Clinical background
Syncope is a common presentation in the emergency department, and the clinical challenge is to separate patients with a cardiac cause, who may require urgent investigation and admission, from those with a reflex-mediated or other benign cause. The history and examination often give clues, but individual findings have only moderate discriminatory power. The EGSYS score was developed precisely to structure the initial assessment and to gather the clinical variables that best distinguish cardiac syncope from other causes, before advanced investigation has been undertaken.
The score is a diagnostic, not a prognostic, instrument. Its purpose is to estimate the probability that the cause of the syncope is cardiac, not to predict death or serious events over a short horizon. This distinguishes EGSYS from other syncope rules, such as the San Francisco Syncope Rule, which are primarily intended to predict serious outcomes.
Calculating the EGSYS score
EGSYS (Evaluation of Guidelines in Syncope Study) is calculated as the sum of six clinical variables, four contributing positively and two negatively:
Each variable is binary (yes = 1, no = 0). The score ranges from to .
The variables scoring negatively — autonomic prodromes (nausea/vomiting) and precipitating/predisposing factors (a warm environment, prolonged standing, fear/pain/emotional distress) — are typical of reflex-mediated syncope. Their absence strengthens the suspicion of a cardiac cause, and they are therefore subtracted from the score.
The derivation study was conducted as a prospective cohort study at 14 general hospitals in Italy [1]. In total 516 consecutive patients with unexplained syncope were referred urgently to the emergency departments. The cohort was divided into a derivation cohort of 260 patients, in which the predictive variables were identified and the scoring developed, and a validation cohort of 256 patients, in which the model was tested. Diagnosis followed the ESC guidelines on syncope, and the outcome was a diagnosis of cardiac syncope established after the work-up was complete. Over a mean follow-up of 614 days, mortality was also recorded.
In the derivation cohort, a score of 3 or higher identified cardiac syncope with high sensitivity (95 per cent in the derivation cohort, 92 per cent in the validation cohort) but moderate specificity (61 and 69 per cent respectively) [1]. Patients with a score of 3 or higher also had considerably higher mortality during follow-up: 17 per cent versus 3 per cent in the derivation cohort and 21 per cent versus 2 per cent in the validation cohort, both with p < 0.001 [1].
Interpretation in practice
A threshold of 3 points is the established cut-off used both in the derivation study and in subsequent validations. The score is interpreted as an estimate of probability, not as a definite diagnosis.
| Score | Interpretation | Clinical action |
|---|---|---|
| < 3 | Low probability of cardiac syncope | Consider another cause. If the clinical picture and the ECG are consistent with reflex-mediated syncope and there are no red flags, the patient can usually be managed as an outpatient. |
| ≥ 3 | Increased probability of cardiac syncope | Admission and cardiac investigation are warranted. An ECG, echocardiography and, where needed, prolonged ECG monitoring or an electrophysiological study should be considered. |
A score below 3 does not exclude cardiac syncope with certainty. Sensitivity is high but not perfect, and clinical assessment of the whole picture must always be weighed in. Conversely, a score of 3 or higher does not confirm a cardiac cause, since specificity is only moderate. False positives occur, particularly in older patients with an abnormal ECG from a cause other than the one provoking the syncope.
Validation and performance
External validation of EGSYS is limited but broadly consistent with the derivation study. In a prospective validation at three university hospitals in Tehran, 198 patients with syncope were included, of whom 115 (58 per cent) had a cardiac cause [2]. The area under the ROC curve was 0.82 for the univariable and 0.81 for the multivariable model, indicating good discrimination [2]. At a threshold of 3 points, sensitivity was 86 per cent (univariable) and 91 per cent (multivariable), and specificity 68 and 57 per cent respectively [2]. Sensitivity was thus somewhat lower than in the derivation cohort, while specificity was in the same range.
A systematic review in JAMA in 2019, which collated 11 studies of clinical examination in suspected cardiac syncope, included two prospective validations of EGSYS [3]. The authors found that a score below 3 was associated with a lower probability of cardiac syncope, with a sensitivity of 89 to 91 per cent and a specificity of 69 to 73 per cent, corresponding to a negative likelihood ratio of 0.12 to 0.17 [3]. The review thus supports the view that EGSYS is strongest as a rule-out instrument: a low score appreciably reduces the probability of cardiac syncope.
In a retrospective comparison of the EGSYS and OESIL scores in 153 patients in the Czech Republic, both scoring systems distinguished cardiac from non-cardiac syncope with a statistically significant difference between the groups [4]. The authors recommended EGSYS for acute management because it combined high sensitivity with good specificity.
Calibration has not been reported systematically in any of the available studies, which is a gap in the evidence.
Limitations
EGSYS was developed for patients presenting acutely with syncope in whom the cause is initially unclear. The score does not apply to patients in whom the history or initial examination already gives a clear diagnosis, and it is not validated in children. A study of EGSYS in the outpatient setting, in which patients had already undergone medical assessment before being scored, showed markedly lower sensitivity (48 per cent) at a threshold of 3, while specificity rose to 78 per cent [5]. This indicates that the instrument loses value when the pre-test probability has already been adjusted by earlier investigation.
The negative points for autonomic prodromes and precipitating factors depend on the presence of specific features (nausea/vomiting and a warm environment, prolonged standing, fear/pain/emotional distress respectively). The absence of these features scores 0, not minus points, which means that a patient with neither cardiac features nor reflex-mediated prodromes can end up at zero points, a value at which interpretation becomes uncertain.
An important pitfall is that the variable "abnormal ECG and/or heart disease" is a composite. An abnormal ECG of non-cardiac origin — for example non-specific ST-T changes in an older patient — can score 3 points and thereby push the score to the threshold for cardiac syncope without the probability actually being raised. Interpretation of the ECG requires clinical judgement and should not be scored mechanically without context.
No calibration data have been published, and external validation is geographically limited to Iran, the Czech Republic and Portugal. Generalisability to Scandinavian populations has not been tested directly.
References
- Del Rosso A, Ungar A, Maggi R, et al. Clinical predictors of cardiac syncope at initial evaluation in patients referred urgently to a general hospital: the EGSYS score. Heart 2008;94(12):1620-6. PMID: 18519550
- Kariman H, Harati S, Safari S, et al. Validation of EGSYS Score in Prediction of Cardiogenic Syncope. Emerg Med Int 2015;2015:515370. PMID: 26649200
- Albassam OT, Redelmeier RJ, Shadowitz S, et al. Did This Patient Have Cardiac Syncope?: The Rational Clinical Examination Systematic Review. JAMA 2019;321(24):2448-2457. PMID: 31237649
- Plasek J, Doupal V, Fürstova J, et al. The EGSYS and OESIL risk scores for classification of cardiac etiology of syncope: comparison, revaluation, and clinical implications. Biomed Pap Med Fac Univ Palacky Olomouc Czech Repub 2010;154(2):169-73. PMID: 20668500
- de Sousa Bispo J, Azevedo P, Mota T, et al. EGSYS score for the prediction of cardiac etiology in syncope: Is it useful in an outpatient setting? Rev Port Cardiol 2020;39(5):255-261. PMID: 32534800
