• 
• 

Three Part Question

In [adult patients with transient loss of consciousness presenting to an ED], is [SFSR better than OESIL, EGSYS or ROSE] at [minimising admission whilst reducing mortality]?

Search Strategy

Medline (1948 to June Week 4 2011) and Embase (1980 to week 26 2011) using the Ovid interface.
{[(transient loss of consciousness.mp) OR (TLOC.mp.) OR (T-LOC.mp.) OR (exp Unconsciousness/) OR (exp Syncope/) OR (exp Syncope, Vasovagal/)] AND [(SFSR.mp.) OR (San Francisco Syncope Rule.mp.) OR (OESIL.mp.) OR (EGSYS.mp.) OR (ROSE.mp.)]} LIMIT to humans AND english language.

Only the first two parts of the three-part question were used in the search, as the results had been narrowed down sufficiently by this point.

Search Outcome

In Medline 88 papers were found of which 18 were relevant, however one was deemed to be of insufficient quality to be included. In Embase 112 papers were found, of which 17 were duplicates.
Of the 17 relevant papers, one was a meta-analysis encompassing 11 of the others (however the paper deriving the ROSE rule was appraised separately, being the only study to evaluate the rule).

Relevant Paper(s)

Author, date and country	Patient group	Study type (level of evidence)	Outcomes	Key results	Study Weaknesses
Serrano, L., et al 2010 United States	SFSR: 4363 patients taken from 9 studies. Two studies included children and one had only ≥65 year old patients. Most studies used an outcome period of 7 days, two used 30 days and one used 10 days. OESIL: 1185 patients taken from 3 studies. One included older children (>12). The outcome periods ranged from 7 days to one year.	Meta-analysis	SFSR: Sensitivity	86%	Variations of syncope definitions. Variations of ECG interpretation (abnormalities included and the individual who interprets). Variations of outcome period.
			SFSR: Specificity	49%
			OESIL: Sensitivity	95%
			OESIL: Specificity	31%
Quinn, J, et al 2005 United States	684 patients screened for syncope, loss of consciousness, fall, collapse, seizure, light-headedness, tachycardia, bradycardia, shortness of breath, and chest pain. The admitting physician was asked to predict the likelihood of serious outcome within 7 days, and this was compared to the SFSR result. Same patient group as used in the previous derivation study.	Prospective cohort study	SFSR: Sensitivity for serious outcome	96%	Not an external study to the group who derived the SFSR, conflict of interest? Includes paediatric patients.
			SFSR: Specificity for serious outcome	62%
			Physician Judgement Sensitivity for serious outcome	94%
			Physician Judgement Specificity for serious outcome	52%
Numeroso, F., et al 2010 Italy	200 patients randomly selected from a group of 597 syncope patients admitted to the ED ward over a three month period. Patients were included by a presentation of T-LOC, and then excluded if they were <18, took over 24 hours to be referred and if a clear alternative diagnosis was found i.e. epilepsy. The patient's outcomes were compared to that predicted by the OESIL score.	Retrospective cohort study	OESIL: Sensitivity for cardiac syncope, ≥2 points	77.7%	No reference to statistical power. A selection bias from only including patients who were admitted to the ED ward - missing those deemed lower risk. Does not include all T-LOC patients having excluded some with an epilepsy diagnosis. Poor presentation of results.
			OESIL: Specificity for cardiac syncope, ≥2 points	59.3%
Del Rosso, A., et al 2008 Italy	516 consecutive patients from 14 hospitals over one month with unexplained syncope. The first 260 patients formed the derivation group. The second 256 patients formed the validation group. Patients with definite non-syncopal causes of T-LOC from initial evaluation were excluded. 52 variables were collected from each patient and used to the derive the EGSYS score.	Prospective cohort study	EGSYS: Sensitivity for cardiac syncope (derivation cohort)	95%	No reference to statistical power. Patients with other causes of T-LOC, i.e. seizure, were not included.
			EGSYS: Specificity for cardiac syncope (derivation cohort)	61%
			EGSYS: Sensitivity for cardiac syncope (validation cohort)	92%
			EGSYS: Specificity for cardiac syncope (validation cohort)	69%
Reed, M., et al 2007 UK	99 patients presenting with syncope to the ED over a 3 month period. Patients were flagged by a triage nurse and confirmed by a doctor, 9 patients were rejected in this way due to other diagnoses or lack of consent. Exclusion criteria were >16 years of age, and a history of seizure with prolonged post-ictal phase. The patient's outcomes were compared with those predicted by the SFSR and OESIL score.	Prospective cohort study	SFSR: Sensitivity for serious outcome	100%	Retrospective analysis of patient records revealed 163 patients were missed, and there was a significant difference in age between the two groups. No statistical power calculation was used, but it was acknowledged that the sample size was too small. No mention of ethical approval.
			SFSR: Specificity for serious outcome	45.5%
			OESIL: Sensitivity for serious outcome, ≥2 points	90.9%
			OESIL: Specificity for serious outcome, ≥2 points	48.9%
Quinn, J, et al 2008 United States	1418 consecutive patients from the ED with syncope over a 45 month period. Patients were enrolled by doctors, students and an electronic system. Exclusion criteria were trauma, alcohol and drugs induced T-LOC, and definite seizures. Same patient group as used in the previous derivation and validation studies. The 12 month mortality was recorded and compared to determine how many the SFSR could predict.	Prospective cohort study	SFSR: Sensitivity for syncope-related mortality at 1 year	93%	Not an external study to the group who derived the SFSR, conflict of interest? Includes paediatric patients.
			SFSR: Specificity for syncope-related mortality at 1 year	53%
Reed, M., et al 2010 UK	1067 patients aged ≥16 presenting with acute syncope to the ED. Patients were excluded for lack of consent, suspected stroke, alcohol consumption, hypoglycaemia and trauma. The outcome period was one month. 529 patients were used to derive the rule, and 538 to validate it. 93 variables were collected for each patient and used to derive the score.	Prospective cohort study	ROSE: Sensitivity for serious outcome (derivation cohort)	92.5%	No external validation.
			ROSE: Specificity for serious outcome (derivation cohort)	73.8%
			ROSE: Sensitivity for serious outcome (validation cohort)	87.2%
			ROSE: Specificity for serious outcome (validation cohort)	65.5%

