User login
BANGKOK – A prediction tool that determines the risk of a pediatric epilepsy diagnosis eventually being made in a child who has had one or more paroxysmal events of possible epileptic origin is now available, and the clarity it provides makes life considerably easier for physicians and worried parents, Kees P. Braun, MD, PhD, said at the International Epilepsy Congress.
This prediction tool is highly practical. It relies upon certain clinical characteristics and a first interictal EEG, all information readily available at the time of the family’s first consultation with a neurologist or pediatrician with access to EEG, noted Dr. Braun, professor of neurology at Utrecht (the Netherlands) University.
The tool is freely available online (http://epilepsypredictiontools.info/first-consultation). The details of how Dr. Braun and coinvestigators developed the prediction tool have been published (Pediatrics. 2018 Dec;142[6]:e20180931. doi: 10.1542/peds.2018-0931), he said at the congress sponsored by the International League Against Epilepsy.
Early and accurate diagnosis or exclusion of epilepsy following a suspicious paroxysmal event deserves to be a high priority. Diagnostic delay is common, with resultant unrecognized recurrent epileptic seizures that can cause cognitive and behavioral impairments. And overdiagnosis of pediatric epilepsy unnecessarily exposes a child to the risks of antiepileptic drug therapy, not to mention the potential social stigma.
The predictive tool was developed through retrospective, multidimensional analysis of detailed data on 451 children who visited the outpatient pediatric neurology clinic at University Medical Center Utrecht for a diagnostic work-up after one or more paroxysmal events that might have been seizures, all of whom were subsequently followed for a year or longer. The resultant predictive model was then independently validated in a separate cohort of 187 children seen for the same reason at another Dutch university.
The model had an area under the receiver operating characteristic curve of 0.86, which statisticians consider to be excellent discriminatory power. The tool’s sensitivity and specificity varied according to the diagnostic probability threshold selected by the parents and physicians. For example, the predictive tool had a sensitivity of 18%, specificity of 99%, positive predictive value of 94%, and negative predictive value of 80% for identification of individuals with a greater than 80% probability of being diagnosed with epilepsy. For identification of all patients with a greater than 20% likelihood of receiving the diagnosis, the sensitivity was 73%, specificity 82%, positive predictive value 76%, and negative predictive value 79%.
The clinical characteristics incorporated in the predictive model include age at first seizure, gender, details of the paroxysmal event, and specifics of the child’s medical history. The relevant features of the standard interictal EEG recorded at the time of consultation include the presence or absence of focal epileptiform abnormalities if focal spikes or spike-wave complexes were detected, generalized epileptiform abnormalities in the presence of generalized spikes or spike-wave complexes, and nonspecific nonepileptiform abnormalities.
Future predictive refinements are under study
Dr. Braun and coworkers have reported that examining EEG functional network characteristics – that is, the functional networks of correlated brain activity in an individual patient’s brain – improves the EEG’s predictive value for epilepsy (PLoS One. 2013;8[4]:e59764. doi: 10.1371/journal.pone.0059764), a conclusion further reinforced in their systematic review and meta-analysis incorporating 11 additional studies (PLoS One. 2014 Dec 10;9[12]:e114606. doi: 10.1371/journal.pone.0114606).
In addition, the Dutch investigators have shown that ripples superimposed on rolandic spikes seen in scalp EEG recordings have prognostic significance. An absence of ripples superimposed on rolandic spikes identified children without epilepsy. In contrast, more than five ripples predicted atypical and symptomatic rolandic epilepsy with a substantial seizure risk warranting consideration of antiepileptic drug therapy (Epilepsia. 2016 Jul;57[7]:1179-89).
A Boston group using a fully automated spike ripple detector subsequently confirmed that ripples occurring in conjunction with epileptiform discharges on scalp EEG constitute a noninvasive biomarker for seizure risk that outperforms analysis of spikes alone and could potentially be useful in guiding medication tapering decisions in children (Brain. 2019 May 1;142[5]:1296-1309).
Dr. Braun reported having no financial conflicts regarding his presentation.
