Written and maintained by CASRAI Editorial Board
Last updated
The San Francisco Syncope Rule (SFSR) is a five-item clinical decision rule, remembered by the mnemonic CHESS, used in the emergency department to help stratify patients presenting with syncope (transient loss of consciousness) by their short-term risk of a serious underlying cause. It was derived to help answer a specific disposition question — does this syncope patient need admission and further workup, or can they be safely discharged — and it remains one of the most widely taught syncope decision rules in emergency medicine, even though its real-world performance has turned out to be a genuinely unsettled question. This guide covers the CHESS criteria, what the rule was originally derived and validated to predict, and — because this matters more for patient-safety purposes than the rule’s popularity suggests — the substantial gap between its original sensitivity figures and what independent, external validation studies have since found.
The CHESS criteria
SFSR flags a patient as higher-risk if any one of the five CHESS criteria is present. A patient negative on all five is classified as lower-risk.
| CHESS letter | Criterion |
|---|---|
| C | History of congestive heart failure |
| H | Hematocrit < 30% |
| E | Abnormal ECG (new changes from a prior tracing, or any non-sinus rhythm) |
| S | Shortness of breath |
| S | Systolic blood pressure < 90 mm Hg at triage |
All five criteria are assessed from information already routinely available at the index ED visit — history, a triage vital sign, and a basic metabolic/CBC panel plus ECG — which is a large part of why the rule was designed to be usable without waiting on further testing.
What SFSR was derived to predict, and how well it did
SFSR was derived by Quinn and colleagues (Quinn JV et al., Annals of Emergency Medicine, February 2004) in a single-center, prospective cohort of 684 ED patients presenting with syncope or near-syncope. The outcome the rule was built to predict was a pre-defined short-term serious outcome — death, myocardial infarction, arrhythmia, pulmonary embolism, stroke, subarachnoid hemorrhage, significant hemorrhage, or any condition causing a related return ED visit and hospitalization — occurring within 7 days of the index visit. In the derivation cohort, the rule (any CHESS criterion positive) achieved 96% sensitivity (95% CI, 92–100%) and 62% specificity for that 7-day serious-outcome definition.
A subsequent prospective validation by the same investigators (Quinn et al., 2006, 791 consecutive patients at the University of California, San Francisco) reported even higher sensitivity: 98% (95% CI, 89–100%). Those two figures — 96% and 98% — are the numbers most commonly quoted when SFSR is introduced, and on their own they make the rule look close to ideal for a rule-out tool: high sensitivity, low miss rate, a clear rationale for discharging CHESS-negative patients.
The validation gap: what happened when other centers tested it
That optimistic picture did not hold up when investigators outside the original UCSF group applied SFSR to their own syncope populations. Several independent external validation studies found meaningfully lower sensitivity than either the derivation or the original prospective validation reported:
- Birnbaum et al. external validation (Annals of Emergency Medicine): sensitivity 89% (95% CI, 81–97%) — notably lower than the 96–98% originally reported, in a cohort the original derivation team had no role in selecting.
- A Los Angeles-based validation cohort: sensitivity around 89%.
- An Australian validation cohort: sensitivity around 90%.
- Montefiore Medical Center (713 patients, 97% follow-up): the lowest reported figure, sensitivity 74% — the rule failed to flag 16 of 61 patients who went on to have a serious outcome, including 1 death, 8 arrhythmias, 3 strokes, and 1 subarachnoid hemorrhage.
That is a consistent pattern across independent centers, not one outlier study: every external cohort tested came in below the 96–98% originally reported, several well below it. A commonly cited review of this evidence base concluded that these external sensitivity figures are too low to rely on SFSR alone for a discharge decision — a conclusion that follows directly from what a mid-70s-to-high-80s sensitivity means operationally: a meaningful share of patients with a genuine serious cause of their syncope would be classified as lower-risk and sent home if the rule’s negative result were treated as a stand-alone discharge criterion.
The likely drivers of the gap are the usual reasons a single-center derivation rule performs worse elsewhere: differences in how “abnormal ECG” and other criteria were interpreted and documented by different clinicians, differences in the underlying case mix and prevalence of serious outcomes across EDs, and the well-known tendency for any clinical decision rule to look better in the population and setting it was built in than in populations it wasn’t. None of that is unique to SFSR — it is the same reason external validation, not just derivation, is the real bar for whether a decision rule is trustworthy in a different hospital.
