIER tool to within 7%, based upon an approximate sensitivity of 90%, as measured by Mohd Nor et al. (2005). The optimum cut-point for stroke diagnosis was determined to be a total ROSIER score of one or above (Mohd Nor et al., 2005). Data were entered into SPSS version 14.0 for Windows and a descriptive analysis performed. Sensitivity, PPV and exact binomial 95% confidence intervals were calculated using Stata 10 for Windows. 3. Results Data were collected from the notes of 106 patients with suspected stroke or TIA. Of these, 28 patients had an alternative diagnosis and 78 a final diagnosis of stroke or TIA (Table 1). 3.1. Time from onset of symptoms to assessment Of the original 106 suspected stroke and TIA patients, 26 (24.5%) presented to hospital within 3 h of symptom onset. Of 78 patients with a final diagnosis of stroke or TIA, 63 (80.7%) had an ischaemic stroke. If the time for availing of rT-PA treatment had been set at 3 h from onset of symptoms, 19 of 63 (30.2%) patients would have been eligible for treatment. Extending the window to 4.5 h would increase this number to 23 (36.5%). Thus, even with a 4.5 h window, 40 patients (63.5%) would have been ineligible for rT-PA treatment.
Following medical consultant assessment and brain imaging of all 106 patients, 63 were diagnosed with ischaemic stroke, nine with TIA, six with intracerebral haemorrhage and 28 patients had an alternative diagnosis. All 106 suspected stroke or TIA patients were assessed by registered nurses using the ROSIER tool. Of the nine patients with a final diagnosis of TIA, six had no neurological signs or symptoms when assessed by registered nurses using the ROSIER and scored zero. The ROSIER has been shown to be insensitive for diagnosing asymptomatic TIA (Mohd Nor et al., 2005), therefore these six patients were excluded from the validation phase leaving a total of 100 study patients. Nine of the 106 suspected stroke cases had a final diagnosis of seizure. Todd’s paresis following seizures has been recognised as one of the most common conditions that produce stroke like signs and symptoms (Somes and Bergman, 2007). Registered nurses using the ROSIER correctly identified six out of nine of the patients with a final diagnosis of seizure compared to two out of nine identified by doctors using traditional neurological assessment. 3.3. Diagnostic performance As noted above, our analysis of diagnostic performance (which excludes the six patients with TIA who had no neurological signs or symptoms when assessed by registered nurses) is based on data from 100 patients. Diagnostic performance by referring doctors working in Accident and Emergency using standard neurological assessment was, sensitivity 97% (95% confidence intervals 87–100), PPV 84% (95% confidence intervals 71–94). It is
B. Byrne et al. / International Journal of Nursing Studies 48 (2011) 979–985
983
Table 2 Time to assessment and diagnostic performance of admitting medical and nursing staff on the stroke unit. Variable
Doctors
Nurses
Median/mean time from stroke unit admission to assessment in minutes (SD)
65/93 (SD 93.5)
Mean = 10/11 (SD 8.1)
Initial diagnosis
Identified with stroke
Final diagnosis: stroke Final diagnosis: no stroke Diagnostic specificity (95%CI) Negative predictive value (95% CI) Diagnostic sensitivity (95% CI) Positive predictive value (95% CI) a
68 17 39% 73% 94% 80%
(22%, (45%, (86%, (70%,
59%) 92%) 98%) 88%)
Not identified with stroke 4a 11
Identified with stroke 69 14 50% 82% 96% 83%
(31%, (57%, (88%, (73%,
Not identified with stroke 3 14
69%) 96%) 99%) 90%)
Includes one unrecorded case assumed to be a false negative.
not known what method of assessment referring primary care doctors utilized to diagnose stroke but their diagnostic performance was, sensitivity 94% (95% confidence intervals 80–99), PPV 82% (95% confidence intervals 66–92). For admitting doctors using traditional neurological assessment, diagnostic sensitivity was 94% (95% CI 86–98), and PPV 80% (95% CI, 70–88). At the optimum cut-point for stroke diagnosis using the ROSIER tool, i.e. a total score of one or above the corresponding diagnostic performance for registered nurses using the ROSIER was a sensitivity of 98% (95% CI, 88–99), and a PPV of 83% (95% CI, 73–90) (Table 2). Admitting doctors made incorrect assessments in 21 cases. Of these, 17 were false positive cases (assessed as stroke but had a final diagnosis of non-stroke) and four false negative cases (assessed as non-stroke diagnosis but had a final diagnosis of stroke). This included one unrecorded case which was assumed to be a false negative, as a CT image of the brain was not ordered until a consultant reviewed the patient later (it is standard practice for brain imaging to be ordered upon admission if a person is suspected to have had a stroke or TIA). Three of the false negatives were lacunar strokes, and one a partial anterior circulation stroke. Nurses made incorrect assessments in 17 cases. Of these, 14 were false positive cases and three false negative cases. Two of the false negatives were lacunar strokes, and one a partial anterior circulation stroke. Admitting doctors and nurses concurred on 13 of the incorrect assessments. 4. Discussion To our knowledge, this study is the first to examine the ability of registered nurses to use the ROSIER tool to identify patients with stroke. Our results show that the ROSIER tool had a sensitivity of 98% (95% confidence intervals 88–99%) and PPV of 83% (95% confidence intervals 73–90%) when used by registered nurses on a stroke unit. The sensitivity achieved by registered nurses was similar to that achieved by doctors using traditional neurological assessment in this study 94% (95% CI; 86–98), and similar to the sensitivity of 93% (95% CI: 89–97) achieved by doctors in an emergency department in the original validation study (Mohd Nor et al., 2005). However, there were insufficient numbers of non-stroke patients in this study to be able to accurately estimate the specificity and NPV.