BANGKOK – A prediction tool that determines the risk of a pediatric epilepsy diagnosis eventually being made in a child who has had one or more paroxysmal events of possible epileptic origin is now available, and the clarity it provides makes life considerably easier for physicians and worried parents, Kees P. Braun, MD, PhD, said at the International Epilepsy Congress.
This prediction tool is highly practical. It relies upon certain clinical characteristics and a first interictal EEG, all information readily available at the time of the family’s first consultation with a neurologist or pediatrician with access to EEG, noted Dr. Braun, professor of neurology at Utrecht (the Netherlands) University.
The tool is freely available online (http://epilepsypredictiontools.info/first-consultation). The details of how Dr. Braun and coinvestigators developed the prediction tool have been published (Pediatrics. 2018 Dec;142[6]:e20180931. doi: 10.1542/peds.2018-0931), he said at the congress sponsored by the International League Against Epilepsy.
Early and accurate diagnosis or exclusion of epilepsy following a suspicious paroxysmal event deserves to be a high priority. Diagnostic delay is common, with resultant unrecognized recurrent epileptic seizures that can cause cognitive and behavioral impairments. And overdiagnosis of pediatric epilepsy unnecessarily exposes a child to the risks of antiepileptic drug therapy, not to mention the potential social stigma.
The predictive tool was developed through retrospective, multidimensional analysis of detailed data on 451 children who visited the outpatient pediatric neurology clinic at University Medical Center Utrecht for a diagnostic work-up after one or more paroxysmal events that might have been seizures, all of whom were subsequently followed for a year or longer. The resultant predictive model was then independently validated in a separate cohort of 187 children seen for the same reason at another Dutch university.
The model had an area under the receiver operating characteristic curve of 0.86, which statisticians consider to be excellent discriminatory power. The tool’s sensitivity and specificity varied according to the diagnostic probability threshold selected by the parents and physicians. For example, the predictive tool had a sensitivity of 18%, specificity of 99%, positive predictive value of 94%, and negative predictive value of 80% for identification of individuals with a greater than 80% probability of being diagnosed with epilepsy. For identification of all patients with a greater than 20% likelihood of receiving the diagnosis, the sensitivity was 73%, specificity 82%, positive predictive value 76%, and negative predictive value 79%.
The clinical characteristics incorporated in the predictive model include age at first seizure, gender, details of the paroxysmal event, and specifics of the child’s medical history. The relevant features of the standard interictal EEG recorded at the time of consultation include the presence or absence of focal epileptiform abnormalities if focal spikes or spike-wave complexes were detected, generalized epileptiform abnormalities in the presence of generalized spikes or spike-wave complexes, and nonspecific nonepileptiform abnormalities.
Future predictive refinements are under study
Dr. Braun and coworkers have reported that examining EEG functional network characteristics – that is, the functional networks of correlated brain activity in an individual patient’s brain – improves the EEG’s predictive value for epilepsy (PLoS One. 2013;8[4]:e59764. doi: 10.1371/journal.pone.0059764), a conclusion further reinforced in their systematic review and meta-analysis incorporating 11 additional studies (PLoS One. 2014 Dec 10;9[12]:e114606. doi: 10.1371/journal.pone.0114606).
In addition, the Dutch investigators have shown that ripples superimposed on rolandic spikes seen in scalp EEG recordings have prognostic significance. An absence of ripples superimposed on rolandic spikes identified children without epilepsy. In contrast, more than five ripples predicted atypical and symptomatic rolandic epilepsy with a substantial seizure risk warranting consideration of antiepileptic drug therapy (Epilepsia. 2016 Jul;57[7]:1179-89).
A Boston group using a fully automated spike ripple detector subsequently confirmed that ripples occurring in conjunction with epileptiform discharges on scalp EEG constitute a noninvasive biomarker for seizure risk that outperforms analysis of spikes alone and could potentially be useful in guiding medication tapering decisions in children (Brain. 2019 May 1;142[5]:1296-1309).
Dr. Braun reported having no financial conflicts regarding his presentation.
