What is the Sudbury Vertigo Risk Score?
The Sudbury Vertigo Risk Score is a seven-item, weighted clinical prediction model developed to estimate the probability that an adult presenting to the emergency department (ED) with acute vertigo, dizziness, or imbalance has a serious underlying diagnosis within a short horizon. In the derivation work, a serious diagnosis was defined as stroke (ischemic or hemorrhagic), transient ischemic attack (TIA), vertebral artery dissection, or brain tumor, adjudicated from hospital records, imaging, and structured follow-up. The score is named for the lead institution’s region (Sudbury, Ontario) and is intended as a bedside adjunct to history, examination, and disposition decisions—not as a substitute for clinical judgment, neuroimaging when indicated, or specialist consultation.
Vertigo and dizziness are among the most common chief complaints in the ED. Most patients have benign peripheral or functional explanations, but a minority harbor posterior circulation stroke, dissection, or other neurologic emergencies. Traditional imaging strategies have imperfect sensitivity for early infarction on non-contrast CT, while advanced imaging is resource-intensive. A transparent risk score that uses routinely available history and examination findings can help teams risk-stratify, communicate pretest probability, and align further testing with patient-specific risk—provided its limitations and derivation context are understood.
Study design and cohort (derivation)
The score was derived in a multicenter prospective cohort conducted over three years at three Canadian university-affiliated urban tertiary care emergency departments. Consecutive alert patients aged 18 years or older were enrolled when the treating emergency physician assessed them for a chief complaint of acute vertigo, dizziness, or imbalance. Key exclusions included symptom onset more than 14 days prior, recent head or neck trauma, reduced level of consciousness, hypotension, recent syncope, and active cancer. Attending emergency physicians or supervised emergency medicine residents recorded structured data on dozens of candidate clinical variables; research staff verified forms and integrated objective data from the chart, nursing documentation, consultants, and imaging reports. Telephone follow-up at multiple time points supplemented ascertainment of outcomes for some patients.
The enrolled cohort was representative of real-world ED vertigo populations in many high-acuity settings: mean age was high, a majority of patients were women, and the overall prevalence of the composite serious outcome was on the order of a few percent—substantially higher than zero but far below “rule-in by prevalence alone.” That spectrum is important when interpreting positive and negative predictive value outside the original sites.
Outcome definition
The primary outcome was a serious diagnosis of stroke, TIA, vertebral artery dissection, or brain tumor, established if diagnosed during the index ED visit or within 30 days thereafter. Stroke and TIA definitions aligned with conventional clinical criteria used in emergency and stroke medicine. Vertebral artery dissection required supportive vascular imaging. Brain tumor required intracranial mass with clinical consequence. Events were reviewed by an adjudication committee blinded to the index visit details, with agreement rules to classify uncertain cases—strengthening outcome validity relative to single-reviewer chart review alone.
Model development and point assignment
Investigators began with a broad set of candidate predictors informed by prior literature and expert opinion, then used multivariable logistic regression to retain a parsimonious final model. Regression coefficients were converted to integer points by dividing each predictor’s beta coefficient by the smallest absolute beta among retained predictors and rounding to the nearest integer—a standard approach to produce a usable paper-and-electronic score without materially harming discrimination in many settings. The resulting model demonstrated strong discrimination in the derivation sample, with a high concordance (C-) statistic and reasonable calibration across score groups in reported analyses.
Scores above a high threshold were collapsed in some presentations because of sparse event counts at the extreme tail; the clinically relevant bands most often quoted in summaries are <5, 5–8, and >8.