A further limitation of this study is that the sample was of patients who already had a possible diagnosis of stroke. However, 28 (26%) of these patients eventually had an alternative diagnosis compared with an estimated norm of one in five (Harraf et al., 2002) and fourteen (50%) of these false positives were correctly identified by nurses using the ROSIER as conditions mimicking stroke. It could be argued that registered nurses employed on a stroke unit have specialised skills in relation to assessment of stroke signs and symptoms compared to registered nurses employed in other departments and so would be more adept at using the ROSIER tool. If this were the case the results from this study would not be widely generalisable to other departments. However, although registered nurses in this study were familiar with caring for patients with stroke, assessing for neurological signs commonly associated with stroke had not been part of their role and they required training in the use of the ROSIER tool. Items contained within the tool are easily taught and should be well within the capabilities of registered nurses employed in any department. Of the original 106 patients assessed using the ROSIER tool, nine (8.4%) had a final diagnosis of TIA. Six of the patients’ initial signs and symptoms had resolved at time of assessment by registered nurses and as a result the registered nurse diagnosed them as non-stroke. Doctors positively diagnosed six of the nine patients with TIA using standard neurological assessment and through obtaining the medical history. A typical TIA resolves within an hour (Albers et al., 2002), so correct diagnosis depends on acquiring an accurate medical history and appropriate neuroimaging (Lewandowski et al., 2008). Although the ROSIER extracts some history items, these focus on excluding conditions that mimic stroke rather than eliciting information which could guide TIA diagnosis. This has been acknowledged as a weakness of the tool by its developers, who recommend development of a separate tool to diagnose TIA (Mohd Nor et al., 2005). On the other hand, registered nurses using the ROSIER correctly identified six out of nine of the patients with a final diagnosis of seizure compared to two out of nine identified by doctors using traditional neurological assessment, perhaps indicating the strength of the ROSIER in identifying seizure as a stroke mimic. Using the ROSIER tool, registered nurses incorrectly diagnosed 17 cases (17%). Doctors’ incorrectly diagnosed
984
B. Byrne et al. / International Journal of Nursing Studies 48 (2011) 979–985
21 cases (21%). False positive rates for doctors and nurses, 17% and 14%, respectively, were slightly higher than the 10% found in the prospective validation phase described by Mohd Nor et al. (2005). It is possible that both doctors and nurses were influenced by the provisional diagnosis of stroke suggested by the referring doctor in primary care or in Accident and Emergency. Conversely these false positive rates were significantly lower than the 31% found in a study that examined hospital-based doctors’ ability to distinguish between stroke and stroke mimic using standard clinical assessment (Hand et al., 2006). Furthermore if the ROSIER score obtained by registered nurses had been used in combination with the doctors’ neurological assessment, the number of false positives could potentially have been reduced by five (four seizures and one syncope) bringing the false positive rate to 11% in line with Mohd Nor et al. (2005). Of the original 106 suspected stroke and TIA patients, 26 (24%) presented to hospital within 3 h of symptom onset. Included in this figure are the 63 patients suffering an ischaemic stroke, of whom 19 (30.2%) presented to hospital within 3 h of symptom onset. This finding is similar to other studies carried out in the UK, with proportions ranging from 25% to 37% (Johnson et al., 1999; Harraf et al., 2002) but lower than findings from studies in the United States (US) where proportions have ranged from 30% to 56% (Kothari et al., 1999; Evenson et al., 2001; Lacy et al., 2001). In the US rT-PA has been licensed for use since 1995 and there have been mass public education programmes emphasising the importance of early arrival to hospital following onset of stroke symptoms. The registered nurse assessment using the ROSIER took place on average 75 min before assessment by the admitting doctor, although the patients had already been assessed by doctors in primary care or in A&E. Time of assessment by doctor on the stroke unit was recorded for 55 (51%) of the original 106 patients. These findings may be partly explained by the timing of this audit which was prior to introduction of active administration of rT-PA for acute ischaemic stroke. As the service was not yet operational doctors may not have considered recording time of assessment as essential. We envisage an rT-PA service where the registered nurse, having obtained a positive score for stroke using the ROSIER, will immediately initiate a paging system which will warrant an immediate response from a member of an emergency response stroke team. This team member will assess the suspected stroke patient regarding their suitability for rTPA treatment. 4.1. Conclusion and implications On admission to hospital, whether to Accident and Emergency, a stroke unit or medical admissions unit, the first professional to assess a stroke patient is usually a registered nurse. This audit shows that registered nurses are able to quickly and effectively assess suspected stroke patients with an acceptable degree of accuracy. Given the importance of early intervention with rt-PA treatment, our study suggests that if registered nurses working in areas
that receive the acute stroke patient were trained to use the ROSIER as a basis for triaging patients, the time from admission to treatment could be minimized and the numbers of patients treated maximized. A larger study to establish the accuracy of assessment of patients with suspected stroke by registered nurses using the ROSIER in Accident and Emergency departments is recommended as this is likely to further reduce time from admission to treatment.Conflict of interest None declared. Funding None. Ethical approval The study proposal was submitted to the hospital’s Research Governance Body and classified as a service evaluation, so formal ethical approval was not sought. Permission to access patients’ notes was granted by consultant medical staff and the relevant senior managers in the hospital.
