The poster can be accessed here [https://mallinckrodt.gcs-web.com/static-files/e0881e7b-d796-4133-b089-7f558db5c3e2] on the company's website.
Using claims data from 10.8 million patients under two years of age and input from IS medical experts, researchers developed a predictive model to identify combinations of factors that may best predict IS. The deductive model found that two or more symptoms, including developmental delays, convulsions, dysphagia and cerebral palsy, coupled with a moderate or high severity emergency department (ED) visit, were strong predictors for identification of IS.
"If we can leverage machine learning algorithms in electronic medical systems, we may be able to identify those children at higher risk of developing infantile spasms, with the possibility of creating alert systems, decreasing time from symptom onset to diagnosis, and ultimately improving outcomes for this potentially devastating form of epilepsy," said Dr. Adam L. Numis, pediatric neurologist and epileptologist, UCSF Benioff Children's Hospital, San Francisco, CA.
Deductive models developed to identify the combinations of variables that predict IS prior to diagnosis and during the treatment pathway: 557 patients had ≥2 symptoms pertaining to IS – seizures, developmental delay, lack of eye contact and lack of muscle tone – and a moderate or high severity ED visit.
In this group, 55 percent of patients (n=304) had an IS diagnosis within a median 0.8 months of the triggering event. The most notable deductive combinations focused on pre-diagnosis of IS.
The study utilized data from the Symphony Health Integrated Dataverse database to identify triggers for early identification of IS patients from medical and pharmacy claims from 10.8 million patients under 2 years of age between May 2017 and April 2018.
International Classification of Diseases (ICD) procedure codes were evaluated to identify the codes most likely to predict a subsequent diagnosis of IS.
Researchers also used input from IS medical experts to determine the clinical, electrographic, radiologic, procedural and medication variables that may predict IS development.
The use of ICD codes to evaluate IS may not identify all IS cases.
The deductive analysis relied on the accuracy of coding for diagnoses and procedures in the ED.
The analysis was supported by Mallinckrodt.
"We're proud to present this data in support of faster diagnosis of infants during IS Awareness Week, which seeks to raise awareness of the subtle signs of IS that are often overlooked and the urgent need for prompt diagnosis and treatment," said Tunde Otulana, M.D., Chief Medical Officer at Mallinckrodt. "Mallinckrodt is committed to conducting research and providing critical services to help caregivers and infants with IS get the proper diagnosis as quickly as possible to enable earlier treatment intervention with the goal of improving long-term outcomes."