The SOAR Stroke Score Predicts Inpatient and 7-Day Mortality in Acute Stroke
Background and Purpose—An accurate prognosis is useful for patients, family, and service providers after acute stroke.
Methods—We validated the Stroke subtype, Oxfordshire Community Stroke Project Classification, Age, and prestroke Rankin stroke score in predicting inpatient and 7-day mortality using data from 8 National Health Service hospital trusts in the Anglia Stroke and Heart Clinical Network between September 2008 and April 2011.
Results—A total of 3547 stroke patients (ischemic, 92%) were included. An incremental increase of inpatient and 7-day mortality was observed with increase in Stroke subtype, Oxfordshire Community Stroke Project Classification, Age, and prestroke Rankin stroke score. Using a cut-off of ≥3, the area under the receiver operator curves values for inpatient and 7-day mortality were 0.80 and 0.82, respectively.
Conclusions—A simple score based on 4 easily obtainable variables at the point of care may potentially help predict early stroke mortality.
The importance of accurate and early prediction of survival in patients with acute stroke is undisputed. For widespread implementation of a prognosis score, it should be accurate, pragmatic, and easy to calculate, such as the Age, Blood Pressure, Clinical Features, Duration, Diabetes score (ABCD2) for transient ischemic attack.1 Previously developed prognostic scores are very complex.2,3 There has yet to be a single, globally accepted stroke mortality prediction score.
We have previously derived and internally validated the Stroke subtype, Oxfordshire Community Stroke Project Classification, Age, and prestroke Rankin (SOAR) score and have shown that it can accurately predict both inpatient and early mortality after acute stroke.4 In this study, we validated the SOAR score using an independent data set routinely collected in 8 National Health Service hospital trusts in the Anglia Stroke & Heart Clinical Network.5
The study sample was drawn from the Anglia Stroke and Heart Clinical Network database which routinely collects data on consecutive stroke admissions since September 2008 in 8 National Health Service hospitals in Norfolk, Suffolk, and Cambridgeshire. The individual institutional approvals were obtained. Patients included were confirmed cases with stroke based on clinical history and neuroimaging. Patients admitted to 1 site before February 2010 were excluded because they were included in the derivation study.
The SOAR is an 8-point scale (0–7), which includes Stroke subtype (ischemic/hemorrhagic), Oxford Community Stroke Project classification, Age, and prestroke modified Rankin Score (Table I in the online-only Data Supplement). The final Oxford Community Stroke Project diagnosis was based on initial clinical assessment and neuroimaging. The score is the sum of the points designated for these 4 variables, which are measured at the time of admission.
The sensitivities, specificities, positive and negative predictive values were calculated for each point of the score for both inpatient and 7-day mortality.
Three thousand five hundred forty seven patients with stroke admitted between September 2008 and April 2011 with complete data for SOAR were included (41% of the sample). Their characteristics are shown in Table II in the online-only Data Supplement. Out of the 23% of those excluded where full data sets were not available, there were a significantly higher rate of hemorrhagic stroke (17.3% versus 7.8%) and fewer lacunar strokes (14.6% versus 25.9%). Table 1 compares the crude rates for inpatient and early (within 7-day) mortality for each score point between the validation and derivation studies (Figure I in the online-only Data Supplement). No statistically significant differences were observed in mortality rates between the studies for each score value for both outcomes.
The sensitivities, specificities, positive and negative predictive values in predicting study mortality outcomes are shown in Table 2 for each cut-off value of SOAR score. The best balance of predictive value was observed for the cut-off score of 3 (ie, ≥3). The area under the receiver operator curves for inpatient mortality and 7-day mortality were 0.80 (95% confidence interval, 0.78–0.82) and 0.82 (95% confidence interval, 0.79–0.84), respectively.
This external validation study suggests that the SOAR stroke score may help in predicting inpatient and 7-day mortality after acute stroke. In contrast to previously proposed stroke prognostic scores, SOAR contains only 4 easily obtainable variables. Furthermore, unlike other scores which cannot be used for both ischemic and hemorrhagic stroke, SOAR can be applied to both (Tables III and IV in the online-only Data Supplement).