References Albers, G.W., Caplan, L.R., Easton, J.D., Fayad, P.B., Mohr, J.P., Saver, J.L., et al., 2002. Transient ischemic attack: proposal for a new definition. New England Journal of Medicine 347, 1713–1716. Carrozzella, J., Jauch, E.C., 2002. Emergency stroke management. Emergency Nursing 37, 35–57. Dawson, J., Walters, M., 2005. Development and validation of a stroke recognition tool. Lancet 4, 691–693. Donnan, G., Davis, S., 2008. Breaking the 3 h barrier for treatment. Lancet Neurology 7, 981–982. Donnan, G., Fisher, M., MacLeod, M., Davis, S., 2008. Stroke. The Lancet 371, 1612–1623. Evenson, K., Rosamond, W., Morris, D., 2001. Prehospital and in-hospital delays in acute stroke care. Neuroepidemiology 20, 65–76. Feigin, V.L., Lawes, C.M.M., Bennett, D.A., Barker-Collo, S.L., Parag, V., 2009. Worldwide stroke incidence and early case fatality reported in 56 population-based studies: a systematic review. Lancet Neurology 8, 355–369. Hacke, W., Donnan, G., Fieschi, C., Kaste, M., von Kummer, R., Broderick, J.P., et al., 2008. Thrombolysis with Alteplase 3–4.5 h after acute ischemic stroke. New England Journal of Medicine 359, 1317–1329. Hand, P.J., Kwan, J., Lindley, R.I., Dennis, M.S., Wardlaw, J.M., 2006. Distinguishing between stroke and mimic at the bedside. The Brain Attack Study. Stroke 37 (3), 769–775. Harraf, F., Sharma, A.K., Brown, M.M., Lees, K.R., Vass, R.I., Kalra, L., 2002. A multicentre observational study of presentation and early assessment of acute stroke. British Medical Journal 325, 17–20. Harbinson, J., Hossain, O., Jenkinson, D., Davis, J., Louw, S.J., Ford, G.A., 2003. Diagnostic accuracy of stroke referrals from primary care. Emergency room physicians, and ambulance staff using the face arm speech test. Stroke 34, 71–76. Hinkle, J.L., Manoj, A., Brook, S., Webb, A., Muangpaisan, W., Kennedy, J., Buchan, A.M., 2007. Nursing triage of patients with acute stroke. Nursing Times 103, 32–33. Johnson, F., Wardlaw, J., Dennis, M.S., Lewis, S., Nimmo, G., Lindley, R.I., Warlow, C.P., 1999. Delays in stroke referrals. Lancet 354, 47–48. Kothari, R., Jaunch, E., Broderick, J., Brott, T., Sauerbeck, L., Khoury, J., Liu, T., 1999. Acute stroke: delays to presentation and emergency department evaluation. Annals of Emergency Medicine 33, 3–8. Lacy, C.R., Suh, C.C., Bueno, M., Kostis, J.B., 2001. Delay in presentation and evaluation for acute stroke: stroke time registry for outcomes knowledge and epidemiology (S.T.R.O.K.E). Stroke 32, 63–69. Lewandowski, C.A., Rao, C.P.V., Silver, B., 2008. Transient ischemic attack: definitions and clinical presentations. Annals of Emergency Medicine 52, S7–S16. Mohd Nor, A.M., Davis, J., Sen, B., Shipsey, D., Louw, S.J., Dyker, A.G., et al., 2005. The recognition of stroke in the emergency room (ROSIER) scale, development and validation of a stroke recognition instrument. Lancet 4, 727–734.
B. Byrne et al. / International Journal of Nursing Studies 48 (2011) 979–985 National Audit Office, 2005, 2005. Reducing Brain Damage: Faster Access to Better Stroke Care. Department Of Health, London. NICE, 2007. Alteplase for the Treatment of Acute Ischaemic Stroke. National Institute for Health and Clinical Excellence, London Available from: http://www.nice.org.uk/nicemedia/pdf/TA122guidance. pdf. NICE, 2008. Stroke: Diagnosis and Initial Management of Acute Stroke And Transient Ischaemic Attack. National Institiute Health and Clinical Excellence, London Available from: http://www.nice.org.uk/ nicemedia/pdf/CG68NICEGuideline.pdf.
985
SIGN (Scottish Intercollegiate Guidelines Network. & NHS Quality Improvement Scotland), 2008. Management of Patients with Stroke or TIA: Assessment, Investigation, Immediate Management and Secondary Prevention: A National Clinical Guideline. Scottish Intercollegiate Guidelines Network, Edinburgh. Somes, J., Bergman, D.L., 2007. ABCDs of acute stroke intervention. Journal of Emergency Nursing 33 (3), 228–234. Strong, K., Mathers, C., Bonita, R., 2007. Preventing stroke: saving lives around the world. Lancet Neurology 6, 182–187.
