Predictive model for diagnosing central lesions in emergency department patients with isolated dizziness who undergo diffusion-weighted magnetic resonance imaging.

Abstract

Only 5% to 10% of patients who visit the emergency department (ED) with isolated dizziness without neurologic abnormalities may have central lesions; however, it is important to distinguish central lesions through brain imaging. This study was conducted to create a nomogram to provide an objective medical basis for selectively performing magnetic resonance imaging (MRI) among patients with isolated dizziness. This retrospective observational study enrolled patients who visited the ED of a tertiary hospital with isolated dizziness and underwent diffusion-weighted MRI and subsequently consulted with the departments of neurology, neurosurgery, or otorhinolaryngology. Multivariable logistic regression analysis was performed to identify risk factors in patients diagnosed with central lesions to create a nomogram with the significant variables. Of the 1,078 patients who were screened, 119 were diagnosed with central lesions. Significant variables in the multivariable logistic regression analysis were albumin levels (odds ratio [OR]= 0.339, 95% confidence interval [CI]= 0.188 to 0.610, p=0.0003), inorganic phosphate levels (OR= 0.891, 95% CI= 0.832 to 0.954, p=0.0010), history of ischemic stroke (OR= 3.170, 95% CI= 1.807 to 5.560, p<0.0001), presyncope (OR= 3.152, 95% CI= 1.184 to 8.389, p=0.0216), and nystagmus (OR= 0.365, 95% CI= 0.237 to 0.561, p<0.0001). The area under the receiver operating characteristic curve of the nomogram created with these variables was 0.7315 (95% CI= 0.6842 to 0.7788, p<0.0001). Albumin, inorganic phosphate, previous stroke, presyncope, and nystagmus were associated with the predictive diagnosis of central lesions among patients admitted to the ED with isolated dizziness. The novel nomogram created using these variables can help in objectively determining the need for MRI in patients presenting with isolated dizziness to the ED.