Response to Letter Regarding Article, “Does Use of the Recognition of Stroke In the Emergency Room Stroke Assessment Tool Enhance Stroke Recognition by Ambulance Clinicians?”
In our study,3 ambulance clinicians had to suspect a stroke before they used the Recognition of Stroke in the Emergency Room (ROSIER). Our clinicians use a range of factors to suspect that a patient is having a stroke including clinical assessments and observations taken on scene, medical history, history of the current event, medications, and the call determinant allocated during triage of the emergency call for help. Our clinicians practice in line with the Joint Royal Colleges Ambulance Liaison Committee’s UK Ambulance Service Clinical Practice Guidelines.4
The average age of patients in our study was indeed relatively young at 65 years; however, this is representative of suspected stroke patients in the area of London where our study was conducted. Our stroke audit figures from last year (2012/13) show that the average age of suspected stroke patients conveyed by our clinicians to the hospital involved in our study was 67 years, which is consistent with the age reported in our article. Information from the 2011 National Census5 confirms that the average age in the London borough where the participating hospital is located is younger than the national average age (30.9 versus 39.4 years). In addition, the higher proportion of black and South Asian individuals (43%) residing in the area served by the participating hospital may have contributed to the relatively young age of patients with stroke in our study attributable to the propensity of stroke at a younger age in these ethnic groups.6,7
In our article, we report that “32 patients were treated by study-trained ambulance clinicians, but as ROSIER assessment was not undertaken, they were not included in the study.” Brandler and Sharma2 ask what became of those patients and furthermore whether we searched our registry to identify confirmed stroke patients attended by ambulance clinicians in whom the ROSIER was not used. Because the 32 patients were not included, hospital outcomes were not collected and thus we do not know what happened to them. Unfortunately, hospital data are not routinely shared with UK ambulance services and as such we do not have access to diagnoses for patients who were outside the remit of our study. It is for this reason that we are also unable to identify retrospectively from our registry the patients attended by ambulance clinicians where the ROSIER was not used, but who were later diagnosed as confirmed stroke.
In response to Brandler and Sharma’s2 interest in the level of agreement between the Face Arm Speech Test (FAST) and ROSIER regarding the presence of stroke, we have interrogated our data further: 100% of the ROSIER-positive patients were also FAST positive and 98% of the FAST-positive patients were also ROSIER positive. All 6 patients (2%) who were FAST positive but ROSIER negative were confirmed as nonstroke. In these instances, the ROSIER correctly identified nonstroke when the FAST did not.
Price et al1 have identified an important potential issue regarding redirection bias, although we are confident that this was not an issue in our study. As per the study protocol, the vast majority of patients for whom the ROSIER was applied were taken to either the hyper-acute stroke unit or the emergency department at the participating hospital. A total of 54 patients who were assessed with the ROSIER were conveyed to an alternative hospital: 22 were FAST negative and 32 were FAST positive. Because they went to a nonparticipating hospital, we were unable to obtain outcome information and they were excluded from the study.
However, if we were to assume the absolute worst case scenario, that is, for all 54 cases our clinicians made the wrong diagnosis (ie, we had an additional 22 false negatives and 32 false positives), the effect on sensitivity and specificity would be as follows. Sensitivity and specificity levels for the FAST would change from 97% to 86% and from 13% to 10%, respectively. Sensitivity and specificity levels for the ROSIER would change from 97% to 86% and from 18% to 14%, respectively. This effect is, of course, greatly exaggerated because it is extremely unlikely that our ambulance clinicians would have made inaccurate diagnoses in all 54 cases. So, although the sensitivity levels might decrease from our reported level of 97%, we could reasonably expect that in reality they would not decrease as far as 86%. Thus, we can be confident that our sensitivity levels are indeed genuinely higher than those observed previously in North East England. Furthermore, any effect of failing to identify false-negative cases is likely to be small and would not alter the conclusions of our study.
We agree with Price et al1 that is it is important to understand whether clinical details and symptoms were present at the point of admission, and this is why we comment in our discussion that “the ROSIER may have failed to out-perform the FAST because of certain signs of stroke not being present during prehospital assessment or not being recorded by ambulance clinicians.” We further agree that positive predictive value is a more meaningful interpretation for clinical practice and service cost-effectiveness than sensitivity. We hope that those reading our article will consider positive predictive value as a practical measure of the effectiveness of the stroke recognition tools.
Rachael T. Fothergill, PhD
Melanie J. Edwards, PhD
Clinical Audit and Research Unit
London Ambulance Service NHS Trust
London, United Kingdom
Patrick Gompertz, MBChB, FRCP
Barts Health NHS Trust
Royal London Hospital
London, United Kingdom
Stroke welcomes Letters to the Editor and will publish them, if suitable, as space permits. Letters must reference a Stroke published-ahead-of-print article or an article printed within the past 3 weeks. The maximum length is 750 words including no more than 5 references and 3 authors. Please submit letters typed double-spaced. Letters may be shortened or edited.
- © 2014 American Heart Association, Inc.
- Price CI,
- Rudd M,
- Ford GA
- Brandler ES,
- Sharma M
- Fothergill RT,
- Williams J,
- Edwards MJ,
- Russell IT,
- Gompertz P
- 4.↵Joint Royal Colleges Ambulance Liaison Committee. UK Ambulance Service Clinical Practice Guidelines. Warwick, United Kingdom: Joint Royal Colleges Ambulance Liaison Committee;2006. http://www2.warwick.ac.uk/fac/med/research/hsri/emergencycare/prehospitalcare/jrcalcstakeholderwebsite/guidelines/clinical_guidelines_2006.pdf. Accessed December 11, 2013.
- 5.↵Office for National Statistics. 2011 Census Data for England and Wales. http://www.nomisweb.co.uk/census/2011/ks102ew. Accessed October 31, 2013.
- Stewart JA,
- Dundas R,
- Howard RS,
- Rudd AG,
- Wolfe CD
- Banerjee S,
- Biram R,
- Chataway J,
- Ames D