Ava L Liberman1, Shyam Prabhakaran2, Faddi G Saleh Velez2, Ronald Alvarado-Dyer2, Camila Bonin Pinto3,4, Jorge G Ortiz García5, Daryl Mchugh1, Jenny Lu6, Oleg Otlivanchik1, Brent L Flusty1. 1. Department of Neurology, Montefiore Medical Center, Albert Einstein College of Medicine, Bronx, NY (D.M., O.O., B.L.F., A.L.L.). 2. Department of Neurology, University of Chicago Medical Center, University of Chicago, IL (F.G.S.V., R.A.-D., S.P.). 3. Institute of Psychology, University of Sao Paulo, Brazil (C.B.P.). 4. Department of Physiology, Northwestern University, Chicago, IL (C.B.P.). 5. Department of Neurology, University of Oklahoma Health Science Center (J.G.O.G.). 6. Albert Einstein College of Medicine, Bronx, NY (J.L.).
Abstract
BACKGROUND: Missed or delayed diagnosis of acute stroke, or false-negative stroke (FNS), at initial emergency department (ED) presentation occurs in ≈9% of confirmed stroke patients. Failure to rapidly diagnose stroke can preclude time-sensitive treatments, resulting in higher risks of severe sequelae and disability. In this study, we developed and tested a modified version of a structured medical record review tool, the Safer Dx Instrument, to identify FNS in a subgroup of hospitalized patients with stroke to gain insight into sources of ED stroke misdiagnosis. METHODS: We conducted a retrospective cohort study at 2 unaffiliated comprehensive stroke centers. In the development and confirmatory cohorts, we applied the Safer Stroke-Dx Instrument to report the prevalence and documented sources of ED diagnostic error in FNS cases among confirmed stroke patients upon whom an acute stroke was suspected by the inpatient team, as evidenced by stroke code activation or urgent neurological consultation, but not by the ED team. Inter-rater reliability and agreement were assessed using interclass coefficient and kappa values (κ). RESULTS: Among 183 cases in the development cohort, the prevalence of FNS was 20.2% (95% CI, 15.0-26.7). Too narrow a differential diagnosis and limited neurological examination were common potential sources of error. The interclass coefficient for the Safer Stroke-Dx Instrument items ranged from 0.42 to 0.91, and items were highly correlated with each other. The κ for diagnostic error identification was 0.90 (95% CI, 0.821-0.978) using the Safer Stroke-Dx Instrument. In the confirmatory cohort of 99 cases, the prevalence of FNS was 21.2% (95% CI, 14.2-30.3) with similar sources of diagnostic error identified. CONCLUSIONS: Hospitalized patients identified by stroke codes and requests for urgent neurological consultation represent an enriched population for the study of diagnostic error in the ED. The Safer Stroke-Dx Instrument is a reliable tool for identifying FNS and sources of diagnostic error.
BACKGROUND: Missed or delayed diagnosis of acute stroke, or false-negative stroke (FNS), at initial emergency department (ED) presentation occurs in ≈9% of confirmed stroke patients. Failure to rapidly diagnose stroke can preclude time-sensitive treatments, resulting in higher risks of severe sequelae and disability. In this study, we developed and tested a modified version of a structured medical record review tool, the Safer Dx Instrument, to identify FNS in a subgroup of hospitalized patients with stroke to gain insight into sources of ED stroke misdiagnosis. METHODS: We conducted a retrospective cohort study at 2 unaffiliated comprehensive stroke centers. In the development and confirmatory cohorts, we applied the Safer Stroke-Dx Instrument to report the prevalence and documented sources of ED diagnostic error in FNS cases among confirmed stroke patients upon whom an acute stroke was suspected by the inpatient team, as evidenced by stroke code activation or urgent neurological consultation, but not by the ED team. Inter-rater reliability and agreement were assessed using interclass coefficient and kappa values (κ). RESULTS: Among 183 cases in the development cohort, the prevalence of FNS was 20.2% (95% CI, 15.0-26.7). Too narrow a differential diagnosis and limited neurological examination were common potential sources of error. The interclass coefficient for the Safer Stroke-Dx Instrument items ranged from 0.42 to 0.91, and items were highly correlated with each other. The κ for diagnostic error identification was 0.90 (95% CI, 0.821-0.978) using the Safer Stroke-Dx Instrument. In the confirmatory cohort of 99 cases, the prevalence of FNS was 21.2% (95% CI, 14.2-30.3) with similar sources of diagnostic error identified. CONCLUSIONS: Hospitalized patients identified by stroke codes and requests for urgent neurological consultation represent an enriched population for the study of diagnostic error in the ED. The Safer Stroke-Dx Instrument is a reliable tool for identifying FNS and sources of diagnostic error.
Authors: Ava L Liberman; Ahmed Hassoon; Mehdi Fanai; Shervin Badihian; Hetal Rupani; Susan M Peterson; Krisztian Sebestyen; Zheyu Wang; Yuxin Zhu; Richard B Lipton; David E Newman-Toker Journal: Acad Emerg Med Date: 2021-08-18 Impact factor: 3.451