Abstract： Objective:To investigate the differences in clinical characteristics, diagnosis, and treatment of pediatric septic shock in pediatric intensive care unit (PICU) among hospitals of different levels.Methods:This retrospective study enrolled 368 children with septic shock treated in the PICU of Beijing Children′s Hospital, Henan Children′s Hospital, and Baoding Children′s Hospital from January 2018 to December 2021. Their clinical data were collected, including the general information, location of onset (community or hospital-acquired), severity, pathogen positivity, consistence of guideline (the rate of standard attainment at 6 h after resuscitation and the rate of anti-infective drug administration within 1 h after diagnosis), treatment, and in-hospital mortality. The 3 hospitals were national, provincial, and municipal, respectively. Furthermore, the patients were divided into the tumor group and the non-tumor group, and into the in-hospital referral group and the outpatient or emergency admission group. Chi-square test and Mann-Whitney U test were used to analyze the data. Results:The 368 patients aged 32 (11, 98) months, of whom 223 were males and 145 females. There were 215, 107, and 46 patients with septic shock, with males of 141, 51, and 31 cases, from the national, provincial, and municipal hospitals, respectively. The difference in pediatric risk of mortality Ⅲ (PRISM Ⅲ) scores among the national,provincial and municipal group was statistically significant (26(19, 32) vs.19(12, 26) vs. 12(6, 19), Z=60.25, P<0.001). The difference in community acquired septic shock among the national,provincial and municipal group was statistically significant (31.6%(68/215) vs. 84.1%(90/107) vs. 91.3%(42/46), χ 2=108.26, P<0.001). There were no significant differences in compliance with guidelines among the 3 groups ( P>0.05). The main bacteria detected in the national group were Klebsiella pneumoniae (15.4% (12/78)) and Staphylococcus aureus (15.4% (12/78)); in the provincial group were Staphylococcus aureus (19.0% (12/63)) and Pseudomonas aeruginosa (12.7% (8/63)), and in the municipal group were Streptococcus pneumoniae (40.0% (10/25)) and Enteric bacilli (16.0% (4/25)). The difference in the proportion of virus and the proportion of 3 or more initial antimicrobials used among the national,provincial and municipal group was statistically significant (27.7% (43/155) vs. 14.9% (13/87) vs. 9.1% (3/33), 22.8%(49/215) vs. 11.2%(12/107) vs. 6.5%(3/46), χ 2=8.82, 10.99, both P<0.05). There was no difference in the in-hospital mortality among the 3 groups ( P>0.05). Regarding the subgroups of tumor and non-tumor, the national group had higher PRISM Ⅲ (31(24, 38) vs. 22 (21, 28) vs.16 (9, 22), 24 (18, 30) vs. 17(8, 24) vs. 10 (5, 16), Z=30.34, 10.45, both P<0.001), and it was the same for the subgroups of in-hospital referral and out-patient or emergency admission (29 (21, 39) vs. 23 (17, 30) vs. 15 (10, 29), 23 (17, 29) vs. 18 (10, 24) vs. 11 (5, 16), Z=20.33, 14.25, both P<0.001) as compared to the provincial and municipal group. There was no significant difference in the in-hospital mortality among the 2 pairs of subgroups (all P>0.05). Conclusion:There are differences in the severity, location of onset, pathogen composition, and initial antibiotics of pediatric septic shock in children′s hospitals of different levels, but no differences in compliance with guidelines and in-hospital survival rate.