Abstract:Objective:To investigate the related indicators associated with poor short-term prognosis of ischemic stroke (IS) in PICU, and to establish a multivariate prognostic prediction model for IS in PICU. Methods:The consecutive hospitalized patients with acute IS who were admitted to PICU in Children's Hospital of Fudan University from January 2013 to December 2018 were included. Short-term prognosis was evaluated according to Pediatric Cerebral Performance Category score (PCPC) at hospital discharge. The poor prognosis was defined as PCPC≥4 at hospital discharge, and the good prognosis was PCPC<4. Univariate analysis and binary logistic stepwise regression analysis were performed to screen the related indicators associated with poor prognosis including patients' basic information, clinical situation, imaging and VEEG characteristic. A multivariate prognostic prediction model was established based on the above predictive indicators, and its internal effectiveness was evaluated. Results:A total of 114 patients with IS who met the inclusion and exclusion criteria of this study were included in the analysis with 64 males (56.1%) and 50 females. The median age was 30.0 (9.0-102.8) months. There were 78 cases (68.4%) with focal manifestations, 101 cases (88.6%) with systemic manifestations and 65 cases (57.0%) with both focal and systemic manifestations. Sixty-one cases (53.5%) had symptomatic ictus epilepticus after IS. The modified GCS was evaluated when IS patients were admitted to PICU, ranging from 3 to 15(10.5±3.2). Arterial IS (112 cases) accounted for 98.2% of all IS. The infarctus laterally involved 42 cases (36.8%) of bilateral brain simultaneously, 37 cases (32.5%) of left brain and 35 cases (30.7%) of right brain respectively. There were 70 patients(61.4%) with Synek classification score of VEEG≥3. Recurrent IS occurred in 8 cases (7%), 7 of which were males. The median interval between recurrence and the first onset was 4.5 (1.3-14.0) months. There were 81 (71.1%) cases with good prognosis and 33 cases with poor prognosis. Seven patients died (mortality rate 6.1% at discharge). Among the rest 107 survivors, 95 cases (88.8%) had different degree and varied forms of dysfunction including dyskinesia, cognitive impairment, speech disorder and symptomatic epilepsy in a descending order of frequency. None of the dead patients were recurrent cases and all of the dead patients had systemic manifestations, 4 of which had acute symptomatic ictus epilepticus after IS. The modified GCS (OR=0.601, 95%CI: 0.460-0.784) and Synek classification score of VEEG (OR=4.518, 95%CI: 1.682-12.137) were independent predictors of poor prognosis of pediatric IS in PICU. The internal validation of the multivariate prognostic prediction model which was established based on the above predictors showed that the prediction accuracy was 89.5% and the area under ROC curve of the prediction probability was 0.946(95%CI: 0.906-0.986). Conclusion:The incidence of pediatric IS in PICU was low (1.38%), the overall short-term prognosis was good, the mortality rate at discharge was low, but the disability rate was high. The combination of modified GCS and Synek classification score of VEEG got high prediction efficiency which had potential clinical application value.