Contents lists available at ScienceDirect
International Journal of Nursing Studies journal homepage: www.elsevier.com/ijns
Accuracy of stroke diagnosis by registered nurses using the ROSIER tool compared to doctors using neurological assessment on a stroke unit: A prospective audit Bronagh Byrne a, Peter O’Halloran b,*, Christopher Cardwell c a
Stroke Nurse Specialist, Stroke Unit, Level 6, Daisy Hill Hospital, 5 Hospital Road, Newry, Co.Down BT35 8DR, UK Nursing and Midwifery Research Unit, School of Nursing and Midwifery, Queen’s University Belfast, 10 Malone Road, Belfast BT9 5BN, UK Centre for Public Health, School of Medicine, Dentistry and Biomedical Sciences, Institute of Clinical Sciences, Block B, Queens University Belfast, Royal Victoria Hospital, Grosvenor Road, Belfast BT12 6BA, UK b c
A R T I C L E I N F O
A B S T R A C T
Article history: Received 18 August 2010 Received in revised form 10 December 2010 Accepted 28 January 2011
Background: Recombinant tissue plasminogen activator (rT-PA) is an effective treatment for acute ischaemic stroke when given within 3 h of symptom onset but can be delayed as patients wait for a diagnosis. The ROSIER assessment tool (Fig. 1) has been found to be effective in diagnosing stroke but to date has only been tested when used by doctors. Objective: To compare registered nurses’ ability to diagnose stroke using the ROSIER assessment tool with doctors’ ability to diagnose stroke using traditional neurological assessment. Design, setting, and participants: A prospective audit of all suspected stroke patients (n = 106) admitted to the stroke unit of a district general hospital over an eight month period, assessed by registered nurses trained to use the ROSIER assessment tool to identify stroke. Main outcome measures: Time from admission to the stroke unit until initial assessment by doctors and registered nurses. Comparison of initial diagnosis by doctors and registered nurses with final diagnosis by a consultant for stroke. Results: Of 106 suspected stroke patients, 78 (73.5%) had a final diagnosis of stroke or transient ischaemic attack (TIA) and 28 (26.4%) had an alternative diagnosis. Six patients with TIA were subsequently excluded as they were asymptomatic at the time of assessment, leaving 100 participants in the validation phase of the study. Using the ROSIER tool registered nurses achieved a diagnostic sensitivity for stroke of 98% (95% confidence interval 88–99), positive predictive value (PPV) 83% (95% confidence interval 73–90). Doctors using standard neurological assessment had a similar diagnostic sensitivity of 94% (95% confidence interval 86–98), PPV 80% (95% confidence interval 70–88). The mean time from initial assessment by registered nurses using the ROSIER tool, until assessment by doctor on the stroke unit was 75 min (SD = 65.8 min). Conclusions: Registered nurses working on a stroke unit using the ROSIER assessment tool are able to diagnose stroke with a degree of accuracy comparable to doctors using clinical neurological assessment. Prompt assessment of suspected stroke patients by registered nurses using the ROSIER tool could reduce delays in eligible stroke patients being assessed for rT-PA treatment. ß 2011 Elsevier Ltd. All rights reserved.
Keywords: Diagnosis Nursing assessment Sensitivity and specificity Stroke Thrombolytic therapy
What is already known on the topic * Corresponding author. E-mail addresses: [email protected] (B. Byrne), [email protected] (P. O’Halloran). 0020-7489/$ – see front matter ß 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.ijnurstu.2011.01.015
rT-PA is an effective thrombolytic treatment for acute ischaemic stroke when given within 3 h of symptom onset but can be delayed as patients wait for a diagnosis.
980
B. Byrne et al. / International Journal of Nursing Studies 48 (2011) 979–985
The ROSIER assessment tool has been shown to be effective in diagnosing stroke but has only been tested when used by doctors. What this paper adds Using the ROSIER tool registered nurses are able to diagnose stroke with a degree of accuracy comparable to doctors using clinical neurological assessment. Training registered nurses to use the ROSIER tool has the potential to reduce delay in diagnosis of stroke and enable more patients to benefit from rT-PA treatment. 1. Background Every year about 16 million people worldwide suffer a first ever stroke, causing a total of 5.7 million deaths (Strong et al., 2007). In 2005 it was estimated that stroke cost the National Health Service in the United Kingdom approximately £2.8 billion a year in direct care costs, and the wider economy some £1.8 billion more in lost productivity and disability (National Audit Office, 2005). Additionally the annual informal care costs (costs of home nursing and care borne by patients’ families) are around £2.4 billion (National Audit Office, 2005). A systematic review of worldwide stroke incidence and early case fatality over the past four decades found that, although stroke incidence has decreased in high income countries it has increased by more than 100% in low to middle income countries (Feigin et al., 2009). 1.1. Recombinant tissue plasminogen activator (rT-PA) treatment in people with acute ischaemic stroke Traditionally treatment following stroke has concentrated on secondary stroke prevention, physical rehabilitation and management of potential complications (Carrozzella and Jauch, 2002). Given that ischaemic stroke accounts for approximately 75% of all cases (Feigin et al., 2009), the introduction of rT-PA for the treatment of acute ischaemic stroke has the potential to revolutionize acute stroke management and become the expected standard of care. Ischaemic stroke is caused by vessel occlusion, cardioembolic, artery to artery embolism, or in situ small vessel disease (Donnan et al., 2008). Occlusion of a vessel disrupts blood flow to an area of the brain unleashing an ischaemic cascade, leading to death of brain tissue. Use of the fibrinolytic intravenous drug rT-PA in eligible patients with acute ischaemic stroke can result in recanalisation of the affected blood vessel, restoring blood flow to the brain and improving outcomes. In the UK rT-PA is recommended for the treatment of acute ischaemic stroke by the National Institute for Health and Clinical Excellence (NICE), with the proviso that it is used in centres with the necessary facilities and appropriately trained staff (NICE, 2007). In particular, the guidance specifies that rT-PA should be administered within 3 h of onset of symptoms and only after brain haemorrhage has been definitively excluded using brain imaging. Disappointingly, low uptake of this acute treatment is apparent in the UK and elsewhere, largely because