Comment(s)

The meta-analysis demonstrates that the pooled sensitivity and specificity of the San Francisco Syncope Rule is much lower than was claimed upon derivation. There are possible reasons given for this, such as variation of how the ECG was interpreted. Interestingly, the pooled result is now lower than that of the physician's judgement from Quinn, 2005, a study published by the rule's authors to strengthen it. Further research conducted with stricter adherence to the original guidelines could be useful to prove the SFSR's worth, however the variation displayed in the different papers can be argued to mimic that found in different hospitals. According to the meta-analysis, the OESIL rule fares better than the SFSR. However when other papers are taken into account, its success is lowered. That said, one of the papers (Numeroso) can be argued to be of a lower quality than the others. If one were to give that paper less weight, the OESIL score is put roughly on a par with SFSR. Of course, SFSR would then still have a much greater level of external validation on it's side. There is a lack of external validation of the EGSYS score. The authors of the score did have good success with it with their group of 516 patients, in what appears to be a strong study conducted over 14 hospitals. The ROSE score gives a similar picture, except in this case there are twice the patients, 1067 but only from one hospital. More research would be welcomed as external validation would be a tremendous reassurance to the reliability of these two rules.

Clinical Bottom Line

It is difficult to choose one decision rule that is clearly superior, but the ROSE score has a very good result of 90%/70% derived from a strong study, and the benefit of being the only rule based in the UK. It is for these reasons it manages to stand out above the others.

References

1. Serrano, L., Hess, E., Bellolio, F., et al. Accuracy and Quality of Clinical Decision Rules for Syncope in the Emergency Department: A Systematic Review and Meta-analysis. Ann Emerg Med 2010; 362-373
2. Quinn, J., Stiell, I., McDermott, D., et al. The San Francisco Syncope Rule vs physician judgment and decision making. Am J Emerg Med 2005; 782-786
3. Numeroso, F., Mossini, G., Spaggiari, E., et al. Syncope in the emergency department of a large northern Italian hospital: incidence, efficacy of a short-stay observation ward and validation of the OESIL risk score. Emerg Med J 2010; 653-658
4. 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; 1620-1626
5. Reed, M., Newby, D., Coull, A., et al. The Risk stratification Of Syncope in the Emergency department (ROSE) pilot study: a comparison of existing syncope guidelines. Emerg Med J 2007; 270-275
6. Quinn, J., McDermott, D., Kramer, N., et al. Death After Emergency Department Visits for Syncope: How Common and Can It Be Predicted? Ann Emerg Med 2008; 585-590
7. Reed, M., Newby, D., Coull, A., et al. The ROSE (Risk Stratification of Syncope in the Emergency Department) Study. J Am Coll Cardiol 2010; 713-721