How to use CHESS without over-relying on it
For a hospital ED, patient-safety, or quality office, the practical implications of the validation gap are straightforward:
- Treat a CHESS-positive result as a genuine red flag. Any single positive criterion identifies a patient the derivation and validation evidence consistently agrees is higher-risk; the disagreement across studies is about the negative (all-criteria-absent) side of the rule, not the positive side.
- Do not treat a CHESS-negative result as license to discharge without clinical judgment. Given that independent cohorts have reported sensitivity as low as 74–90%, a “CHESS-negative” patient still carries a non-trivial residual risk that the original 96–98% figures understate. SFSR is decision support, not a disposition mandate — the same principle CASRAI’s other emergency-medicine decision-rule guides apply to a documented clinical gestalt override.
- Document the reasoning, not just the score. “SFSR: CHESS-negative; disposition based on rule plus clinical assessment” is more defensible on retrospective chart review than a bare “low risk,” and it creates the audit trail a patient-safety office needs if an adverse outcome is later reviewed.
- Use SFSR alongside — not instead of — a broader syncope workup that already accounts for age, cardiac history, and exam findings; SFSR was never intended to replace a clinician’s overall assessment, only to structure one input into it.
For a related example of a widely used ED decision rule whose reported performance shifted once it was tested outside its derivation population, see CASRAI’s PERC Rule for Pulmonary Embolism guide, where the pattern runs the opposite direction: PERC’s external and pooled sensitivity has stayed close to its original figures. The contrast is itself useful for a quality office — it illustrates why “the derivation study reported X% sensitivity” is never sufficient justification on its own for building a discharge pathway around a rule; the external-validation literature has to be checked criterion by criterion.
Where SFSR fits in hospital risk stratification
SFSR sits in the same family of ED clinical decision rules CASRAI covers elsewhere — instruments meant to structure a specific triage or disposition decision using data already available at the bedside, rather than requiring new testing. See CASRAI’s Wells Criteria for DVT and PE, CHA2DS2-VASc Score, and NIH Stroke Scale (NIHSS) guides for other examples in this category, and CASRAI’s TIMI Score guide for a comparable risk-stratification instrument used for a different presenting complaint. For the incident-review side of what happens when a discharged patient returns with a missed serious diagnosis, see CASRAI’s root cause analysis and sentinel event guides. For the wider hospital patient-safety context these instruments sit inside, see CASRAI’s Patient Safety & Infection Prevention pillar.
Frequently asked questions
What does CHESS stand for in the San Francisco Syncope Rule?
Congestive heart failure history, Hematocrit under 30%, abnormal ECG, Shortness of breath, and Systolic blood pressure under 90 mm Hg at triage. Any one of the five present classifies the patient as higher-risk under the rule.
What is the San Francisco Syncope Rule actually used for?
It’s a bedside decision aid for ED risk-stratification of syncope patients — helping decide whether a patient needs admission and further cardiac/neurologic workup, or can reasonably be considered for discharge, based on their short-term risk of a serious underlying cause of the syncopal episode.
Is the San Francisco Syncope Rule reliable enough to base a discharge decision on by itself?
Not on its own. The original derivation (96% sensitivity) and prospective validation (98% sensitivity) numbers are frequently cited, but multiple independent external validation studies have reported meaningfully lower sensitivity — around 89–90% in Los Angeles-based and Australian cohorts, and as low as 74% at Montefiore Medical Center, where the rule missed 16 of 61 patients with a genuine serious outcome. A published review of this evidence concluded those external figures are too low to rely on the rule alone for a discharge decision.
Why does the rule perform worse outside the hospital where it was developed?
The gap is consistent with the general pattern seen across clinical decision rules: performance measured in the population and setting where a rule was derived and initially validated tends to look better than performance in an independent, external population, due to differences in case mix, criterion interpretation, and local documentation practice. External validation, not derivation-study performance alone, is what determines whether a rule is trustworthy in a different hospital.