of the difficulties imposed by the 3 h time window (Dawson and Walters, 2005; Donnan and Davis, 2008). However, following the results of the ECASS III trial (Hacke et al., 2008) this 3 h time window has recently been extended to 4.5 h in some regions within the UK (SIGN, 2008). Nevertheless, timely intervention is dependent on an early and accurate diagnosis of stroke, which is best done using a validated assessment tool (NICE, 2008). To date, research into the use of these tools has concentrated on their use by paramedics and doctors. Electronic literature searches of the following databases: Cumulative Medicine Index to Nursing and Allied Health Literature (CINAHL), Pubmed, MEDLINE, Cochrane Database, Journals at Ovid and British Nursing Index (BNI) revealed no articles examining registered nurses’ ability to diagnose stroke using the ROSIER tool. However, when a patient with a suspected stroke is admitted to hospital the first member of staff to make an assessment is usually a registered nurse (Harbinson et al., 2003; Mohd Nor et al., 2005; Hinkle et al., 2007). If registered nurses can effectively use a standard assessment tool to diagnose stroke upon admission to hospital this has the potential to accelerate the assessment process for possible administration of rT-PA. Our aim was to compare the speed and accuracy of diagnosis of stroke by registered nurses in a hospital stroke unit using a standard assessment tool with that of the admitting doctors obtained by performing standard neurological assessment. We chose to use the Recognition of Stroke in the Emergency Room (ROSIER) tool (Mohd Nor et al., 2005) because it has demonstrated higher PPV and higher negative predictive value (NPV) than any of the other stroke recognition tools and is recommended by NICE for use in Accident and Emergency departments (NICE, 2008). 2. Methods 2.1. Objective The objective of the study was to compare registered nurses’ ability to diagnose stroke using the ROSIER assessment tool with doctors’ ability to diagnose stroke using traditional neurological assessment. 2.2. Design and setting This was a prospective audit using routinely collected data following the introduction of the ROSIER assessment tool to the practice of registered nurses in the stroke unit of an acute hospital in Northern Ireland. Registered nurses are those who have passed a course at a higher education institution in pre-registration nursing, leading to registration with the Nursing and Midwifery Council in the UK. Twenty registered nurses, with nursing experience ranging between six months and 30 years, were trained in the use of the ROSIER assessment tool (Mohd Nor et al., 2005) by BB. Training was carried out with small groups of three to four registered nurses and involved a PowerPoint presentation and practical demonstration of the ROSIER tool using actual stroke patients with their consent. The ROSIER can be taught in 15 min and performed in 5–10 min (Mohd Nor et al., 2005). The admitting doctors used standard neurological
B. Byrne et al. / International Journal of Nursing Studies 48 (2011) 979–985
assessment to assess the patients. Depending on the experience of the doctor this can take between 10 and 20 min. One of the admitting doctors was a ‘staff’ grade (that is a non-consultant middle-grade doctor, in this case with more than 15 years experience) and the remainder either in the first or second years of a ‘Foundation Programme’ (the overall training structure for the first two years of clinical practice post-medical school). The final, ‘gold-standard’ diagnosis was made by either of two medical consultants for stroke. The audit took place before the introduction of rT-PA as a standard treatment option in the hospital. 2.3. Participants and eligibility criteria Patients with a suspected stroke or transient ischaemic attack (TIA) were admitted to the stroke unit following medical assessment in the Accident and Emergency
department or following direct telephone referral by primary care doctors (but prior to brain imaging). The following were excluded from the audit: patients who had a stroke or TIA whilst an in-patient in hospital and patients admitted to the stroke unit from another ward or hospital with an established diagnosis of stroke or TIA. 2.4. Outcomes and data collection Patients admitted to the stroke unit were assessed by the admitting registered nurse using the ROSIER tool. This tool requires the assessor to score for the presence or absence of a number of common clinical features, including loss of consciousness or syncope; seizure activity; asymmetric facial or limb weakness; speech disturbance; and visual field defect (Fig. 1). A validation study for this tool demonstrated that, for a stroke cut-off rating of +1 or above, the ROSIER
ROSIER scale proforma
Assessment
Date:
Symptom onset
Date Time
Time:
1
GGS E=□
M= □ V= □
BP
*B BM
* If BM < 3.5 mmol/l treat urgently and reassess once blood glucose normal Has there been loss of consciousness or syncope?
Y (-1) □
N (0) □
Has there been seizure activity?
Y (-1) □
N (0) □
I. Asymmetric facial weakness
Y (+1) □
N (0) □
II. Asymmetric arm weakness
Y (+1) □
N (0) □
III. Asymmetric leg weakness
Y (+1) □
N (0) □
IV. Speech disturbance
Y (+1) □
N (0) □
V. Visual field defect
Y (+1) □
N (0) □
Is there a NEW ACUTE onset (or on awakening from sleep)
*Total Score _____ (-2 to +5) Provisional diagnosis: □ Stroke
□
981
Non-stroke (specify)
___________________
*Stroke is unlikely but not completely excluded if total scores are ≤ 0.
BM=blood glucose; BP=blood pressure (mm Hg); GCS=Glasgow Coma Scale; E=eye; M=motor; V=verbal component. Fig. 1. ROSIER scale proforma.
982
B. Byrne et al. / International Journal of Nursing Studies 48 (2011) 979–985
scale had a sensitivity of 93%, (95% confidence interval 89– 97), specificity 83% (77–89), PPV 90% (95% confidence interval 85–95), and NPV 88% (95% confidence interval 83– 93) (Mohd Nor et al., 2005). In our study, if a score of one to five was obtained, a patient was considered to have had a stroke or TIA. The doctor on duty was then informed of the patient’s arrival and condition but was blind to the ROSIER score at this stage. The admitting doctor then made a provisional diagnosis using clinical neurological assessment, as standardized in a local Stroke Integrated Care Pathway document. The pathway offered the admitting doctor four options as to how confident they were that the patient had a stroke: ‘definite’, ‘probable’, ‘possible’ or ‘other’. The first three options were regarded as a positive diagnosis of stroke or TIA. Data were collected from the notes of all patients admitted to the stroke unit with suspected stroke or TIA between July 2008 and February 2009 by BB using a predesigned audit form. This included basic demographic data; previous stroke history; time of onset of symptoms; route and time of admission; time and outcome of assessment by the admitting doctor and outcome of assessment by nursing staff using the ROSIER tool; and the final medical diagnosis. The final medical diagnosis was made by either of two consultants for stroke based on clinical findings and the result of brain imaging, normally within 24 h of admission.