BANGKOK – A prediction tool that determines the risk of a pediatric epilepsy diagnosis eventually being made in a child who has had one or more paroxysmal events of possible epileptic origin is now available, and the clarity it provides makes life considerably easier for physicians and worried parents, Kees P. Braun, MD, PhD, said at the International Epilepsy Congress.
This prediction tool is highly practical. It relies upon certain clinical characteristics and a first interictal EEG, all information readily available at the time of the family’s first consultation with a neurologist or pediatrician with access to EEG, noted Dr. Braun, professor of neurology at Utrecht (the Netherlands) University.
The tool is freely available online (http://epilepsypredictiontools.info/first-consultation). The details of how Dr. Braun and coinvestigators developed the prediction tool have been published (Pediatrics. 2018 Dec;142[6]:e20180931. doi: 10.1542/peds.2018-0931), he said at the congress sponsored by the International League Against Epilepsy.
Early and accurate diagnosis or exclusion of epilepsy following a suspicious paroxysmal event deserves to be a high priority. Diagnostic delay is common, with resultant unrecognized recurrent epileptic seizures that can cause cognitive and behavioral impairments. And overdiagnosis of pediatric epilepsy unnecessarily exposes a child to the risks of antiepileptic drug therapy, not to mention the potential social stigma.
The predictive tool was developed through retrospective, multidimensional analysis of detailed data on 451 children who visited the outpatient pediatric neurology clinic at University Medical Center Utrecht for a diagnostic work-up after one or more paroxysmal events that might have been seizures, all of whom were subsequently followed for a year or longer. The resultant predictive model was then independently validated in a separate cohort of 187 children seen for the same reason at another Dutch university.
The model had an area under the receiver operating characteristic curve of 0.86, which statisticians consider to be excellent discriminatory power. The tool’s sensitivity and specificity varied according to the diagnostic probability threshold selected by the parents and physicians. For example, the predictive tool had a sensitivity of 18%, specificity of 99%, positive predictive value of 94%, and negative predictive value of 80% for identification of individuals with a greater than 80% probability of being diagnosed with epilepsy. For identification of all patients with a greater than 20% likelihood of receiving the diagnosis, the sensitivity was 73%, specificity 82%, positive predictive value 76%, and negative predictive value 79%.
The clinical characteristics incorporated in the predictive model include age at first seizure, gender, details of the paroxysmal event, and specifics of the child’s medical history. The relevant features of the standard interictal EEG recorded at the time of consultation include the presence or absence of focal epileptiform abnormalities if focal spikes or spike-wave complexes were detected, generalized epileptiform abnormalities in the presence of generalized spikes or spike-wave complexes, and nonspecific nonepileptiform abnormalities.
Future predictive refinements are under study
Dr. Braun and coworkers have reported that examining EEG functional network characteristics – that is, the functional networks of correlated brain activity in an individual patient’s brain – improves the EEG’s predictive value for epilepsy (PLoS One. 2013;8[4]:e59764. doi: 10.1371/journal.pone.0059764), a conclusion further reinforced in their systematic review and meta-analysis incorporating 11 additional studies (PLoS One. 2014 Dec 10;9[12]:e114606. doi: 10.1371/journal.pone.0114606).
In addition, the Dutch investigators have shown that ripples superimposed on rolandic spikes seen in scalp EEG recordings have prognostic significance. An absence of ripples superimposed on rolandic spikes identified children without epilepsy. In contrast, more than five ripples predicted atypical and symptomatic rolandic epilepsy with a substantial seizure risk warranting consideration of antiepileptic drug therapy (Epilepsia. 2016 Jul;57[7]:1179-89).
A Boston group using a fully automated spike ripple detector subsequently confirmed that ripples occurring in conjunction with epileptiform discharges on scalp EEG constitute a noninvasive biomarker for seizure risk that outperforms analysis of spikes alone and could potentially be useful in guiding medication tapering decisions in children (Brain. 2019 May 1;142[5]:1296-1309).
Dr. Braun reported having no financial conflicts regarding his presentation.
REPORTING FROM IEC 2019