Approximately 76% and 64% of patients scoring 6 points died as inpatient in the derivation and validation studies, respectively (Table 1). Thus, should a patient deteriorate rapidly, consideration of the initial SOAR score might support future management decisions. Further studies should compare the value of the SOAR score with clinical judgment and test the usefulness of well-defined treatment pathways for severe stroke.
The score can be calculated by both clinical and nonclinical staff (for administrative purposes) once the clinical assessment has been made, all 4 factors being fixed at time of evaluation. In addition, we used the data from 8 hospitals in the United Kingdom which would capture variations in stroke services6 making the results more likely to be generalizable.
Our study has limitations. First, the score was only validated using hospital-based data. Second, differences were observed in hemorrhage rates, age, and prestroke Rankin between included and excluded patients. However, the internal relationship between the score and outcomes examined would not have been affected. A further potential limitation is the possibility of inter-rater variability and thus score reliability (eg, modified Rankin Scale).7 However, the prestroke modified Rankin scale was categorized into 0 to 2, 3 to 4, and 5, thus minimizing the impact of inter-rater variability. Although the sensitivity and specificity of the SOAR score are reasonable when cut-off point ≥3 is used, the positive predictive value was less impressive (only 23% with a SOAR score of 3 died) and this may result in some patients being assigned a misleading prognosis. As a result, clinical judgment still has a role in prognostication as the score has yet to be tested against clinical judgment.
In summary, we found that the SOAR stroke score could predict inpatient and 7-day mortality in acute stroke; however, care must be taken in interpreting these results in view of the relatively low positive predictive values of the score. It is, however, simple to administer and can be implemented at time of admission. Future studies should evaluate if the score has value in predicting stroke mortality in different populations along with the ability to predict physical or cognitive function, discharge destination, and long-term mortality.
We thank our funders and stroke teams based at the participating NHS Trusts.
Sources of Funding
The Anglia Stroke Clinical Network Evaluation Study (ASCNES) is funded by the National Institute of Health Research (NIHR) for Patient Benefit Program (PB-PG-1208-18240). This paper presents independent research commissioned by the NIHR under its Research for Patient Benefit (RfPB) Programme (Grant Reference Number PB-PG-1208-18240). The views expressed are those of the author(s) and not necessarily those of the NHS, the NIHR or the Department of Health. The Anglia Stroke and Heart Clinical Network is funded by the National Health Service Improvement Program. Funding was also received from the Cambridge NIHR Biomedical Research Grant.
John F. Potter received funding from British Heart Foundation, National Institute of Health Research and Dunhill Medical Trust. Anthony K. Metcalf received funding from National Institute of Health Research and from Bohringer Ingelheim for talks to general practitioners (<$10k). Dr Shekhar received speaker fees from Boehringer and Bayer. He sits in the advisory board for Astellas Incontinence Services. Dr Warburton received funding from British Heart Foundation and she has received honoraria for various talks and has been involved in medical reports for medicolegal claims/cases. Dr Myint received funding from National Institute of Health Research and Dunhill Medical Trust. The other authors have no conflict to report.
The Norfolk Research Ethics Committee approved the study.
The online-only Data Supplement is available with this article at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.113.001148/-/DC1.
- Received February 11, 2013.
- Accepted March 28, 2013.
- © 2013 American Heart Association, Inc.
- Smith EE,
- Shobha N,
- Dai D,
- Olson DM,
- Reeves MJ,
- Saver JL,
- et al
- Saposnik G,
- Kapral MK,
- Liu Y,
- Hall R,
- O’Donnell M,
- Raptis S,
- et al
- Myint PK,
- Clark AB,
- Kwok CS,
- Davis J,
- Durairaj R,
- Dixit AK,
- et al
- Rodgers H,
- Dennis M,
- Cohen D,
- Rudd A
- Quinn TJ,
- McArthur K,
- Dawson J,
- Walters MR,
- Lees KR