Boston syncope criteria

Clinical background

Syncope is a common presentation in the emergency department, and in approximately 10 per cent of patients there is a serious or potentially life-threatening underlying cause that is not always evident at the initial assessment [4]. The decision to admit or discharge a patient with syncope is therefore difficult: too restrictive an approach leads to unnecessary admissions, and too permissive an approach misses patients with occult cardiac or neurological disease. The Boston syncope criteria were developed precisely to structure this decision. The instrument is designed as a screening tool with high sensitivity, that is, its primary purpose is to identify patients who can be discharged at low risk of suffering a serious event within 30 days, rather than to classify precisely those who are unwell.

Applying the Boston syncope criteria

The Boston syncope criteria form a binary decision rule with eight criteria. Each criterion is a yes or no, and if at least one criterion is met the patient is classified as high risk. There is no graded scoring system; the rule works as a threshold:

$$\text{High risk} = \begin{cases} 1 & \text{if any of the 8 criteria is met} \ 0 & \text{if no criterion is met} \end{cases}$$

The eight criteria are:

1. Signs/symptoms of acute coronary syndrome
2. Signs of a conduction disturbance
3. A worrying cardiac history (structural heart disease, a low ejection fraction)
4. Valvular disease by history or examination
5. A family history of sudden death
6. Persistently abnormal vital signs in the emergency department
7. Signs of volume depletion
8. A primary central nervous system event

The rule was derived in a prospective observational study at Beth Israel Deaconess Medical Center in Boston, published in 2007 [1]. The study included 362 consecutive patients aged 18 years or older presenting with syncope, of whom 293 (81 per cent) completed 30 days of follow-up. Of these, 201 (69 per cent) were admitted, and 68 patients (23 per cent) had either a critical intervention or a serious event within 30 days. The primary outcome measure was composite: a critical intervention (for example a pacemaker, a defibrillator, urgent coronary intervention, anticoagulation) or a serious event (death, myocardial infarction, arrhythmia, stroke, pulmonary embolism, haemorrhage) within 30 days.

Interpretation in practice

Since the rule is binary there are only two outcomes, and interpretation concerns disposition rather than a risk percentage:

The rule is not intended to replace clinical judgement. A patient classified as low risk in whom the treating physician has a strong clinical suspicion should nonetheless be managed according to the clinical picture. Conversely, the rule is most useful when the physician is considering discharging a patient: if no criterion is met, the rule supports that decision.

Validation and performance

In the derivation cohort the rule achieved a sensitivity of 97 per cent (95 per cent CI 93 to 100) and a specificity of 62 per cent (95 per cent CI 56 to 69) [1]. Of the 68 patients with an outcome, the rule identified 66. The negative predictive value was high, which is what the instrument is designed for.

A prospective implementation study at the same emergency department tested the rule in routine clinical use [2]. In a before-and-after design, the proportion of patients admitted fell from 69 per cent (201 of 293) to 58 per cent (160 of 277), an absolute reduction of 11 percentage points. In real-time use, sensitivity was 100 per cent (95 per cent CI 94 to 100), specificity 57 per cent (95 per cent CI 50 to 63), and the negative predictive value 100 per cent. No serious events occurred among the patients discharged. The study was, however, single-centre and conducted at the same institution as the derivation, which limits generalisability.

An external validation was carried out retrospectively at a tertiary hospital in Jerusalem [3]. Of 198 patients with complete follow-up, 21 had a serious event or a critical intervention. The rule identified 20 of the 21, giving a sensitivity of 95 per cent, a specificity of 66 per cent and a negative predictive value of 99 per cent. This is the only published external validation in an independent population outside Boston, and the results confirm that the rule retains high sensitivity but low specificity in another care setting as well.

A Chinese retrospective study from 2024 compared five risk stratification tools in 221 patients aged 60 years or older with syncope [5]. The Boston criteria performed with a sensitivity of 94.9 per cent, a specificity of 67.3 per cent and a negative predictive value of 97.3 per cent. This was comparable to the FAINT score (sensitivity 93.2 per cent) and better than ROSE, SFSR and the CSRS in the same cohort. The study was, however, retrospective and limited to older patients, so the results cannot be generalised to younger populations.

A systematic review and meta-analysis from 2025, carried out for SAEM GRACE, identified 13 clinical decision rules for syncope [4]. Only three of these (the San Francisco Syncope Rule, the Canadian Syncope Risk Score and OESIL) had been validated in more than two studies. The Boston criteria were not among them, and the review concludes that the evidence for most decision rules is of low quality for routine clinical use. No randomised controlled trial has compared structured risk stratification with unstructured clinical judgement in syncope.

Limitations

The Boston criteria are designed to be highly sensitive, which means that specificity is low. A substantial proportion of patients classified as high risk do not have a serious event, and the rule can therefore lead to unnecessary admissions if applied mechanically. In the derivation cohort, specificity was 62 per cent, and in the implementation study it fell to 57 per cent [1, 2].

The rule does not apply to patients in whom the cause of the syncope is already established on arrival, for example with a documented tachycardia or bradycardia that explains the episode. In these cases management should be guided by the underlying diagnosis, not by the screening rule. The 2018 ESC guidelines emphasise precisely this distinction between syncope with an established diagnosis and syncope of uncertain cause, in which prognostic stratification is relevant [6].

The rule has not been validated in children or young adults under 18 years. The Chinese study was restricted to patients aged 60 years or older, and the derivation cohort had a mixed age distribution but did not include paediatric patients [1, 5]. Nor has any of the validation studies tested the performance of the rule in patients with known severe structural heart disease, in whom the risk is so high that a screening tool is superfluous.

The individual criteria are partly subjective. "A worrying cardiac history" and "signs of volume depletion" require clinical interpretation, which can reduce reproducibility between assessors. In the derivation study, the assessment was made by emergency physicians in real time, and inter-observer variability was not reported systematically [1].

References

1. Grossman SA, Fischer C, Lipsitz LA, et al. Predicting adverse outcomes in syncope. J Emerg Med. 2007;33(3):233-9. PMID: 17976548
2. Grossman SA, Bar J, Fischer C, et al. Reducing admissions utilizing the Boston Syncope Criteria. J Emerg Med. 2012;42(3):345-52. PMID: 21421292
3. Muhtaseb O, Alpert EA, Grossman SA. A Tale of Two Cities: Applying the Boston Syncope Criteria to Jerusalem. Isr Med Assoc J. 2021;23(7):420-425. PMID: 34251124
4. Wakai A, Sinert R, Zehtabchi S, et al. Risk-stratification tools for emergency department patients with syncope: A systematic review and meta-analysis of direct evidence for SAEM GRACE. Acad Emerg Med. 2025;32(1):72-86. PMID: 39496561
5. Mu H, Liu J, Huang C, et al. Application of five risk stratification tools for syncope in older adults. J Int Med Res. 2024;52(1). PMID: 38190847
6. Numeroso F, Mossini G, Lippi G, et al. Emergency department management of patients with syncope according to the 2018 ESC guidelines: Main innovations and aspect deserving a further improvement. Int J Cardiol. 2019;283:119-121. PMID: 30826198