Table 1 Demographic details and final diagnosis. Variable
Stroke/TIA (n = 78 (%))
No stroke (n = 28 (%))
Age years Mean (SD) Men Route of admission A&E Primary care Previous history of stroke Stroke classification Total anterior circulation stroke Partial anterior circulation stroke Lacunar stroke Posterior circulation stroke Primary intracerebral haemorrhage TIA
72 (14.6) 47 (60.3)
64 (21) 16 (57.1)
42 (53.8) 37 (47.4) 19 (24.4)
15 (53.6) 12 (42.9) 15 (53.6)
15 (19.2) 22 (28.2) 12 (15.4) 14 (17.9) 6 (7.7) 9 (11.5)
Time of assessment by doctor on the stroke unit was recorded for 55 (51%) of the original 106 patients. Registered nurses recorded time of onset of patients’ symptoms and time of completion of ROSIER on 103 (97%) patients. Where documented, the mean time from initial assessment by nurses using the ROSIER tool, until doctors assessment on the stroke unit was 75 min (SD = 65.8 min). 3.2. Final diagnosis
2.5. Sample size and analysis In a sample of 100 patients admitted to the stroke unit with suspected stroke or TIA we would expect approximately 80 to have had a stroke or TIA providing a 95% probability of estimating the true sensitivity of the ROSIER tool to within 7%, based upon an approximate sensitivity of 90%, as measured by Mohd Nor et al. (2005). The optimum cut-point for stroke diagnosis was determined to be a total ROSIER score of one or above (Mohd Nor et al., 2005). Data were entered into SPSS version 14.0 for Windows and a descriptive analysis performed. Sensitivity, PPV and exact binomial 95% confidence intervals were calculated using Stata 10 for Windows. 3. Results Data were collected from the notes of 106 patients with suspected stroke or TIA. Of these, 28 patients had an alternative diagnosis and 78 a final diagnosis of stroke or TIA (Table 1). 3.1. Time from onset of symptoms to assessment Of the original 106 suspected stroke and TIA patients, 26 (24.5%) presented to hospital within 3 h of symptom onset. Of 78 patients with a final diagnosis of stroke or TIA, 63 (80.7%) had an ischaemic stroke. If the time for availing of rT-PA treatment had been set at 3 h from onset of symptoms, 19 of 63 (30.2%) patients would have been eligible for treatment. Extending the window to 4.5 h would increase this number to 23 (36.5%). Thus, even with a 4.5 h window, 40 patients (63.5%) would have been ineligible for rT-PA treatment.
Following medical consultant assessment and brain imaging of all 106 patients, 63 were diagnosed with ischaemic stroke, nine with TIA, six with intracerebral haemorrhage and 28 patients had an alternative diagnosis. All 106 suspected stroke or TIA patients were assessed by registered nurses using the ROSIER tool. Of the nine patients with a final diagnosis of TIA, six had no neurological signs or symptoms when assessed by registered nurses using the ROSIER and scored zero. The ROSIER has been shown to be insensitive for diagnosing asymptomatic TIA (Mohd Nor et al., 2005), therefore these six patients were excluded from the validation phase leaving a total of 100 study patients. Nine of the 106 suspected stroke cases had a final diagnosis of seizure. Todd’s paresis following seizures has been recognised as one of the most common conditions that produce stroke like signs and symptoms (Somes and Bergman, 2007). Registered nurses using the ROSIER correctly identified six out of nine of the patients with a final diagnosis of seizure compared to two out of nine identified by doctors using traditional neurological assessment. 3.3. Diagnostic performance As noted above, our analysis of diagnostic performance (which excludes the six patients with TIA who had no neurological signs or symptoms when assessed by registered nurses) is based on data from 100 patients. Diagnostic performance by referring doctors working in Accident and Emergency using standard neurological assessment was, sensitivity 97% (95% confidence intervals 87–100), PPV 84% (95% confidence intervals 71–94). It is
B. Byrne et al. / International Journal of Nursing Studies 48 (2011) 979–985
983
Table 2 Time to assessment and diagnostic performance of admitting medical and nursing staff on the stroke unit. Variable
Doctors
Nurses
Median/mean time from stroke unit admission to assessment in minutes (SD)
65/93 (SD 93.5)
Mean = 10/11 (SD 8.1)
Initial diagnosis
Identified with stroke
Final diagnosis: stroke Final diagnosis: no stroke Diagnostic specificity (95%CI) Negative predictive value (95% CI) Diagnostic sensitivity (95% CI) Positive predictive value (95% CI) a
68 17 39% 73% 94% 80%
(22%, (45%, (86%, (70%,
59%) 92%) 98%) 88%)
Not identified with stroke 4a 11
Identified with stroke 69 14 50% 82% 96% 83%
(31%, (57%, (88%, (73%,
Not identified with stroke 3 14
69%) 96%) 99%) 90%)
Includes one unrecorded case assumed to be a false negative.
