ROSE rule (risk stratification of syncope in the emergency department)

Clinical background

Syncope is a common condition in the emergency department and accounts for 1 to 6 per cent of all hospital admissions. The decision between admission and discharge is difficult because the clinical picture is often normal by the time the patient arrives, and the underlying cause may be anything from benign vasovagal syncope to a life-threatening arrhythmia or myocardial infarction. Existing instruments such as the San Francisco Syncope Rule and the OESIL score showed shortcomings in the pilot study: the San Francisco rule had good sensitivity but at the cost of a high admission rate, while the OESIL score was not sensitive enough to identify reliably the patients who could be sent home [2]. The ROSE rule was developed to provide an instrument with high sensitivity and a high negative predictive value, so that low-risk patients can be identified reliably and unnecessary admissions reduced.

Applying the ROSE rule

The ROSE rule is a binary risk rule with seven criteria, in which any single criterion that is met classifies the patient as high risk. The acronym BRACES summarises the criteria:

$$\text{High risk} = \begin{cases} 1 & \text{if any of the following is present} \ 0 & \text{if no criterion is present} \end{cases}$$

The seven criteria are:

The rule was derived in a prospective observational study at the Royal Infirmary of Edinburgh, United Kingdom [1]. The study included adult patients presenting to the emergency department with syncope. A derivation cohort of 550 patients was used to identify independent predictors of a serious outcome or death from any cause within one month. The outcome occurred in 40 patients (7.3 per cent) in the derivation cohort. The independent predictors with the strongest associations were positive faecal occult blood on rectal examination (OR 13.2), a BNP $\geq 300$ pg/mL (OR 7.3), a haemoglobin $\leq 90$ g/L (OR 6.7), an oxygen saturation $\leq 94$ per cent (OR 3.0) and a Q wave on the ECG (OR 2.8). The rule was then tested in an internal validation cohort of a further 550 patients, in which the outcome occurred in 39 patients (7.1 per cent).

Interpretation in practice

The ROSE rule has only two outcomes, which makes interpretation straightforward:

The rule is constructed to maximise sensitivity and the negative predictive value. Its purpose is therefore primarily to identify reliably the patients who do not need admission, rather than to quantify precisely the risk in those classified as high risk. BNP was the single most important predictor: in the validation cohort, a raised BNP accounted for 36 per cent of all serious cardiovascular events and 89 per cent of the deaths [1]. A raised BNP should therefore prompt particular vigilance even when no other criterion is met.

Validation and performance

In the internal validation cohort, the ROSE rule showed a sensitivity of 87.2 per cent, a specificity of 65.5 per cent and a negative predictive value of 98.5 per cent [1]. The high negative predictive value is the rule's strength: few patients classified as low risk suffer a serious outcome.

A prospective cohort study from 2016 with 159 syncope patients at two American hospitals supported the central role of BNP in the rule [3]. Patients with a BNP above 250 pg/mL had an 8.8-fold increased risk of a serious outcome within one month (OR 8.8, 95 per cent CI 3.3 to 23.8), a 14.8-fold increased risk of myocardial infarction and a 4.5-fold increased risk of a life-threatening arrhythmia. The authors concluded that BNP measurement combined with the ROSE rule can improve risk stratification compared with clinical judgement alone.

In a geriatric population, however, the rule performed considerably less well. A retrospective study from 2023 comprising 350 patients aged 65 years or older found that the ROSE rule had a sensitivity of only 19 per cent for outcomes within 48 hours and 55 per cent for outcomes within 30 days [4]. This was the lowest of all four rules tested (CSRS, EGSYS, SFSR and ROSE). The Canadian Syncope Risk Score performed best, with an AUC of 0.73 at 48 hours and 0.75 at 30 days. All the rules performed suboptimally in the older population, suggesting that the ROSE rule cannot be transferred uncritically to geriatric patients.

Large external validation studies in mixed adult populations are lacking in the literature. The derivation study was single-centre, and the few published studies evaluating the rule have been small or retrospective.

Limitations

The ROSE rule has several important limitations:

It does not apply to geriatric patients. In the population aged 65 years and over, sensitivity falls dramatically and the rule misses a large proportion of patients with a serious outcome [4]. In older patients, other instruments or a more conservative admission strategy should be considered.

Single-centre derivation. The rule was developed at a single British hospital, and external validation in other populations and health systems is lacking in larger prospective studies.

BNP requires a laboratory assay. BNP is the single most important predictor in the rule, but the assay is not always available around the clock in every emergency department. Without BNP the rule loses a substantial part of its predictive power.

Rectal examination. The criterion of positive faecal occult blood presupposes that a rectal examination is performed, which is not routine in every syncope work-up. If the examination is omitted, the criterion cannot be assessed, which in practice lowers the sensitivity of the rule.

The rule does not quantify risk. Unlike score-based instruments, the ROSE rule gives only a binary result. Patients with several positive criteria cannot be expected to have a higher risk than those with a single criterion under the rule, even though such a trend is clinically plausible.

Bradycardia and chest pain. These two criteria are included in the rule but were not independent predictors on multivariable analysis in the derivation cohort [1]. They were probably included on grounds of clinical relevance. Their contribution to the performance of the rule is less well documented.

References

1. Reed MJ et al. The ROSE (risk stratification of syncope in the emergency department) study. J Am Coll Cardiol 2010. PMID: 20170806
2. Reed MJ et al. The Risk stratification Of Syncope in the Emergency department (ROSE) pilot study: a comparison of existing syncope guidelines. Emerg Med J 2007. PMID: 17384381
3. Isbitan A et al. Utility of brain natriuretic peptide assay as a predictor of short term outcomes in patients presenting with syncope to the emergency department. Cardiovasc Diagn Ther 2016. PMID: 27280086
4. Kiradoh SA et al. Predicting short-term adverse outcomes in the geriatric population presenting with syncope: a comparison of existing syncope rules and beyond. J Geriatr Cardiol 2023. PMID: 36875169