RESUMO
STUDY OBJECTIVE: To determine the sensitivity of dysphagia screening by emergency physicians on acute stroke patients. METHODS: To develop a 2-tiered dysphagia screen and performed it on a convenience sample of acute stroke patients. Tier 1 examined voice quality, swallowing complaints, facial asymmetry, and aphasia. Tier 2 involved a water swallow test, with evaluation for swallowing difficulty, voice quality compromise, and pulse oximetry desaturation (>or=2%). We classified patients passing both tiers as "low risk" and compared the screen's sensitivity to a formal assessment by speech language pathologists. To assess reproducibility, we performed 2 consecutive, blinded ED screens on a convenience sample of 32 patients. RESULTS: During 16 months, we enrolled a convenience sample of 103 patients, excluding 19 patients from data analysis for lack of a stroke discharge diagnosis (n=11), an incomplete speech language pathologist evaluation within 24 hours (n=7), or pneumonia on emergency department (ED) chest radiography (n=1). Of the 84 remaining patients, speech language pathologists identified dysphagia in 48. The sensitivity of the ED dysphagia screen was 96% (95% confidence interval [CI] 85% to 99%), with a negative likelihood ratio of 0.08 (95% CI 0.02 to 0.3). Reproducibility testing yielded a kappa for the overall screen result of 0.9 (95% CI 0.9-1.0) and a simple agreement of 97%. CONCLUSION: Preliminary data on the sensitivity and reliability of our ED dysphagia screening tool are promising. The simple screen provides an easy way for emergency physicians to identify acute stroke patients eligible for early oral medications and nutrition. Further validation and refinement of our screen are needed before its widespread adoption.