Score components and point values
The published seven components and their weights are:
| Domain | Component | Points |
|---|---|---|
| Vascular risk / demographics | Male sex | +1 |
| Age greater than 65 years | +1 | |
| Diabetes | +1 | |
| Hypertension (history or prevalent hypertension at presentation) | +3 | |
| Focal or posterior circulation examination findings | Motor and/or sensory deficit on examination | +5 |
| Cerebellar signs or symptoms (e.g., diplopia, dysarthria, dysphagia, dysmetria, ataxia) | +6 | |
| Benign peripheral positional diagnosis | Clinical diagnosis of benign paroxysmal positional vertigo (BPPV)—for example positive Dix–Hallpike or supine roll testing, or a clear ED discharge diagnosis of BPPV in the source studies | −5 |
The total score is the algebraic sum of applicable items. Theoretical extremes are approximately −5 (isolated BPPV with no positive risk items) to +17 (all positive risk items without BPPV). Negative totals can occur when positional testing supports BPPV and few competing risk factors are present.
Interpretation: risk strata from the derivation cohort
Published summaries report the following observed proportions of the serious composite outcome by total score in the derivation data:
- Score <5: 0% of patients in that stratum experienced the adjudicated serious outcome in the reported analysis. Using <5 as a “low risk” band was associated with very high sensitivity for excluding the composite serious diagnosis in that cohort, at the cost of moderate specificity—meaning many patients without serious disease still score ≥5.
- Score 5–8: approximately 2.1% had a serious diagnosis—an intermediate band where shared decision-making, serial reassessment, and targeted investigation are commonly considered.
- Score >8: approximately 41% had a serious diagnosis—a high-risk band where urgent neuroimaging (often MRI with vascular imaging when available), neurology involvement, and cautious disposition are typically emphasized.
These percentages describe empirical risks in one prospective multicenter cohort; they are not biological constants. Pretest probability changes if the patient does not match the enrollment criteria, if the visit is subacute, or if the presentation is dominated by findings (such as an obvious cerebellar syndrome) that already mandate a central workup regardless of the numeric total.
Validation and transportability
A subsequent retrospective validation applied the same score and cutoffs to a large cohort of patients evaluated at the same urban Canadian tertiary sites over a different time window. In that analysis, a score <5 again identified a group with no adjudicated serious outcomes in the reported results, with sensitivity near 100% for the serious composite at that threshold and specificity on the order of roughly two-thirds—similar in spirit to the derivation performance though not identical in every numeric detail. Such “internal” or overlapping-site validation is a necessary intermediate step but does not replace prospective, multicenter, geographically diverse validation with standardized outcome assessment and clear imaging pathways.
Users should assume that real-world performance will depend on how faithfully each element—especially BPPV testing and documentation of focal deficits—is performed and recorded.
Clinical caveats and health information technology considerations
- BPPV is a double-edged lever in the score. A true BPPV diagnosis subtracts five points and can move a patient into a low numeric band; an incorrect label of BPPV (false positive positional testing or premature diagnostic closure) can falsely reassure. Conversely, failure to perform or document positional testing may omit a legitimate negative adjustment.
- Focal neurologic findings carry large positive weights. Many clinicians would pursue central evaluation even before summing the score; the model quantifies risk rather than replacing escalation rules.
- Imaging and follow-up bias affected the original studies: not every patient underwent MRI, and incomplete follow-up can distort sensitivity estimates in either direction depending on who is lost.
- Scope of “vertigo.” The derivation explicitly treated vertigo, dizziness, and imbalance as a single ED presentation family. The score may not apply to isolated presyncope, obvious cardiac syncope, medication toxicity, or other excluded conditions.
- Children, anticoagulated patients with head trauma, and hyperacute large-vessel occlusion presentations are not the target population; use condition-specific pathways when they apply.
Using this calculator on CalcMD
This tool implements the published integer weights and displays the component-wise contribution to the total, the aggregate score, and a short narrative aligned with the three derivation-cohort risk bands. It is offered for education, documentation, and shared understanding of published probabilities. It does not store patient identifiers, does not interpret imaging, and does not replace institutional protocols, telestroke consultation, or shared decision-making with the patient regarding radiation, contrast, transport, and follow-up.
When the numeric score and your bedside impression diverge, act on the higher level of concern, obtain appropriate specialty input, and document the rationale. Serial reassessment remains essential because posterior circulation symptoms may evolve during the ED stay.