not known what method of assessment referring primary care doctors utilized to diagnose stroke but their diagnostic performance was, sensitivity 94% (95% confidence intervals 80–99), PPV 82% (95% confidence intervals 66–92). For admitting doctors using traditional neurological assessment, diagnostic sensitivity was 94% (95% CI 86–98), and PPV 80% (95% CI, 70–88). At the optimum cut-point for stroke diagnosis using the ROSIER tool, i.e. a total score of one or above the corresponding diagnostic performance for registered nurses using the ROSIER was a sensitivity of 98% (95% CI, 88–99), and a PPV of 83% (95% CI, 73–90) (Table 2). Admitting doctors made incorrect assessments in 21 cases. Of these, 17 were false positive cases (assessed as stroke but had a final diagnosis of non-stroke) and four false negative cases (assessed as non-stroke diagnosis but had a final diagnosis of stroke). This included one unrecorded case which was assumed to be a false negative, as a CT image of the brain was not ordered until a consultant reviewed the patient later (it is standard practice for brain imaging to be ordered upon admission if a person is suspected to have had a stroke or TIA). Three of the false negatives were lacunar strokes, and one a partial anterior circulation stroke. Nurses made incorrect assessments in 17 cases. Of these, 14 were false positive cases and three false negative cases. Two of the false negatives were lacunar strokes, and one a partial anterior circulation stroke. Admitting doctors and nurses concurred on 13 of the incorrect assessments. 4. Discussion To our knowledge, this study is the first to examine the ability of registered nurses to use the ROSIER tool to identify patients with stroke. Our results show that the ROSIER tool had a sensitivity of 98% (95% confidence intervals 88–99%) and PPV of 83% (95% confidence intervals 73–90%) when used by registered nurses on a stroke unit. The sensitivity achieved by registered nurses was similar to that achieved by doctors using traditional neurological assessment in this study 94% (95% CI; 86–98), and similar to the sensitivity of 93% (95% CI: 89–97) achieved by doctors in an emergency department in the original validation study (Mohd Nor et al., 2005). However, there were insufficient numbers of non-stroke patients in this study to be able to accurately estimate the specificity and NPV.
A further limitation of this study is that the sample was of patients who already had a possible diagnosis of stroke. However, 28 (26%) of these patients eventually had an alternative diagnosis compared with an estimated norm of one in five (Harraf et al., 2002) and fourteen (50%) of these false positives were correctly identified by nurses using the ROSIER as conditions mimicking stroke. It could be argued that registered nurses employed on a stroke unit have specialised skills in relation to assessment of stroke signs and symptoms compared to registered nurses employed in other departments and so would be more adept at using the ROSIER tool. If this were the case the results from this study would not be widely generalisable to other departments. However, although registered nurses in this study were familiar with caring for patients with stroke, assessing for neurological signs commonly associated with stroke had not been part of their role and they required training in the use of the ROSIER tool. Items contained within the tool are easily taught and should be well within the capabilities of registered nurses employed in any department. Of the original 106 patients assessed using the ROSIER tool, nine (8.4%) had a final diagnosis of TIA. Six of the patients’ initial signs and symptoms had resolved at time of assessment by registered nurses and as a result the registered nurse diagnosed them as non-stroke. Doctors positively diagnosed six of the nine patients with TIA using standard neurological assessment and through obtaining the medical history. A typical TIA resolves within an hour (Albers et al., 2002), so correct diagnosis depends on acquiring an accurate medical history and appropriate neuroimaging (Lewandowski et al., 2008). Although the ROSIER extracts some history items, these focus on excluding conditions that mimic stroke rather than eliciting information which could guide TIA diagnosis. This has been acknowledged as a weakness of the tool by its developers, who recommend development of a separate tool to diagnose TIA (Mohd Nor et al., 2005). On the other hand, registered nurses using the ROSIER correctly identified six out of nine of the patients with a final diagnosis of seizure compared to two out of nine identified by doctors using traditional neurological assessment, perhaps indicating the strength of the ROSIER in identifying seizure as a stroke mimic. Using the ROSIER tool, registered nurses incorrectly diagnosed 17 cases (17%). Doctors’ incorrectly diagnosed
984
B. Byrne et al. / International Journal of Nursing Studies 48 (2011) 979–985
21 cases (21%). False positive rates for doctors and nurses, 17% and 14%, respectively, were slightly higher than the 10% found in the prospective validation phase described by Mohd Nor et al. (2005). It is possible that both doctors and nurses were influenced by the provisional diagnosis of stroke suggested by the referring doctor in primary care or in Accident and Emergency. Conversely these false positive rates were significantly lower than the 31% found in a study that examined hospital-based doctors’ ability to distinguish between stroke and stroke mimic using standard clinical assessment (Hand et al., 2006). Furthermore if the ROSIER score obtained by registered nurses had been used in combination with the doctors’ neurological assessment, the number of false positives could potentially have been reduced by five (four seizures and one syncope) bringing the false positive rate to 11% in line with Mohd Nor et al. (2005). Of the original 106 suspected stroke and TIA patients, 26 (24%) presented to hospital within 3 h of symptom onset. Included in this figure are the 63 patients suffering an ischaemic stroke, of whom 19 (30.2%) presented to hospital within 3 h of symptom onset. This finding is similar to other studies carried out in the UK, with proportions ranging from 25% to 37% (Johnson et al., 1999; Harraf et al., 2002) but lower than findings from studies in the United States (US) where proportions have ranged from 30% to 56% (Kothari et al., 1999; Evenson et al., 2001; Lacy et al., 2001). In the US rT-PA has been licensed for use since 1995 and there have been mass public education programmes emphasising the importance of early arrival to hospital following onset of stroke symptoms. The registered nurse assessment using the ROSIER took place on average 75 min before assessment by the admitting doctor, although the patients had already been assessed by doctors in primary care or in A&E. Time of assessment by doctor on the stroke unit was recorded for 55 (51%) of the original 106 patients. These findings may be partly explained by the timing of this audit which was prior to introduction of active administration of rT-PA for acute ischaemic stroke. As the service was not yet operational doctors may not have considered recording time of assessment as essential. We envisage an rT-PA service where the registered nurse, having obtained a positive score for stroke using the ROSIER, will immediately initiate a paging system which will warrant an immediate response from a member of an emergency response stroke team. This team member will assess the suspected stroke patient regarding their suitability for rTPA treatment. 4.1. Conclusion and implications On admission to hospital, whether to Accident and Emergency, a stroke unit or medical admissions unit, the first professional to assess a stroke patient is usually a registered nurse. This audit shows that registered nurses are able to quickly and effectively assess suspected stroke patients with an acceptable degree of accuracy. Given the importance of early intervention with rt-PA treatment, our study suggests that if registered nurses working in areas
that receive the acute stroke patient were trained to use the ROSIER as a basis for triaging patients, the time from admission to treatment could be minimized and the numbers of patients treated maximized. A larger study to establish the accuracy of assessment of patients with suspected stroke by registered nurses using the ROSIER in Accident and Emergency departments is recommended as this is likely to further reduce time from admission to treatment.Conflict of interest None declared. Funding None. Ethical approval The study proposal was submitted to the hospital’s Research Governance Body and classified as a service evaluation, so formal ethical approval was not sought. Permission to access patients’ notes was granted by consultant medical staff and the relevant senior managers in the hospital.
