Abstract for presentation at The Australian and New Zealand Association of Neurologists Annual Scientific Meeting 2007

Utility of MRI in Adult First Seizure Patients - Development & Validation of a Clinical Algorithm

  • Dr Raju Yerra, Department of Neurology, Royal Melbourne Hospital & Department of Medicine, University of Melboure, Australia
  • Mr Adrian Lowe, Department of Neurology, The Royal Melbourne Hospital, Australia
  • Associate Prof Christine Kilpatrick, Department of Neurology, Royal Melbourne Hospital, VIC 3050 & Department of Medicine, University of Melboure VIC 3050, Australia
  • Dr Daniel Ghougassian, Department of Neurology, Alfred Hospital, VIC, Australia
  • Dr Zelko Matkovic, Department of Neurology, The Royal Melbourne Hospital, Australia
  • A/Prof Terence O'Brien, Department of Neurology, Royal Melbourne Hospital, Department of Medicine, University of Melboure, Australia

    • Rationale: There is a growing practice to perform MRI on patients who have experienced a first seizure. However, the majority of these studies are normal, raising questions about the utility of this practice. This study aimed to develop and validate an algorithm to select patient in whom the positive yield of MRI is highest, and conversely those in whom it is low.
    Methods: Using data from a study of 105 consecutive adult patients from a hospital-based First Seizure Clinic, using logistic regression, a clinical algorithm was developed, to determine which patients should be offered an MRI. Data from 100 patients from a second hospital based First Seizure Clinic were used to validate the predictive value of the algorithm.
    Results:
    Development cohort: MRI provided useful information additional to CT in 16.2% of patients. Independent predictors of added value from MRI were (i) a CT showing lesion that required clarification (p<0.001), (ii) partial or unclassifiable seizure type (p=0.001), and (iii) history of head-injury (p=0.011). A clinical algorithm was proposed where MRI was only offered to patients with one or more of these factors. The algorithm had a sensitivity of 100%, and a specificity of 79.6%. The negative predictive value if all the above three features were not present was 98.57.
    Validation cohort: 9% of patients gained additional value from MRI, and the algorithm had a sensitivity of 100%, and a specificity of 47.25%. The negative predictive value if all the above three features were not present was 95.16.
    Conclusions: The results of this study identify subgroups of patients in whom MRI has high yield for adding value to CT following first seizure and in whom performance of MRI should be given priority, and also allows identification of a sub group where MRI can be performed non-urgently without compromising patient management and safety

    Conference Organiser - ICMS Pty Ltd