Article Text

other Versions

Quantitative study - other
Nurses trained in the use of the ROSIER tool can assess signs and symptoms of stroke with comparable accuracy to doctors performing standard neurological assessment.
  1. Holli A DeVon1,
  2. Patricia Zrelak2
  1. 1Betty Irene Moore School of Nursing, University of California Davis, Sacramento, California, USA
  2. 2Stroke Program, University of California Davis Medical Center, Sacramento, California, USA
  1. Correspondence to Holli A DeVon
    Betty Irene Moore School of Nursing, University of California Davis, 4610 X St., Suite 4202, Sacramento, CA 95817, USA; Holli.DeVon{at}

Statistics from

Commentary on: OpenUrlCrossRefPubMed

Implications for practice and research

  • Rapid assessment of patients with potential stroke is imperative for triage nurses.

  • The Recognition of Stroke in the Emergency Room (ROSIER) scale may help triage nurses rapidly recognise stroke and differentiate between potential stroke and common stroke mimics.

  • The following guidelines are recommended for patients with ischaemic stroke1:

  • Administration of recombinant tissue plasminogen activator (rt-PA) within 3 h of ischaemic stroke and up to 4.5 h in a select subgroup of patients.

  • Evaluation of stroke symptoms by a physician within 10 min of arrival.

  • Door-to-initiation of rt-PA in 60 min or less for stroke.


A major barrier to worldwide use of thrombolytic therapy is a lack of stroke recognition.2 Accurate triage of patients with potential stroke is critical to facilitate the administration of rt-PA. Byrne et al compared the ability of stroke nurses to diagnose stroke using the ROSIER tool with that of the traditional neurological assessment performed by physicians.


This study took place in the stroke unit where patients were admitted after an initial evaluation in the accident and injury (A and E) department or by direct referral from their primary care physician. Patients triaged to the stroke unit for admission had an 80% chance of having a stroke. The authors determined that a sample size of 100 would provide a 95% probability of estimating the true sensitivity of the ROSIER within 7 percentage points, although it is not clear how this was established. Nurses were asked to determine, based on the ROSIER score, whether or not the patient had a stroke. The physician on duty then independently evaluated the patient based on a standard neurological examination. Finally, patients were evaluated by a stroke doctor for final diagnosis using additional information such as brain imaging. Sensitivity, specificity, positive and negative predictive values were computed for each group.


Screening with the ROSIER tool had similar results when compared with the full neurological assessment performed by the physician. Sensitivity and positive predictive value (PPV) for nurses were 98% (95% CI 88 to 99%) and 83% (95% CI 73 to 90%), respectively. Results were similar for physicians with a sensitivity of 94% (95% CI 86 to 99%) and PPV of 80% (95% CI 70 to 88%). However, caution must be taken in interpreting these results beyond test comparability. Evaluations did not account for the additional information from the first examination (which would have been positive or the patients would not have been triaged to the stroke unit). In addition, patients with transient ischaemic attack, who were asymptomatic when the ROSIER was administered, need to be recognised.


Stroke hospitals should treat 80% of ischaemic strokes with thrombolytic therapy within 60 min of the patient's arrival. The mean time from nurse to physician assessment in the stroke unit was 75 min (SD=65.8 min). This adds to the delay in administration of rt-PA for eligible patients. Additionally, the patient's condition may have changed during this time, necessitating re-evaluation. For the 26 (24.5%) patients who presented to the hospital within 3 h of symptom onset, time to physician evaluation was quite long and would need to be improved for safe and effective use of rt-PA.

The comparison of results by referring physicians working in A and E and primary care is confusing. Seeing all types of patients versus those already screened with a tool such as the ROSIER will make a difference in the test results, particularly predictive values. Additionally, a κ statistic or other measure of inter-rater reliability would be helpful in determining the amount of agreement between the nurse's ROSIER score and physician ratings. The diagnostic specificity was low at only 50%, meaning many strokes could be missed, leading to under treatment. The ROSIER tool can be used effectively by nurses to assess symptoms of stroke and may be best used in the emergency department as intended by the developers.


View Abstract


  • Competing interests None.

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.