References Albers, G.W., Caplan, L.R., Easton, J.D., Fayad, P.B., Mohr, J.P., Saver, J.L., et al., 2002. Transient ischemic attack: proposal for a new definition. New England Journal of Medicine 347, 1713–1716. Carrozzella, J., Jauch, E.C., 2002. Emergency stroke management. Emergency Nursing 37, 35–57. Dawson, J., Walters, M., 2005. Development and validation of a stroke recognition tool. Lancet 4, 691–693. Donnan, G., Davis, S., 2008. Breaking the 3 h barrier for treatment. Lancet Neurology 7, 981–982. Donnan, G., Fisher, M., MacLeod, M., Davis, S., 2008. Stroke. The Lancet 371, 1612–1623. Evenson, K., Rosamond, W., Morris, D., 2001. Prehospital and in-hospital delays in acute stroke care. Neuroepidemiology 20, 65–76. Feigin, V.L., Lawes, C.M.M., Bennett, D.A., Barker-Collo, S.L., Parag, V., 2009. Worldwide stroke incidence and early case fatality reported in 56 population-based studies: a systematic review. Lancet Neurology 8, 355–369. Hacke, W., Donnan, G., Fieschi, C., Kaste, M., von Kummer, R., Broderick, J.P., et al., 2008. Thrombolysis with Alteplase 3–4.5 h after acute ischemic stroke. New England Journal of Medicine 359, 1317–1329. Hand, P.J., Kwan, J., Lindley, R.I., Dennis, M.S., Wardlaw, J.M., 2006. Distinguishing between stroke and mimic at the bedside. The Brain Attack Study. Stroke 37 (3), 769–775. Harraf, F., Sharma, A.K., Brown, M.M., Lees, K.R., Vass, R.I., Kalra, L., 2002. A multicentre observational study of presentation and early assessment of acute stroke. British Medical Journal 325, 17–20. Harbinson, J., Hossain, O., Jenkinson, D., Davis, J., Louw, S.J., Ford, G.A., 2003. Diagnostic accuracy of stroke referrals from primary care. Emergency room physicians, and ambulance staff using the face arm speech test. Stroke 34, 71–76. Hinkle, J.L., Manoj, A., Brook, S., Webb, A., Muangpaisan, W., Kennedy, J., Buchan, A.M., 2007. Nursing triage of patients with acute stroke. Nursing Times 103, 32–33. Johnson, F., Wardlaw, J., Dennis, M.S., Lewis, S., Nimmo, G., Lindley, R.I., Warlow, C.P., 1999. Delays in stroke referrals. Lancet 354, 47–48. Kothari, R., Jaunch, E., Broderick, J., Brott, T., Sauerbeck, L., Khoury, J., Liu, T., 1999. Acute stroke: delays to presentation and emergency department evaluation. Annals of Emergency Medicine 33, 3–8. Lacy, C.R., Suh, C.C., Bueno, M., Kostis, J.B., 2001. Delay in presentation and evaluation for acute stroke: stroke time registry for outcomes knowledge and epidemiology (S.T.R.O.K.E). Stroke 32, 63–69. Lewandowski, C.A., Rao, C.P.V., Silver, B., 2008. Transient ischemic attack: definitions and clinical presentations. Annals of Emergency Medicine 52, S7–S16. Mohd Nor, A.M., Davis, J., Sen, B., Shipsey, D., Louw, S.J., Dyker, A.G., et al., 2005. The recognition of stroke in the emergency room (ROSIER) scale, development and validation of a stroke recognition instrument. Lancet 4, 727–734.
B. Byrne et al. / International Journal of Nursing Studies 48 (2011) 979–985 National Audit Office, 2005, 2005. Reducing Brain Damage: Faster Access to Better Stroke Care. Department Of Health, London. NICE, 2007. Alteplase for the Treatment of Acute Ischaemic Stroke. National Institute for Health and Clinical Excellence, London Available from: http://www.nice.org.uk/nicemedia/pdf/TA122guidance. pdf. NICE, 2008. Stroke: Diagnosis and Initial Management of Acute Stroke And Transient Ischaemic Attack. National Institiute Health and Clinical Excellence, London Available from: http://www.nice.org.uk/ nicemedia/pdf/CG68NICEGuideline.pdf.
985
SIGN (Scottish Intercollegiate Guidelines Network. & NHS Quality Improvement Scotland), 2008. Management of Patients with Stroke or TIA: Assessment, Investigation, Immediate Management and Secondary Prevention: A National Clinical Guideline. Scottish Intercollegiate Guidelines Network, Edinburgh. Somes, J., Bergman, D.L., 2007. ABCDs of acute stroke intervention. Journal of Emergency Nursing 33 (3), 228–234. Strong, K., Mathers, C., Bonita, R., 2007. Preventing stroke: saving lives around the world. Lancet Neurology 6, 182–187.