An Early Prediction Algorirthm For Response to Treatment in Kawasaki Patients

Objectives: Can data from electronic medical records (including patient demographics, in-patient medications records, fever profile, laboratory profile, and echocardiograms) improve early prediction of response to treatment in Kawasaki patients? Specifically, we want to predict (1) whether patients with KD are responding to 1st line treatment (primary objective) and (2) if not, what kind of non-responsiveness is involved, and (3) what is likely to be the most appropriate second line therapy (secondary objectives).

Description

Kawasaki Disease (KD) is a self-limiting acute vasculitis of unknown etiology that affect pediatric patients. Left untreated or when treatment is delayed, KD can result in coronary artery aneurysms. Children with coronary artery aneurysms require lifelong cardiac care and are at high risk of coronary artery thrombosis and stenosis, both of which can be life threatening. In the US, 7,500 children are affected every year, and the incidence of KD has been increasing.

Once diagnosed, all patients received a treatment called intravenous immunoglobulin (IVIG), to which they can have one of three responses: complete response which results in the end of fever, partial response in which the fever subsides but then reoccur, and non-response in which fever is not affected by IVIG. Given that longer duration of fever is associated with exponentially increased risk of coronary complications, a faster determination of IVIG responsiveness is critical. Current predictive models for non-responsiveness to IVIG are poorly predictive (accuracy <60%) and only take in account the pre-treatment information. We aim to use the early fever profile of KD patients at Johns Hopkins to create a predictive model that will, in real time, warn treating physicians that a patient is or is not responding treatment; in the latter case, the model would also be trained to recommend a specific therapy in patients who are non-responsive.

Through this study we will certainly address an important clinical question which could help address the single largest cause of acquired heart disease in children in the developed world. The work we are doing could prevent the development of lifelong cardiac complications in young patients which often have devastating consequences.